STACTICALLY LOADED MOTION CONTROL SYSTEM by Manik Kapoor A Thesis Submitted to the Faculty of The Graduate College in partial fulfillment of the requirements for the Degree of Master of Science in Engineering Department of Mechanical and Aeronautical Engineering Western Michigan University Kalamazoo, Michigan August 2007 ACKNOWLEDGEMENT I would like to express my gratitude towards my parents for giving me a great education and experiences which served as a strong foundation needed to complete this project. It is by the virtue of their blessings that I had the strength and ability to embark on this latest endeavor. I wish to express my thanks to my advisor Dr. James Kamman, Professor of Mechanical &Aeronautical Engineering for his time and guidance. It would not have been possible to make this project a reality without his vision. I would also like to express my appreciation to the members of my committee Dr. Muralidhar Ghantasala and Dr. K. Ro for their valuable time and suggestions, Mr. Glen Hall for his assistance and employees of Western Michigan University for their help. Manik Kapoor ii TABLE OF CONTENTS ACKNOWLEDGEMENT ............................................................................................ II LIST OF TABLES ..................................................................................................... VII LIST OF FIGURES ................................................................................................. VIII LIST OF EQUATIONS ........................................................................................... XIII CHAPTER 1: INTRODUCTION ................................................................................. 1 1.1 Introduction to Motion Control System…………………………………….1 1.2 Classification of the Motion Control System………………………………..2 CHAPTER 2: CONTROL SYSTEMS OVERVIEW ................................................... 6 iii 2.1 Introduction to Compensators……………………………………………6 2.2 Digital Control……………………………………………………………9 2.3 System Identification……………………………………………………….14 2.4 Nonlinearities/Linearities in a Hydraulic Motion Control System………….15 Table of Contents-Continued CHAPTER 3: ACTUATION SYSTEM ..................................................................... 18 3.1 Mechanical System…………………………………………………………18 3.3 Assembly…………………………………………………………………...20 3.4 Hydraulic Circuit…………………………………………………………...22 3.5 Control Hardware and Software…………………………………………...23 CHAPTER 4: EQUATION OF MOTION AND TRANSFER FUNCTION OF THE WEIGHTED SLED SYSTEM .................................................................................... 25 iv 4.1 Derivation of Equation of Motion………………………………………….25 4.2 Open Loop System Identification………………………………………….29 CHAPTER 5: CLOSED LOOP MOTION CONTROL SYSTEM SIMULATION .. 34 5.1 Uncompensated Closed Loop Loaded Sled System Simulation……………34 5.2 Compensated Loaded Closed Loop Motion Control System Simulation….36 5.3 Digital Controller Implementation…………………………………………41 Table of Contents-Continued CAHPTER 6:EXPERIMENTAL RESULTS & SYSTEM CHARACTERIZATION. ..................................................................................................................................... 45 CHAPTER 7 : CONCLUSIONS AND RECOMMENDATIONS ............................. 56 REFERENCES ........................................................................................................... 58 APPENDIX ................................................................................................................. 62 v A.1: Comparison of Model Output with Actual System Response…………….62 A.2: Closed Loop Response of Loaded Sled System When Compensator pole is at 155 rad/s and Compensator zero is at 26 rad/s at Different Gains…………..64 A.3: Closed Loop Response of Loaded Sled System When Compensator pole is at 65 rad/s and Compensator zero is at 21 rad/s for Different Gains…………..65 A.4: Closed Loop Response of Loaded Sled System When Compensator pole is at 50 rad/s and Compensator zero is at 18 rad/s for Different Gains…………66 A.5: LabVIEW Code for Phase Lead Closed Loop Control of the Loaded Sled System………………………………………………………………………….67 vi LIST OF TABLES Table 4.1 List of Modeling Parameters ...................................................................... 26 Table 4.2 Value of Unknown Parameters Obtained Using System Identification ..... 31 vii LIST OF FIGURES 1.1 Open Loop Motion Control System ....................................................................... 3 1.2 Closed Loop Motion Control System ..................................................................... 4 2.1 (a) Cascade Compensation ; (b) Feedback Compensation; (c) Input Compensation ....................................................................................................................................... 6 2.2Analog to Digital Conversion: Sampling................................................................. 9 2.3 Comparison of Continuous Signal with Corresponding Digital Signal Obtained After Zero Order Hold ................................................................................................ 10 2.4 Figure Converting Transfer Function to Difference Equation............................. 12 2.5 Saturation ............................................................................................................ 16 2.6 Dead-Zone ........................................................................................................... 17 3.1 Sled Carriage Assembly ...................................................................................... 20 viii List of Figures-Continued 3.2 Assembled Loaded System ................................................................................. 21 3.3 Loaded Sled Attached with LVDT (...................................................................... 21 3.4 Hydraulic Circuit for Loaded Sled....................................................................... 22 3.5 Working of the Closed Loop Loaded Sled System.............................................. 24 4.1 Two Stage Model of Loaded Sled System........................................................... 25 4.2 Cylinder Extension .............................................................................................. 27 4.3 Flow Chart of Open Loop System Identification.................................................. 29 4.4 Comparison of Measured and Simulated Valve and Cylinder Response When Input Command Voltage is 9V ................................................................................... 32 4.5 Variation of Valve Parameters as a Function Input Command ........................... 33 4.6 Variation of Cylinder Dynamics Parameters as a Function Input Command ....... 33 ix List of Figures-Continued 5.1 Root Locus Plot of Uncompensated Closed Loop Loaded Sled System ............. 34 5.2 (a) Voltage Command to Valve (b) Uncompensated Position Response ............. 35 5.3 Root Locus Plot of the Compensated Closed Loop Loaded Sled System ............ 37 5.4 (a) Comparison of Compensated &Uncompensated System Closed Loop Step Response (b) Comparison of Compensated & Uncompensated Closed loop Step response with Saturation and Dead-Zone effects ....................................................... 38 5.5 (a) Compensated Frequency response (b) Phase lead Compensator bode plot ... 40 5.6 Compensated Root Locus Plot ............................................................................. 40 5.7 Closed Loop Loaded Sled Motion Control Model with Embedded MATLAB Function containing the difference equation .............................................................. 42 5.8 Closed Loop Loaded Sled Motion Control Model with Discrete Controller ...... 42 x List of Figures-Continued 5.9 (a) Comparison of Input Command to the Valve from discrete Controller with Command from Embedded MATLAB function (b) Position Response with Discrete Controller Vs Position Response with Embedded MATLAB Function Sample Time=0.01s ................................................................................................................. 43 5.10 (a) Comparison of Input Command to the Valve from discrete Controller with Command from Embedded MATLAB function (b) Position Response with Discrete Controller Vs Position Response with Embedded MATLAB Function Sample Time=0.001s ............................................................................................................... 43 5.8 (a) Effect of Discretization on Command to Valve ;(b) Position Response of the Loaded Sled System with Digital Controller at Different Sample Rates ................... 44 6.1 Feedback Subroutine for Phase Lead Control ..................................................... 46 6.2 (a) Input Command to valve (b) Position Response ............................................ 47 6.3 Compensated Position Responses of Loaded Sled System with Gain=51 and Different Tolerances. .................................................................................................. 48 6.4 Compensated Position Response of Loaded Sled System with Tolerance=0.01 inches and Different Gains. .............................................................. 49 xi List of Figures-Continued 6.5(a) Comparison of Input Command to Valve, (b) Comparison of the Position Response ..................................................................................................................... 50 6.6(a) Comparison of Input Command to Valve, (b) Comparison of the Position Response ..................................................................................................................... 50 6.7 (a) Comparison of Input Command to Valve, (b) Comparison of the Position Response ..................................................................................................................... 51 6. 8 Comparisons of Position Responses Corresponding to Phase Lead Controllers against Proportional Control ....................................................................................... 52 6.9 Compensated Position Response of Loaded Sled with Varying Load ................. 53 6.10 Compensated Position Response of Loaded Sled with Varying Load .............. 54 6.11 Compensated Position Response of Loaded Sled with Varying Load .............. 55 xii LIST OF EQUATIONS 2.1 General Compensator Transfer Function .............................................................. 7 2.2 Transfer Function of Phase Lead Compensator ..................................................... 7 2.3 Transfer Function of a Phase Lead Compensator .................................................. 8 2.4 Transfer Function of a Lead-Lag Compensator .................................................... 8 2.5 Z-Transform of a sampled function ...................................................................... 11 2.6 Equation for Converting Discrete Transfer Function to Difference Equation...... 11 2.7 Tustin’s Method .................................................................................................... 12 2.8 Flow through a Sharp Edged Orifice ................................................................... 17 4.1 Model Transfer Function for the Valve ................................................................ 26 xiii List of Equations-Continued 4.2 Equation of Motion for Cylinder Extension ......................................................... 28 4.3 Model Transfer Function Representing Cylinder Dynamics ............................... 28 4.4 Transfer Function relating Spool Position X(s) and Command Voltage U(s) ..... 30 4.5 Transfer Function relating Cylinder Position Y(s) and Spool Position X(s) ........ 31 5.1 Difference Equation of a Phase Lead Compensator ............................................. 41 xiv CHAPTER 1 INTRODUCTION 1.1 Introduction to Motion Control System Mechanical systems are mainly designed to convert one form of energy (hydraulic, electrical) to another (translational, rotational, etc). Some of the initial systems only had the capability to transmit power. However, as technology developed it gave rise to complex machines requiring mechanical systems which can be controlled to produce accurate motion in terms of position, velocity or acceleration. These mechanical systems were later named “motion control systems”. In other words a motion control system can be defined as a system capable of moving a given load from one point in space to another specified point by accelerating the load to a specific velocity and then decelerating such that the desired position is achieved [1]. A motion control system consists of three main components: the actuation system, the controller, and the sensors. 1.1.1 Actuation System An actuation system contains a power source which can be a hydraulic pump, air compressor or an electric motor and an end effector which can be as complex as a robotic arm or as simple as a sled on a set of parallel guides .The end effector is usually connected to the power source through a cylinder or a lead screw. 1 2 1.1.2 Controller A controller acts as the brain of the motion control system. It can be a stand alone analog unit , a computer or a human .The main function of a controller is to make sure the output of the system is as close to the desired value as possible. 1.1.3 Sensors Sensors act as the eyes and ears of the controller. The main function of the sensors is to measure the parameters being controlled (displacement or velocity or both). 1.2 Classification of the Motion Control System A motion control system can be classified based on how power is transmitted and on whether it is an open loop or closed loop system. Based on the type of power source used we can classify a system as hydraulic, pneumatic, or electromechanical. 1.2. 1 Hydraulic Motion Control System As the name suggests a hydraulic motion control system is one where power is transmitted by circulating a hydraulic fluid through a circuit using a hydraulic pump. These systems have found application in a large variety of fields such as heavy duty industrial applications, automated machining, agriculture, aeronautics and transportation [2] - [4]. Some of the advantages of using a hydraulic power source include [5] long life, easier dissipation of internal heat, and faster response to input changes. Unfortunately these systems can be bulky and messy when compared to pneumatic or electromechanical motion control systems. 1.2.2. Pneumatic Motion control system A pneumatic motion control system is similar to a hydraulic motion control system except compressed air is used to transmit power to the actuator. A pneumatic 3 motion control system is more compact and cleaner when compared to hydraulic systems and has lower operating costs. However, compressibility effects and friction reduce the effectiveness of the system when higher loads are involved [6], [7]. More detailed information about pneumatic systems is provided in references [6]-[8]. 1.2.3. Electromechanical Motion Control System Electromechanical motion control systems use an electric motor to provide power to the end effecter. The type of motor used can be stepper motors, brushless dc motor or gear motors depending on the field of application. Some of the main advantages of electromechanical motion control systems are high functionality to weight ratio, less noise, no compressibility effects, high availability of wide variety of motors at low cost.[9]-[11]. These systems however have shorter life span, tend to overheat if operated for long durations and are difficult to implement when heavy loads are involved[9],[11]. Some of the main areas of application for these systems are in the field of robotics, mechatronics, space craft applications [10], medical instrumentation and micro electromechanical systems. 1.2.4 Open Loop System A motion control system is said to be of open loop type when the input is independent of output. As we can see in Figure 1.1, input is given to the actuation system to work on the load and the output is then measured using a sensor. Figure 1.1 Open Loop Motion Control System (Figure reproduced from [12] with Permission) 4 1.2.5 Closed Loop System When the input to the motion control system changes as a function of the output, it is said to be a closed loop system (Figure1.2). In a closed loop system the output is measured using a sensor and fed back into the control device to generate an error signal which becomes the new input to the actuation system. Figure 1.2 Closed Loop Motion Control System (Figure reproduced from [12] with permission) 1.3 Current Trends and Research Objective As stated earlier hydraulic motion control systems have found applications in wide variety of fields ranging from aeronautics, industrial machining, agriculture, and transportation. However, one of the major limitations faced in developing a hydraulic motion control system is the presence of nonlinearities arising from a variety of sources such as fluid flow, valve dynamics, type of loading, and friction[3],[4],[13]- [18]. Therefore, researchers have been trying to develop and implement controllers capable of maintaining high output accuracy in spite of unknown nonlinearities as well as parametric uncertainties. Preliminary research studied methods of applying linear control theory to these systems [2],[13] - [15]. Fitzsimons suggested use of conic section bound method to account for fluid compressibility and servo valve dynamics while modeling a single degree of freedom hydraulic mount. Research done by Donath et al. and 5 Luigi Del Re et al. suggests the application of feedback linearization to overcome nonlinearities in flow control valves. Although this approach is easier to implement its effectiveness is based on how accurately a system can be approximated using linear theory. In recent years however the main focus of research has been on applying nonlinear control theory to electro-hydraulic motion control systems [3],[4],[16]-[18] . The two main approaches being researched are adaptive control theory and variable structure control [3]. Although the approach suggested in [3], [4] provides an effective method for developing a controller for motion control system, it tends to be complex. Research done by Bin yao et al. [16],[17] applies adaptive robust control theory to develop a controller capable of compensating for valve nonlinearities such as dead-zone and nonlinear flow gains. Alternately Babrow et al. has proposed a controller based on Lyapunov theory for a single degree of freedom hydraulic motion control system. The main focus of this research however, is to model and build a staticallyloaded, closed loop hydraulic motion control system with one degree of freedom based on linear control theory. The hydraulic flow in the proposed system will be regulated using a solenoid-operated flow-control valve. A discrete dynamic phase lead controller will be developed for the continuous plant with a constant load while considering the effects nonlinearities such as saturation and dead-zone. The control system will be implemented in a LabVIEW program to communicate with the motion system using data acquisition hardware. The performance of the system will be compared against a loaded system with proportional control. Robustness of the resulting compensator will be discussed. CHAPTER 2 CONTROL SYSTEMS OVERVIEW 2.1 Introduction to Compensators Efficiency of a linear proportional closed loop system is determined by quantifying the defined performance indices (such as settling time, percent overshoot) and comparing them against desired specifications [5], [19]. However, if these indices fall short of the desired value, performance can be improved to a certain extent by varying system parameters. A more effective way is to alter the transfer function of the actual system by adding a controller designed such that the desired performance indices are achieved. The above process of altering the transfer function to improve system performance is called “compensating” and the controller is known as a “compensator”. A compensator can be introduced into the loop before the plant (cascade compensation), in the feedback loop (feedback compensation) or after the input signal (input compensation) as shown in the Figure 2.3 below. (a) (b) (c) Figure 2.1 (a) Cascade Compensation ; (b) Feedback Compensation; (c) Input Compensation (Figure reproduced from [12] with permission) 6 7 2.1.1 Types of Compensators A compensator can be represented by a general transfer function presented in equation 2.1. M GC ( s ) = ∏ K (s + z ) i =1 N i i ∏ (s + p ) j =1 j Equation 2.1 General Compensator Transfer Function [19] Here, the parameters K i , zi and pi represent the compensator gains, zeros and poles. The properties of a compensator depend on the values of these parameters. Phase lead, phase lag, and lead-lag compensators are the most commonly used. 2.1.1.1 Phase Lead Compensator A compensator is of phase lead type when its transfer function is given by equation 2.2 such that pole is greater than zero in magnitude. GC ( s ) = K (s + z) ( s + p) p>z Equation 2.2 Transfer Function of Phase Lead Compensator Introduction of a phase lead compensator generally causes an increase in the bandwidth and phase margin. This leads to a faster system response, i.e. smaller rise and settling times. However, the increase in bandwidth can make the system more susceptible to high frequency noise. 8 2.1.1.2 Phase Lag Compensator A compensator is of phase lag type when its transfer function is given by equation 2.3 such that pole is smaller than zero in magnitude. GC ( s ) = K (s + z) ( s + p) p<z Equation 2.3 Transfer Function of a Phase Lead Compensator While a phase lag compensator generally leads to higher rise and settling times, the system response is improved by increasing stability and reducing plant’s susceptibility to high frequency noise. 2.1.1.3 Lead-Lag Compensator A lead-lag compensator is represented by the transfer function given in equation 2.4. GC (s) = K 1 K 2 ( s + z1 )( s + z 2 ) ( s + p1 )( s + p 2 ) Equation 2.4 Transfer Function of a Lead-Lag Compensator [12] A lead –lag compensator improves system performance by increasing stability while reducing rise and settling times. It is generally less susceptible to high frequency noise than a phase-lead compensator. 9 2.2 Digital Control 2.2.1 Sampling, Quantization and Zero Order Hold (ZOH) Originally a compensator was practically implemented into the plant by building an analog circuit to perform the required compensation. However, the recent developments in software technology, increased reliability and lower cost of computers have made it easy to implement compensators using digital hardware [20]. The main advantage of using digital hardware is that properties of the compensator can be changed easily to adapt the control law for optimum performance [19]-[21]. To implement digital compensators with a continuous (analog) plant, analog signals must be converted into digital ones. The conversion is done by first “sampling” the incoming analog signal and then quantifying it as a corresponding digital value (“quantization”). The device performing this conversion is known as an analog to digital (A/D) converter [21]. As shown in Figure 2.4 a continuous signal x(t) is converted into specific values x(kT) where k=0,1,2,3,4 recorded at particular times t=0,T,2T,3T with T as sample time. Y Y Y = x (kT) Y = x (t) A/D Converter Time(s) Figure 2.2 Analog to Digital Conversion: Sampling Time(s) 10 The digital compensator then uses the digitized input signal to compute an error signal .The digital error signal is converted back into an analog signal by applying zero order hold (ZOH). As the name suggests a ZOH maintains a constant value of the error signal over the duration of one sample time .This value is updated at the end of each sample time. Hence over time a ZOH results in a staircase signal approximating the continuous signal as shown in the Figure 2.5. [12], [21] 10 9 Discretised Command Signal Continuous Command Signal 8 Function Value(V) 7 6 5 4 3 2 1 0 0 0.1 0.2 0.3 0.4 Time(s) 0.5 0.6 0.7 Figure 2.3 Comparison of Continuous Signal with Corresponding Digital Signal Obtained After Zero Order Hold 2.2.2 Z -Transforms & Difference Equations As the name suggests a linear constant coefficient difference equation represents the difference between the output at a particular time and a finite number of past outputs and inputs [22]. A difference equation is used to approximate the response of a model differential equation representing the plant. Therefore it can also be used to approximate a continuous transfer function. 11 A digital compensator is represented by a discrete transfer function obtained by using the concept of Z-transforms. The Z-transforms is the counter part of Laplace transforms in discrete domain. Therefore for example the Z-transform of a sampled function y=x(kT) can be written as shown in equation 2.5[21]. ∞ Z ( x(kT )) = Y ( z ) = ∑ x(kT ) z − k k =0 Equation 2.5 Z-Transform of a sampled function The term z-k in the above equation signifies the point in time at which the value of y is x(kT) i.e. after a delay of kT seconds. Z-transforms provide an effective way of obtaining a difference equation from a discrete transfer function. From equation 2.5 we can further write [4]: Z ( x ( (k − 1)T ) ) = z −1Y ( z ) x ( (k − 1)T ) = Z −1 ( z −1Y ( z )) Equation 2.6 Equation for Converting Discrete Transfer Function to Difference Equation Equation 2.6 is used to convert a discrete transfer function into a difference equation as shown in Figure 2.4. A more detailed explanation of Z-transforms is given in references [21] and [22]. For a continuous compensator to be implemented using digital hardware, the compensator transfer function must be converted into a corresponding discrete transfer function and then into a difference equation. Tustin’s method is one of the most common methods for finding approximate discrete transfer functions for continuous compensators. The accuracy of the discrete transfer function (and its corresponding difference equation) depends on the method of discretization and the sampling time. 12 2.2.2.1 Tustin’s method As stated in references [21] and [22] Tustin’s method involves performing trapezoidal integration to approximate the error signal at each sample time. Thus if T is the sample time, the continuous transfer function can be discretised by making the following substitution. 2(1 − z −1 ) s= T (1 + z −1 ) Equation 2.7 Tustin’s Method For example a continuous transfer function of a critically damped second order system having natural frequency equal to a rad/s can be discretised and converted into a difference equation as shown in the Figure 2.4. Second Order Continuous Transfer Function X(s) 1 = U(s) (s + a)2 Discretisation : s = 2 (1 − z −1 ) T (1 + z ) Discretised Transfer function −1 X (z) T 2 (z +1) 2 = U (z) (aT + 2) 2 z 2 + 2(a 2T 2 − 4)z + (aT − 2) 2 [( aT + 2) 2 + 2( a 2T 2 − 4) z −1 + ( aT − 2) 2 z −2 ] X ( z ) = (T 2 + 2T 2 z −1 + T 2 z −2 )U ( z ) Conversion to Difference Equation X(k) = T2 2T2 T2 2(a2T2 −4) (aT−2)2 X k − − U ( k ) U k U ( k − 2 ) − ( 1 ) + ( − 1 ) + X(k −2) (aT+2) (aT+2)2 (aT+2)2 (aT+2)2 (at+2)2 Figure 2.4 Figure Converting Transfer Function to Difference Equation 13 In addition to the Tustin’s method, the matched pole zero (MPZ) and modified matched pole zero (MMPZ) methods are also commonly used for deriving difference equations from a continuous transfer function. The MPZ and MMPZ methods are similar, since both involve mapping pole and zeros from the s plane to the z plane using the relation z = e st . A step by step summary of MPZ and MMPZ is given in reference [21] At this point it should be noted that the methods mentioned above develop discrete transfer functions (and difference equations) that “emulate” their continuous transfer function counterparts. Another approach for implementing a digital controller is to design the digital controller directly. This method accounts for the discretization and the sampling time in the design process and usually provides a more robust design. In this research, continuous compensators are designed and then converted into discrete transfer functions using Tustin’s method. 14 2.3 System Identification As written by Gaines the concept of “identification” was introduced by Zadeh [23] as a problem of “relating input and output of a black box by experimental means.” In other words this concept aimed at determining the characteristic equation of any system by utilizing the input signal and the corresponding system response data. As soon as this concept was introduced it found applications in a wide variety of fields ranging from cybernetics to philosophy of science [23]-[26]. One such area where this concept has had a huge impact is that of control systems. The significance of the role system identification plays in designing a closed loop motion control system can never be understated. As confirmed in [3],[4],[16],[17] system identification is an integral step in designing and modeling motion control systems. System identification has enabled researchers to verify the derived model equations (transfer functions, state space model) representing the physics of the system against experimental data thereby allowing them to develop more accurate dynamic models. The experimental data used for validation is normally collected by operating the system as an open loop system. The data collection can also be done by operating the system in closed loop if the system is unstable in open loop [23]. Once a system equation that fits the experimental data is obtained the output of the obtained relationship should be validated against the assumptions and physical limitations of the system. Different methods such as root locus analysis, frequency response analysis, and state space approach can be used. 15 2.4 Nonlinearities/Linearities in a Hydraulic Motion Control System A hydraulic motion control system can be referred to as “a control system with hydraulic power source” [5].Therefore dynamic behavior of a hydraulic motion control system can be studied by developing a transfer function representing the system (plant). This transfer function can then be used to derive principal differential equations for the physical system. The main assumption while deriving the transfer function is that the system is linear. Hence it will be able to predict an accurate behavior of a linear motion control system. However, in practice a hydraulic motion control system is nonlinear [5], [4], and [16]-[18]. These nonlinearities effect system performance and cause the actual system response to be different from that predicted by the transfer function. Therefore in order to design a closed loop motion control system which would correlate to physical performance one needs to take into account the effect of these nonlinearities. Some of the nonlinearities which can be experienced individually or in combination, in a hydraulic motion control system are saturation, dead-zone, nonlinear flow equations, nonlinear gain, backlash, hysteresis and friction [5], [3]. The three simplest and most common types of nonlinearities observed in a hydraulic motion control system are discussed below. 16 2.4.1 Saturation A linear closed loop transfer function assumes that any parameter that has a direct effect on the system performance can be varied without limits i.e. if the input of a closed loop transfer function is increased or decreased the system response should change proportionally. However, if the increase in gain or any other input parameter exceeds the physical limitations of a hydraulic system (such as maximum flow rate of the power pack, operating voltage range of the controller) the system response will saturate. Figure 2.5 below shows the phenomenon of saturation. In the figure n is the output parameter while m is the input parameter, +S is upper saturation limit and –S is lower saturation limit of n. n S m -S Figure 2.5 Saturation [5] 2.4.2 Dead-zone: A dead-zone can be defined as the range of input signal within which there is no system response. A dead-zone can be caused mainly by overlapping (due to machining tolerances) of valve spool lands and valve hydraulic ports. The effect of this nonlinearity becomes prominent when the magnitude of the error signal becomes smaller than the machining tolerances of the valve. This nonlinearity can also arise 17 due to friction between the valve spool and the walls of the valve. Figure 2.6 below is the graphical representation of dead-zone. In the figure n is the output parameter while m is the input signal and D is range of input (m) on either side for which there is no response. n Dead zone m D Figure 2.6 Dead-Zone [5] 2.4.3 Nonlinearity from Valve Flow Equations: A solenoid operated flow control valve will be used for regulating flow through the hydraulic cylinder. The flow rate Q of fluid through the valve is often modeled using Equation 2.5 which relates the flow rate Q to the pressure drop ΔP across a sharp-edged orifice. [5], [12], [19]. Q = Cd A 2( ΔP ) ρ Equation 2.8 Flow through a Sharp Edged Orifice [19] Here A represents the orifice area, ρ the fluid mass density, and Cd a discharge coefficient. Conventionally flow dynamics are accounted for by linearizing the flow model about a nominal point using the concept of Taylor series. Alternate approaches based on non-linear control theory have also been applied to model complex valve flow dynamics [3],[4]and[16]-[18]. 18 CHAPTER 3 ACTUATION SYSTEM 3.1 Mechanical System A weighted sled system was designed (but not built) by a group of senior students as a part of their senior design project [27]. The current system is based on their design. The statically loaded mass sled is made of the following components: trainer stand, hydraulic power drive unit, hydraulic cylinder, and directional control valve, tracks and sensor. A short description of each of the components is provided in the following sections. 3.1.1 Trainer Stand The structure is similar to an ordinary table. However, the system was designed for loads that will range between 50lbs-250lbs; therefore extruded aluminum frame (40 series profile) from Parker Hannifin specifically designed for heavy duty functionalities was chosen. Furthermore, care was taken to reinforce the stand by placing Parker Hannifin’s 20-101 series gussets at each joint. [28] 3.1.2 Hydraulic Power Unit A variable displacement pressure compensating pump (VPAK) capable of delivering a maximum of 6 GPM at 1210 psi will be used as hydraulic power source. The pump is driven by an electric motor capable of providing 5 HP at 1725 RPM. The idea of using variable displacement pressure compensating pump is to be able to adjust flow rates such that the set pressure is always maintained. [29] 19 3.1.3 Hydraulic Cylinder The hydraulic cylinder chosen is a single rod, double acting type with 1.5 inches bore diameter and stroke length of 18 inches. The cylinder has a rod diameter of 5/8 inches. Since the power drive unit can produce a maximum pressure of 1210 psi Parker Hannifin series 2H with maximum pressure limit of 3000 psi was chosen to provide a reasonable factor of safety. [27] 3.1.4 Directional Control Valve The flow control in the system is done using Parker DIFX control valve. D1FX is a spring loaded, solenoid actuated, three position four way valve with a closed center. Furthermore, it’s on board electronics also provides the ability to reduce the valve’s dead-zone range. 3.1.5 Tracks and Pillow Blocks The sled is mounted on 4 SSUPBO-16-XS pillow blocks which move along SRA-16-XS rails from NB Corporation. Each rail is 42 inches long and is capable of withstanding a maximum load of about 1000 lbs. 3.1.6 Sensors Sensors are needed to collect data for the system identification process and to provide feedback to the compensator for closed loop control. The position of the sled will be measured by an R-series linear variable differential transformer (LVDT) manufactured by MTS. The chosen LVDT has a length of 40 inches with an operating voltage range of -10V to +10Vand a resolution of 0.0006 inches. A separate LVDT placed on the valve itself will be used to measure spool position. 20 3.3 Assembly The frame was designed to be easily assembled in a lab environment without the use of special tools or machining. As shown in the Figure 3.1 below the sled carriage assembly consists of a base plate, pillow blocks attached at the bottom, clevis bracket, long bolts and nuts. Nut Long Bolt Clevis Bracket Base Plate Pillow Block Figure 3.1 Sled Carriage Assembly (Figure reproduced from [27] with permission) Once the frame has been assembled the cylinder and other components can be mounted on the top surface as shown in Figure 3.2 below. The rod eye of the cylinder fits into the clevis bracket attached to the sled. Care was taken to mount the cylinder as parallel to the rails as possible. Moreover, there is a small amount of swivel in the rod eye to compensate for misalignment. LVDT Load 21 Hydraulic Cylinder Limit Switch Rails Figure 3.2 Assembled Loaded System (Figure reproduced from [27] with permission) The mass carrying sled slides on the parallel rails as the cylinder is extended or retracted. A position sensor (or LVDT) is mounted directly on the table such that it can be easily connected to the base plate and thereby provide feedback on actual position of the load (Figure 3.3). LVDT Figure 3.3 Loaded Sled Attached with LVDT (Figure reproduced from [27] with permission) 22 3.4 Hydraulic Circuit A hydraulic motion control system is one where power is transmitted by circulating hydraulic fluid through a circuit using a hydraulic pump. The hydraulic actuation circuit used to move the statically-loaded sled system is shown in Figure 3.4. ? A B Ps Ta Vent Valve Pressure Gauge 0.00 Bar ? Electric Motor Relief Valve Fixed/Variable Displacement Pump Figure 3.4 Hydraulic Circuit for Loaded Sled We can observe from Figure 3.4 that the direction of the flow is controlled using a 4 way, 3 position, closed-center flow control valve. The valve is spring loaded and operated using a solenoid. As long as the valve is centered, no flow is permitted. When the flow control valve spool moves to left, flow from the pump is directed towards port A (connected to the cap end of the cylinder) while flow through port B (connected to the rod end of the cylinder) is directed to the tank thereby extending the cylinder. When the valve spool moves to right, the flow direction is reversed i.e. pump flow is directed to port B while flow through port A is 23 directed to the tank thereby causing the cylinder to retract. 3.5 Control Hardware and Software The hydraulic circuit in Figure 3.4 is incorporated into a closed loop feedback system such that flow rate to the cylinder through directional flow control valve is a function of error signal being generated by the controller. However, as stated in Chapter 2, the digital controller and continuous plant speak two different languages. Hence for the system to work, the controller has to be integrated and synchronized with the plant. This will be achieved using a LabVIEW code which will allow real time closed loop control by taking in sampled data and sending out corresponding valve command through a National Instrument’s DAQ MX data acquisition card. The block diagram in Figure 3.5 shows how information exchange between the controller and the plant controls hydraulic flow to the cylinder. The position of the sled is measured using an LVDT and fed back into the computer through data acquisition hardware. Depending on this feedback a new voltage command is given to the control valve thereby controlling cylinder movement. The compensator is modeled using a difference equation and is implemented in a LabVIEW program. 24 Hydraulic Power Unit (Variable Displacement) Data sampling and error signal computation using LabVIEW DIFX Flow Control Valve Hydraulic Cylinder National Instrument 6251 Card Loaded Mass Sled LVDT (To measure Current Position) Figure 3.5 Working of the Closed Loop Loaded Sled System CHAPTER 4 EQUATION OF MOTION AND TRANSFER FUNCTION OF THE WEIGHTED SLED SYSTEM 4.1 Derivation of Equation of Motion The previous section described the physical plant for which the controller will be developed. In this section, dynamic equations of motion are derived for the sled system. The continuous transfer functions relating the input command to valve position and valve position to cylinder/sled position are also derived. The unknown parameters involved in the developed transfer function will be approximated by applying the process of system identification discussed in more detail in section 4.2. For simplicity it is assumed that the fluid is incompressible, that temperature changes through out the system are negligible, and that friction between the moving components of the system is small. Moreover, the flow control valve is assumed to be perfectly matched and symmetrical. The subsequent model is a linearized approximation of the loaded sled motion control system. The model is defined by deriving the relationship between command voltage and valve spool position followed by defining the relationship between valve spool position and sled position as shown in Figure 4.1. Command Voltage Second order Valve Dynamics Spool Position K s( s + a) Cylinder Position y(t) Hydraulic Power Unit & Cylinder Figure 4.1 Two Stage Model of Loaded Sled System 25 26 Q = Flow Rate P = Pressure Difference x = Change in Spool Position y = Change in Load Position m = Mass ACap = Damping Coefficient = Force = Area at the cap end of the cylinder ARod = Area at the rod end of the cylinder C F ζ = Damping Ratio Tw , T p = Time Constants Table 4.1 List of Modeling Parameters The first step in this two stage approach is to obtain a linear model approximation for the Parker D1FX proportional directional control valve. Parameters used for developing the transfer functions approximating the flow control valve and cylinder dynamics are defined in Table 4.1. The equation of motion for the flow control valve is derived by visualizing it as a linear spring-mass-damper system. Assuming v(t ) represents the valve input voltage and x(t ) represents the valve spool position, the transfer function can be written as X (s) K = 2 2 V ( s ) Tw s + 2ζ Tw s + 1 Equation 4.1 Model Transfer Function for the Valve The validation of the above approximation using open loop system identification is discussed in section 4.2. 27 PS Tank Tank Valve Spool x(t) A B QB QA • Pcap Acap y(t) Cy m PRod ARod Figure 4.2 Cylinder Extension The second step in modeling is to derive a transfer function capturing the single rod cylinder dynamics shown in the Figure 4.3. The flow rate Q into and out of the cylinder during extension can be defined as a function of change in spool position and pressure gradient P as [12]; Q = f ( x, P ) As stated earlier the flow equation is generally nonlinear, hence for simplicity the above model is linearized about a nominal spool position and pressure (xo,Po) using a Taylor series expansion. Therefore the change in flow rate can be approximately written as: ⎛ ∂f ( x, P) ⎞ ⎛ ∂f ( x, P) ⎞ Q=⎜ x⎟ p⎟ = Gx x − G p P −⎜ ⎠ ( xo , p o ) ⎠ ( xo , po ) ⎝ ∂P ⎝ ∂x 28 Thus for an incompressible fluid we can write Q = Ay = G x x − G p P or P= − Ay + G x x Gp Applying Newton’s second law to mass m in Figure 4.3 and substituting from the equations above, the equation of motion for cylinder extension can be written as in Equation 4.2. ⎛ Acap G x ( cap ) ARod G x ( rod ) my + Cy = ⎜ + ⎜ G G p ( rod ) p ( cap ) ⎝ ⎞ ⎛ A 2 cap A 2 rod ⎟x − ⎜ + ⎟ ⎜G ⎠ ⎝ p ( cap ) G p ( rod ) ⎞ ⎟ y ⎟ ⎠ Equation 4.2 Equation of Motion for Cylinder Extension The above derived equation of motion is then used to derive a corresponding transfer function (Equation 4.3) using Laplace transforms. Y (s) K = X ( s) s( s + a) K = a = 1 ⎛ Acap Gx ( cap ) ARod Gx ( rod ) ⎞ + ⎜ ⎟ m ⎜⎝ G p ( cap ) G p ( rod ) ⎟⎠ ⎛ A2 cap A2 rod ⎞ ⎞ 1⎛ + ⎜C + ⎜ ⎟⎟ ⎟ ⎜G ⎟ m ⎜⎝ G p ( rod ) ⎠ ⎠ ⎝ p ( cap ) Equation 4.3 Model Transfer Function Representing Cylinder Dynamics 29 4.2 Open Loop System Identification The next step is to experimentally characterize the above derived transfer function using actual system data. This is done by performing open loop system identification. This two-stage process is shown in Figure 4.4. The sections below provide more detail on each stage. Open Loop Data Acquisition LabVIEW VI Loaded Sled System NI DAQ 6251 Card Computer Data Processing System Identification Toolbox Figure 4.3 Flow Chart of Open Loop System Identification 30 4.2.1 Open Loop Data Acquisition: Experimental data is collected using an existing LabVIEW program in the Parker Motion Control Lab. The program acts as a graphical user interface which allows the user to select a constant command voltage (step input) and the sample rate at which data is to be collected. The program records the time, command voltage, spool position voltage and sled position voltage. Seven data sets were collected at a sample rate of 1000 Hz for command voltages varying from 3-9V in increments of 1V with the sled load of 50 lbs. The collected data corresponds to cylinder extension. 4.2.2 System Identification: The command voltage, spool position voltages and cylinder position voltages were extracted separately from each of the seven saved data files for the individual command voltages and averaged to reduce noise effects. The averaged variables were then imported into MATLAB’s system identification tool box. The general transfer functions relating spool position with command voltage and cylinder position with spool position are given by equations 4.4 and 4.5 respectively. X (s) K = 2 2 V ( s ) Tw s + 2ζ Tw s + 1 Equation 4.4 Transfer Function relating Spool Position X(s) and Command Voltage U(s) 31 K1 Y (s) = X ( s ) s (Tp s + 1) Equation 4.5 Transfer Function relating Cylinder Position Y(s) and Spool Position X(s) The values of unknown parameters Tw, K, ζ in equation 4.4 and K1 , Tp in equation 4.5 are shown in Table 4.2 below. The values were obtained when the sled was loaded with 50 lbs. Command Voltage(V) K Tw ζ K1 Tp 3 -1.3614 0.01138 0.68317 -0.36853 0.0065187 4 -1.2529 0.0084079 0.54244 -0.49726 0.013889 5 -1.1814 0.0095218 0.61304 -0.58478 0.030369 6 -1.1334 0.007983 0.59853 -0.62924 0.037097 7 -1.0993 0.0075337 0.65203 -0.65119 0.040797 8 -1.0736 0.0076073 0.68822 -0.65192 0.046561 9 -1.0551 0.0079625 0.71174 -0.63902 0.047813 Table 4.2 Value of Unknown Parameters Obtained Using System Identification The output obtained from the above transfer functions with parameter values corresponding to a 9V command is plotted against the experimental data in Figure 4.4. The output of the derived transfer functions matches the system response very closely at each of the commanded voltages. The comparison of model output with experimental data for command voltages between 3-8 V is presented in Appendix A.1. 32 2.5 0 2 Sled Position(in) -2 Spool Position(V) Simulated Output Measured Output Measured Output Simulated Output -4 -6 -8 -10 1.5 1 0.5 0 0.1 0.2 Time(s) 0.3 0.4 0 0 0.1 0.2 Time(s) 0.3 0.4 Figure 4.4 Comparison of Measured and Simulated Valve and Cylinder Response When Input Command Voltage is 9V Unfortunately, the values of the parameters Tw , K and ζ in equation 4.4 and the parameters K1 and Tp in equation 4.5 are not constant over the command input range, indicating the valve and cylinder dynamics are both non-linear. The valve and cylinder parameters are plotted as a function of command voltage in Figures 4.5 and 4.6. The values for each command voltage are normalized with respect to the values associated with a 3 volt command. The small variability in the valve parameters in the 7-9V range indicates fairly linear behavior there; however, the cylinder dynamics seems not to have a linear range. In spite of the lack of linearity, the transfer functions associated with a 9V command are used in the compensator design process. 33 1.2 Norm alised Param eter Value 1 0.8 0.6 "Normalised Valve Gain(K)" 0.4 "Normalised Valve Time Constant(Tw)" 0.2 Normalised Damping Ratio(Zeta) 0 3 4 5 6 7 8 9 Input Com m and(V ) Figure 4.5 Variation of Valve Parameters as a Function Input Command. Note: Normalization is done against parameters corresponding to 3V Command Normalised Parameter Value 8 7 6 5 4 Normalised Cylinder Gain(K1) 3 2 Normalised Cylinder Dynamics Time Constant 1 0 3 4 5 6 7 8 9 Input Com m and(V ) Figure 4.6 Variation of Cylinder Dynamics Parameters as a Function Input Command. Note: Normalization is done against parameters corresponding to 3V Command 34 CHAPTER 5 CLOSED LOOP MOTION CONTROL SYSTEM SIMULATION 5.1 Uncompensated Closed Loop Loaded Sled System Simulation The first step in designing a phase lead compensator was to study the performance of the closed loop loaded sled system with proportional control for a specific closed loop gain. A root locus analysis of the system indicated a crossover gain of 133. The root locus plot of the loaded sled system carrying a load of 50 lbs is shown in Figure 5.1. Closed loop pole locations for a chosen gain of 51 are also shown. The performance of the system is further studied by simulating a closed loop model of the loaded sled system. The simulation is done to evaluate performance indices (such as settling time and percent overshoot) while considering the effects of saturation and dead-zone, thereby providing a more realistic approximation. 250 0.76 0.64 0.5 0.34 0.16 200 0.86 150 Uncompensated Closed Loop Poles for K=51 Imaginary Axis 100 0.94 50 0.985 0 250 200 150 100 50 -50 0.985 -100 0.94 -150 0.86 -200 0.76 -250 -300 0.64 -200 0.5 0.34 0.16 -100 0 100 200 Real Axis Figure 5.1 Root Locus Plot of Uncompensated Closed Loop Loaded Sled System 35 Simulation began by creating a model approximating the original continuous plant with proportional control (uncompensated system). The model also takes into account saturation and dead-zone thereby allowing the user to see their corresponding effects on performance. The saturation limits are set between +10 V to -10 V since this is the operating range of the data acquisition board and the LVDT. Figure 5.2 shows input command to the valve and corresponding position response of the uncompensated system corresponding to a gain value of 51. 1.5 50 Input Valve Command without Saturation & Dead-Zone Input Valve Command with Saturation & Dead-Zone 30 1 Sled Position(in) Input Valve Command(V) 40 20 10 0 0.5 -10 -20 -30 0 Uncompensated Step Response without Saturation &Dead-Zone Uncompensated Step Response with Saturation & Dead-Zone 0.1 0.2 0.3 0.4 (a) 0.5 0.6 Time(s) 0.7 0.8 0.9 1 0 0 0.1 0.2 0.3 0.4 0.5 Time(s) 0.6 0.7 (b) Figure 5.2 (a) Voltage Command to Valve (b) Uncompensated Position Response The uncompensated system overshoots by 15.5 % (when the effect of saturation is considered) from the original value for the considered gain and has a settling time of 0.58 seconds. Although performance can be improved to a certain degree by reducing gain a better approach will be to change the structure of transfer function by adding a compensator. Hence, the performance of the system was improved by designing a phase lead controller discussed in the next section. 0.8 0.9 1 36 5.2 Compensated Loaded Closed Loop Motion Control System Simulation 5.2.1 Compensator Design Using Root Locus Approach In the last section the performance of the uncompensated closed loop system was determined. A phase lead controller was designed using root locus method to improve the system response by reducing the settling time and increasing the region of stability. The compensator is designed such that the settling time is less than 0.3 seconds for an input command of 1 inch with damping ratio equal to 0.7 to reduce the overshoot to less than 5%. Since the system is of type 1(from equations 4.4 &4.5) and the input signal will always be a step command the steady state error will be theoretically zero [2]. In practice however, the error will only be as small as the actuation system and sensor will allow .The specified performance was achieved by placing the compensator zero at 26 rad/s and compensator pole at 155 rad/s on the real axis. A root locus analysis of the compensated closed loop transfer function was done to study the effect of the compensator. The root locus plot for the compensated system is shown in Figure 5.3.The effect of adding the compensator was to shift the asymptotic center to the left of its original position. This results in an increase of crossover gain from 133 for the uncompensated system (Figure 5.1) to 184 for the compensated system. The increase in the crossover gain is an indication that the system is more stable. 37 200 0.7 150 Compensated Close Loop Poles 100 For K = 51 Imaginary Axis 50 0 -50 -100 -150 0.7 -200 -300 -250 -200 -150 -100 -50 0 50 100 150 Real A x is Figure 5.3 Root Locus Plot of the Compensated Closed Loop Loaded Sled System The step response for the chosen gain of 51 without considering the effects of saturation is compared in Figure 5.4 (a). The system performance was then further measured by simulating uncompensated and compensated model while considering the effect of saturation and dead-zone. Figure 5.4 (b) compares the uncompensated and compensated position response for the same gain. 38 1.5 1.2 Uncompensated Step Response 1 Compensated Step response 0.8 Amplitude Position(V) 1 0.6 0.4 Uncompensated Stept Response with Saturation Compensated Step Response with Saturation 0.5 0.2 0 0 -0.2 0 0.1 0.2 0.3 0.4 Time (sec) (a) 0.5 0.6 0.7 0.8 0 0.1 0.2 0.3 0.4 0.5 0.6 Time (s) 0.7 0.8 (b) Figure 5.4 (a) Comparison of Compensated &Uncompensated System Closed Loop Step Response (b) Comparison of Compensated & Uncompensated Closed loop Step response with Saturation and Dead-Zone effects The compensated closed loop loaded sled system overshoots by only 4% as compared to 15% overshoot observed in case of uncompensated system. Furthermore, the settling time is reduced from 0.58 s to 0.15 s indicating that the chosen compensator has made system more responsive and stable. The effect of saturation was to increase the settling time to 0.27 s .while eliminating the overshoot. The phase and gain margins for the compensated are evaluated to be 54 degrees and 11.3 db respectively. Although the required performance was achieved, the addition of the phase lead compensator might make the loaded system prone to high frequency noise. 0.9 1 39 5.2.2 Compensator Design using Frequency Response Method The compensator was initially designed by following the methodology for frequency response method given in reference [19]. The phase margin for the uncompensated system was calculated to be 26.9 degrees and the gain margin of 8.29 db. The compensator is designed such that the compensated system will have a phase margin greater than or equal to 53 degrees. The pole and zero locations are calculated by considering a loop gain of 51 (same as that for the root locus method) and a shift of 43 degrees in the phase margin. The application of above methodology yielded a phase margin of 46.4 degrees and reduced the gain margin to 7.74 db .Hence the original methodology was modified such that logarithmic mean frequency is changed keeping the phase shift constant until the required phase margin is obtained. By following the modified procedure the pole is calculated to be at 158.7 radians /s and the zero at 26.6 radians /s. The gain margin increased to 11.3 db and a phase margin of 53.5 degrees was achieved. Figure 5.5 (a) below shows a comparison of frequency response of the open loop uncompensated and compensated system. Frequency response of the phase lead compensator is shown in 5.5(b). Bode Diagram 15 Magnitude (dB) Magnitude (dB) 0 -50 -100 -150 10 5 -180 Uncompens ated Respons e Compens ated Respons e -270 Phase (deg) 0 60 -360 0 10 1 10 2 10 3 10 30 0 4 10 Frequenc y (rad/s ec) 0 10 1 10 2 10 3 Frequency (rad/sec) (a) (b) Figure 5.5 (a) Compensated Frequency response (b) Phase lead Compensator bode plot The root locus plot (Figure 5.6) was drawn to study the effect of the added pole and zero. Note that the root locus of the compensated system transfer function derived using the modified frequency response approach is comparable to the one obtained in root locus method. Root Locus 200 0.8 0.68 0.54 0.38 0.18 150 0.89 100 0.95 Compensated Close Loop Poles 50 0.986 Imaginary Axis Phase (deg) -200 -90 10 40 Bode Diagram 20 50 0 250 200 150 100 50 -50 0.986 -100 0.95 -150 0.89 0.8 -200 -300 -250 0.68 -200 0.54 -150 0.38 -100 0.18 -50 0 50 Real A xis Figure 5.6 Compensated Root Locus Plot 100 150 10 4 41 5.3 Digital Controller Implementation The implementation of the derived compensator was done by digitizing the continuous transfer function using Tustin’s method (section 2.3) to obtain a difference equation. A general form of difference equation representing the continuous transfer function of a phase lead compensator is shown below. x( k ) = α ( zT + 2) pT + 2 u (k ) + α ( zT − 2) pT + 2 u (k − 1) − ( pT − 2) x(k − 1) pT + 2 Equation 5.1 Difference Equation of a Phase Lead Compensator Here α =p/z, p and z are compensator pole and zero respectively. Before implementing the above derived transfer function as a digital controller the effect of discretization on the continuous compensator transfer function was studied. This was done by creating a model having an embedded MATLAB function block containing the difference equation (difference equation model) shown in Figure 5.7. The user can choose a sample rate of the block thereby making it to function as a digital controller performing sampling and zero order hold. The critical sample rate for the system is 0.005 seconds .This number was arrived at by calculating the closed loop system bandwidth and multiplying a factor of 20. Before studying the effects of implementing a digital controller, response obtained from difference equation model (Figure 5.7)is validated against the response obtained from a model having a discrete compensator transfer function in z-domain (discrete controller model) shown in Figure 5.8. 42 Command To Workspace1 Valve Voltage Command Position To Workspace 51 Step1 Proportional Gain3 u Tustin In1 y Embedded MATLAB Function Saturation from DAQ board Out1 Continuous Plant Cylinder Position Figure 5.7 Closed Loop Loaded Sled Motion Control Model with Embedded MATLAB Function containing the difference equation Com m and T o Workspace1 Valve Voltage Com m and Position T o Workspace 51 5.603z-5.459 In1 Out1 z-0.8561 Step1 Proportional Gain3 Dis crete Trans fer Fcn Saturation from DAQ board Continuous Plant Cylinder Position Figure 5.8 Closed Loop Loaded Sled Motion Control Model with Discrete Controller The models shown in Figures 5.7 and 5.8 were simulated for sample time of 0.01 second and 0.001 seconds .The input command to the continuous plant and corresponding position response for the considered sample times are comparable. The results are shown in Figures 5.9 and 5.10. 43 1.4 10 Input Command w ith Embedded MATLAB Function 6 1 Sled Position(in) 1.2 4 2 0.6 0.4 -2 0.2 -4 0 0.1 0.2 0.3 Time(s) 0.4 0.5 0 0.6 Position Response with Embd MATLAB Function Position Response with Discrete Controller 0.8 0 0 0.1 0.2 (a) 0.3 Time(s) 0.4 0.5 0.6 (b) Figure 5.9 (a) Comparison of Input Command to the Valve from discrete Controller with Command from Embedded MATLAB function (b) Position Response with Discrete Controller Vs Position Response with Embedded MATLAB Function Sample Time=0.01s 10 Input Command from Embedded MATLAB Function Input Command from Discrete Controller 1 8 0.8 6 Sled Position(in) In p u t V a lv e C o m m an d (V ) Input Valve Command(V) Input Command w ith Discrete Controller 8 4 2 0.6 0.4 0 0.2 -2 -4 Position Response with Embedded MATLAB Function Position Response with Discrete Controller 0 0.1 0.2 0.3 Time(s) (a) 0.4 0.5 0.6 0 0 0.1 0.2 0.3 Time(s) 0.4 (b) Figure 5.10 (a) Comparison of Input Command to the Valve from discrete Controller with Command from Embedded MATLAB function (b) 0.5 0.6 44 Position Response with Discrete Controller Vs Position Response with Embedded MATLAB Function Sample Time=0.001s The difference equation model was then simulated to test the difference equation at different sample rates above and below the critical sample rate. Figure 5.11 displays the input command to the continuous plant and corresponding position response when sample time for the controller was varied from 0.01 - 0.001 seconds. For sample time above the critical value of 0.005 seconds a highly serrated input command having a larger undershoot is observed. However, as the sample time is reduced a more accurate approximation of the continuous signal is obtained. 10 Continuous Command Sample Rate=0.001 s Sample Rate=0.003 s Sample Rate=0.01 s 1 6 0.8 Sled Position(in) Input Valve Command(V) 8 4 2 0.6 0.4 0 -2 -4 Continuous Response Sample Time=0.001 s Sample Time=0.003 s Sample Time=0.01s 0.2 0 0.1 0.2 0.3 Time(s) (a) 0.4 0.5 0.6 0 0 0.1 0.2 0.3 Time(s) (b) Figure 5.8 (a) Effect of Discretization on Command to Valve ;(b) Position Response of the Loaded Sled System with Digital Controller at Different Sample Rates 0.4 0.5 0.6 45 CHAPTER 6 EXPERIMENTAL RESULTS & SYSTEM CHARACTERIZATION In the previous chapter, phase lead compensator for the loaded system was designed and performance of the compensated closed loop loaded sled system was simulated. The simulated compensated performance was then compared with uncompensated simulated performance. Furthermore, the designed compensator was digitized and then its response for the chosen gain and at different sample rates was presented. In the present chapter actual system response is determined by applying the derived difference equation (Equation 5.1) as a controller for the loaded sled system and performing closed loop control. Implementation was done by modifying the feedback subroutine of the present LabVIEW program for proportional control to include the phase lead controller. Figure 6.1 shows the modified feedback subroutine containing the difference equation. 46 Figure 6.1 Feedback Subroutine for Phase Lead Control The program provides the ability to exercise closed loop position control on the loaded sled system .It also has the ability to record time, input command, spool position and cylinder position in text format. The user is required to choose gain, sample rate, desired position, and pole and zero values for the compensator as well as the file path where data is to be stored. The above controller was used to perform closed loop control on the sled carrying a load of 50 lbs. The position response for a command of 9 inches is shown in Figure 6.2. The loop gain is equal to 51, a tolerance of 0.01 inches above and below the desired position is allowed and sampling is done at a frequency of 1000 47 Hz. The hydraulic supply pressure is maintained at 280 psi. Even though theoretical analysis and simulation in the previous sections indicated a stable response, the loaded sled tends to oscillate about the desired position of 9 inches (Figure 6.2 (b)). This is mainly due to magnification of the inherent noise by the compensator which moves the noise floor outside the chosen tolerance thereby destabilizing the system. (Note: The input command to the valve and corresponding position response for the individual gains and tolerances are presented in appendix A.2). 10 8 5 Sled Position(in) Input Valve Command(V) 10 0 6 4 -5 2 -10 0 0 0.5 1 1.5 Time(s) 2 2.5 3 0 0.5 1 1.5 Time(s) (a) 2 2.5 3 (b) Figure 6.2 (a) Input Command to valve (b) Position Response Therefore in order to reduce system susceptibly to noise and improve performance the allowed tolerance was increased while keeping the gain value of 51 constant and a maintaining a sample rate of 1000 Hz. The tolerances were varied between 0.01”0.09”.The position response for different tolerances is shown in Figure 6.3(a). The minimum tolerance that can be achieved for the chosen gain without destabilizing the system is 0.05 inches above and below the desired value of 9 inches (Figure 6.3(b)). 48 10 10 9 9.5 8 9 Sled Position(in) Sled Position(in) 7 6 5 4 8.5 8 3 1 0 7.5 Tolerance=0.01" Tolerance=0.05" Tolerance=0.07" Tolerance=0.09" 2 0 0.5 1 1.5 Time(s) 2 2.5 Tolerance=0.01" Tolerance=0.05" Tolerance=0.07" Tolerance=0.09" 7 3 1.2 1.4 1.6 Time(s) 1.8 2 (a) (b) Figure 6.3 Compensated Position Responses of Loaded Sled System with Gain=51 and Different Tolerances. Another approach was to reduce the gain value while keeping the tolerance of 0.01 inches constant. As shown in Figure 6.4 (b) the stability of the loaded sled system improves as the gain is lowered from 51 to 10. A good response is obtained for a gain value of 10. 10 10 9 9.5 8 9 Sled Position(in) Sled Position(in) 7 6 5 4 8 3 Gain=51 Gain=20 Gain=15 Gain=10 2 1 0 8.5 0 0.5 1 1.5 Time(s) (a) 2 2.5 Gain=51 Gain=20 Gain=15 Gain=10 7.5 3 7 1.2 1.4 1.6 1.8 Time(s) (b) 2 2.2 2.2 49 Figure 6.4 Compensated Position Response of Loaded Sled System with Tolerance=0.01 inches and Different Gains. Since the above designed compensator made the loaded sled system highly responsive and prone to inherent noise in the system, two alternatives to the pole and zero combination used above are suggested. The first alternative has compensator pole location at 65 rad/s and compensator zero was shifted to 21 rad/s. The second compensator has pole at 50 rad/s and zero at 18 rad/s. These values were arrived at by using methods discussed in chapter 5. This was done to determine if the effect of noise can be reduced by changing the compensator pole and zero locations such that system is less responsive. The two designed compensators were individually used to exercise closed loop control on the loaded sled (carrying a load of 50 lbs). The position response of each of the suggested two phase lead compensators was studied for different gains and tolerances. Shifting compensator pole from 155 to 50 rad/s and zero from 26 to 18 rad/s allowed using higher gains for a constant tolerance and vice versa thereby suggesting an increase in noise tolerance to a certain degree. The position responses and corresponding input commands to the valve for the two phase lead controllers are presented in Appendix A.3 and A.4 The above analysis allowed us to identify input parameters (Gain, Tolerance) for optimum performance of the loaded sled system. The developed compensated model (Figure 5.7) was then simulated for a gain of 10 and tolerance of 0.01 inches for each of the three phase lead compensators. Figures 6.5-6.7 compares the simulated position response and corresponding input command to the valve with actual data for each of the three phase lead compensators. 50 Compensator Pole= 155 rad/s Compensator Zero=26 rad/s Compensator Pole=155 rad/s Compensator Zero=26 rad/s 12 10 Actual response Simulated response 10 Sled Position(in) Input Valve Command(V) 8 8 6 4 2 6 4 2 0 Simulated Response Actual Response -2 0 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 (b) (a) Figure 6.5(a) Comparison of Input Command to Valve, (b) Comparison of the Position Response Compensator Pole =65 rad/s Compensator Zero=21 rad/;s Compensator Pole =65 rad/s Compensator Zero =21rad/s 10 12 Actual Response Simulated Response 10 9 7 8 S led P os ition(in) Input Valve Command(V) 8 6 4 6 5 4 3 2 2 0 Actual Response Simulated Response 1 0 0.5 1 1.5 2 Time(s) (a) 2.5 3 3.5 0 0 0.5 1 1.5 2 Time(s) (b) Figure 6.6(a) Comparison of Input Command to Valve, (b) Comparison of the Position Response 2.5 3 3.5 51 Compensator Pole=50 rad/s Compensator Zero=18 rad/s Compensator Pole=50 rad/s Compensator Zero=18 rad/s 12 10 Actual Response Simualted Response 9 10 7 8 Sled Position(in) Input Valve Command 8 6 4 6 5 4 3 2 2 Actual Response Simulated Response 1 0 0 0.5 1 1.5 2 Time(s) (a) 2.5 3 3.5 0 0 0.5 1 1.5 2 Time(s) 2.5 (b) Figure 6.7 (a) Comparison of Input Command to Valve, (b) Comparison of the Position Response The response predicted by the developed model is comparable with the actual response indicating that the developed model is a good approximation of the actual system. Furthermore, the simulated and actual response for the third phase lead controller is a close match proving that the system tolerance to noise is increased. 3 3.5 52 The responses of the three controllers are further analyzed by comparing them against that obtained using the proportional controller in Figure 6.8. A gain of 10 and a tolerance of 0.01 inches above and below the desired position of 9 inches were 10 9.3 9 9.2 8 9.1 7 9 Sled Position(in) Sled Position(in) chosen for each response shown below. 6 5 4 8.9 8.8 8.7 8.6 3 1 0 8.5 Pole=155 rad/s Zero=26 rad/s Pole=65 rad/s Zero=221 rad/s Pole=50 rad/s Zero=18 rad/s Proportional Control 2 0 0.5 1 1.5 Time(s) (a) 2 2.5 Pole=155 rad/s Zero=26 rad/s Pole=65 rad/s Zero=221 rad/s Pole=50 rad/s Zero=18 rad/s Proportional Control 8.4 8.3 3 1.4 1.6 1.8 2 Time(s) 2.2 2.4 (b) Figure 6. 8 Comparisons of Position Responses Corresponding to Phase Lead Controllers against Proportional Control The system response is slowest (for the above combination of gain and tolerance) when proportional control is used(Figure 6.8(b)) .Comparing the position response of the three phase lead controllers indicates that system reached within 1% of the desired position in approximately 1.65 seconds when pole and zero was at 155 rad/s and 26 rad/s respectively. However, as observed earlier the above compensator made the system more responsive thereby causing it to overshoot by 5% .On the other hand the other two compensators made the system less responsive (no overshoot) . However, it also took longer i.e. 1.7 seconds for the second controller (compensator pole= 65 rad/s and zero =21 rad/s) and 1.75 seconds for the third controller 53 (compensator pole=50 rad/s and compensator zero =18 rad/s) to reach within 1% of the desired position (relative to the original compensator). Although as stated above the controllers are able to achieve the desired position in a reasonable time. The above three phase lead controllers were designed based on open loop data collected when the sled was carrying a load of 50 lbs. Hence, further study was done to determine the robustness of the suggested controllers as the load on the sled is changed. Figures 6.9-6.11 show the position response for the three suggested controllers when the load was varied between 40 – 70 lbs at an increment of 10 lbs. A gain of 10 and a tolerance of 0.01 inches above and below the desired position of 9 inches were chosen for each response shown below. Compensator Pole=155 rad/s Compensator Zero=26 rad/s Compensator Pole=155 rad/s Compensator Zero=26 rad/s 10 9 9.2 8 9 Sled Position(in) Sled Position(in) 7 6 5 4 8.8 8.6 3 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 2 1 0 0 0.5 1 1.5 Time(s) 2 2.5 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 8.4 8.2 3 1.3 1.4 1.5 Time(s) 1.6 (b) (a) Figure 6.9 Compensated Position Response of Loaded Sled with Varying Load 1.7 54 Compensator Pole=65 rad/s Compensator Zero=21 rad/s Compensator Pole=65 rad/s Compensator Zero=21 rad/s 10 9.6 9 9.4 8 9.2 Sled Position(in) Sled Position(in) 7 6 5 4 9 8.8 8.6 3 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 2 1 0 0 0.5 1 1.5 2 Time(s) 2.5 3 8.4 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 8.2 3.5 1.4 1.45 1.5 1.55 (a) 1.6 1.65 Time(s) 1.7 1.75 1.8 (b) Figure 6.10 Compensated Position Response of Loaded Sled with Varying Load Compensator Pole=50 rad/s Compensator Zero=18 rad/s Compensator Pole=50 rad/s Compensator Zero=18 rad/s 10 10.5 9 10 8 9.5 Sled Position(in) Sled Position(in) 7 6 5 4 8.5 8 3 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 2 1 0 9 0 0.5 1 1.5 2 Time(s) (a) 2.5 3 Load=40 lbs Load=50 lbs Load=60 lbs Load=70 lbs 7.5 7 3.5 1.2 1.4 1.6 1.8 Time(s) (b) 2 2.2 2.4 55 Figure 6.11 Compensated Position Response of Loaded Sled with Varying Load As the load on the sled is varied we observe that the corresponding steady state position response is nearly the same as the one corresponding to 50 lbs. It was possible to achieve the desired position of 9 inches in case of all the three controllers for the same combination of gain and tolerance. Moreover the time taken to reach the required position did not increase with increasing load. Thus for the chosen gain and tolerance the three suggested phase lead controllers are considerably robust to load changes between 40-70 lbs. 56 CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS 7.1 Conclusion The range of this research project included modeling and building a statically loaded closed loop hydraulic motion control system with a discrete phase lead controller. The research project developed and integrated the different steps required to design, implement and test a phase lead/lag digital controller for a continuous plant. Mathematical models were developed based on derived transfer functions to simulate the effect of continuous as well as digital phase lead compensator on the continuous plant while including nonlinearities such as dead-zone and saturation. The analysis indicated an improvement in settling time from 0.58 seconds to 0.27 seconds for step input of 1” .Furthermore; a frequency response analysis indicated an increase in gain margin of the system with phase lead controller from 8.29 db to 11.3 db while phase margin increased from 26.9 degrees to 54 degrees. The designed digital phase lead controller was implemented into a LabVIEW program and system parameters (such as gain, tolerance) for optimum performance were identified. Experiments show that optimum performance was obtained for a gain value of 10 or a tolerance of 0.05”. Subsequently, two alternative phase lead controllers were suggested and studied. For a given set of inputs the response predicted by the simulation is comparable to the actual response. Finally a robustness study was done to determine the load range within which the suggested controllers 57 will perform satisfactorily. A good response was obtained for loads between 40 -70 lbs. 7.2 Recommendations The inherent noise in the system was not considered during compensator design .The introduction of the compensator however magnified the noise thereby producing a different response than what was predicted. Hence a more comprehensive noise analysis is required. The noise analysis would involve identifying the lowest noise frequency that overlaps with the frequency response of the loaded closed loop system. Moreover power spectrum density should also be plotted to provide a better understanding of how power varies as a function of frequency. Based on the noise analysis a filter transfer function can be derived and incorporated into the developed models for a more realistic approximation. Signal conditioners must be incorporated in the loop to obtain a cleaner signal from the sensor. Presently the controller developed in this project is only capable of performing point to point position control .The next step should be to model a controller for the loaded sled system capable of tracking a ramp input and then implementing it in a LabVIEW program. This would provide the ability to compare simulated closed loop response of the loaded sled system to a ramp input against the actual response. The hydraulic flow in the loaded motion control system is regulated by D1FX flow control valve. Further study needs to be done to determine how the loaded system behaves when D1FX is replaced by a D1FH flow control valve or a servo valve. Completion of this study would result in a system which can be used to test 58 and compare different types of hydraulic motion control system. Finally, research needs to be done on how to include pneumatic and electromechanical drives to actuate the loaded sled. This would increase the versatility of the loaded sled system and provide the ability to compare different types of motion control system. REFERENCES [1] Johnson, Jack L., “Design of Electro hydraulic Systems for Industrial Motion Control”, 2nd ed., 1995. [2] FitzSimons, P.M., and J. J. Palazzolo, “Part 2: Control”, ASME Journal of Dynamic Systems, Measurement, and Control, Vol. 118, No.3 pp 443448.September 1996. [3] Sohl, Garett A., and James E. Bobrow, “Experiments and Simulations on the Nonlinear Control of a Hydraulic Servosystem”, IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL.7, No.2, March 1999. [4] Yao,Bin, Bu, Fanping, Reedy, John and George T.-C. Chiu, “Adaptive Robust Motion Control Of Single rod Hydraulic Actuators”, IEEE/ASME Transaction on Mechatronics,Vol. 5,No.1,March 2000. [5] Meritt, H.E., Hydraulic control systems; New York: Wiley, 1967. [6] Kenji Okiayama and Ken Ichiryu, “Study of Pneumatic Motion base Control Characteristics”, Tokyo University of Technology, Tokyo, Japan. [7] Bolton, W., Pneumatic and Hydraulic Systems, Oxford; Boston: Butterworth Heinemann, 1997. [8] Stoll, Kurt, “New Developments in Pneumatics”,Festo AG & Co., Ruiterstr. 82, D73734 Esslingen, Germany. 59 [9] Canfield, Eugene B., Electromechanical Control Systems & Devices / Eugene B. Canfield, Huntington, N.Y.: R E Krieger Pub. Co. 1977, c1965. [10] Roth, Mary Ellen, “Electromechanical Actuation for Thrust Vector Control Applications/Mary Ellen Roth : Prepared for the National Aerospace & Electronics Conference Sponsored by Institute of Electrical & Electronics Engineers, Dayton Ohio,May1990”,pp 21-25, [Washington D.C]: NASA,1990. Web link: http:// purl.access.gpo.gov/GPO/LPS66770 [11] Chironis, Nicholas P., Machine devices and instrumentation: mechanical, electromechanical, hydraulic, thermal, pneumatic, pyrotechnic, photoelectric and optical /edited by Nicholas P. Chironis., New York : McGraw-Hill,1966. [12] Kamman, James, ME4710 Motion and Control: Selected Course Notes http://www.mae.wmich.edu/faculty/kamman/ME471course_notes.htm [13] FitzSimons, P.M., and J. J. Palazzolo, “Part 1: Modeling of a One Degree –of – Freedom Active Hydraulic Mount”, ASME Journal of Dynamic Systems, Measurement, and Control , Vol. 118, no.3 pp 439- 442.September 1996. [14] Gholamreza, Vossoughi, and Max Donath, “Dynamic Feedback Linearization for Electrohydraulically Actuated Control Systems”, ASME Journal of Dynamic Systems, Measurement, and Control , Vol. 117, No. 4, pp 468- 477,1996. [15] Luigi Del Re and Alberto Isidori, “Performance Enhancement of Nonlinear Drives by Feedback Linearization of Linear-Bilinear Cascade Models”, IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY,VOL.3 NO.3,SEPTEMBER 1995. [16] Yao,Bin, Bu, Fanping ,”Performance Improvement of Proportional Directional Control Valves Methods and Experiments”, . [17] Yao,Bin, Bu, Fanping, “Adaptive Robust Precision Motion Control of Single-Rod 60 Hydraulic Actuators with Time-Varying Unknown Inertia: A Case Study”. [18] Yao,Bin, Bu, Fanping ,”Nonlinear adaptive robust control of hydraulic actuators regulated by proportional directional control valves with dead band and nonlinear flow gains ”. [19] Dorf, Richard C., Bishop, Robert H., Modern Control Systems. 10th ed. Upper Saddle River: Pearson Prentice Hall, 2005. [20] Franklin, Gene F., J. David Powell, and Abbas Emami- Naeini. Feedback Control of Dynamic Systems.4th ed. Upper Saddle River: Pearson Prentice Hall,2002. [21] Cassell, Douglas A., Microcomputers and modern control engineering, Reston, Va.: Reston Pub. Co.,c1983. [22] Phillips, Charles L., and H. Troy Nagle, Digital Control System Analysis and Design, Englewood Cliffs, N.J.: Prentice Hall, c1995. [23] Gaines, B.R., “General System Identification- Fundamentals and Results “, In Klir,G.J.,Ed. Applied General Systems Research., pp. 91-104 New York, USA: Plenum Press,1978. [24] Sage, Andrew P., System Identification, New York Academic Press, 1971. [25] Kagiwada, Harriet H., 1937-, System Identification: Methods & Applications, Reading, Mass, Addison-Wesley Pub. Co. 1979. [26] Klerk, Elsa de, and Ian K. Craig, “A Laboratory Experiment to Teach Closed – Loop System Identification”, IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY,VOL.47.NO. 2, MAY 2004. [27] Alberts, Aaron, Blower Doug and John Karas, “Design of a Variable Sliding Mass Motion Control Test Machine”, Senior Design Project, Western Michigan 61 University, 2003. [28] Industrial Profile Systems, Parker Hannifin Corporation,2002. [29] Industrial Hydraulic Pumps, Motors and Fluid power System Products, Parker Hannifin Corporation,2002. 62 APPENDIX A.1: Comparison of Model Output with Actual System Response Comparision of Measured & Simulated Response for Input Command=3 V Comparision of Measured & Simulated Response for Input Command=3 V 1 0 Actual Response Simultated Response -1 0.8 -1.5 0.7 -2 -2.5 -3 0.6 0.5 0.4 -3.5 0.3 -4 0.2 -4.5 0.1 0 -5 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 Actual Response Simulated Response 0.9 Sled Position(in) Valve Response (V) -0.5 0.5 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 Comparison of Measured & Simulated Respnse for Input Command =4V Comparison of Measured & Simulated Respnse for Input Command =4V 0 Measured Response Simulated Response 1.2 -1 -2 Sled Position(in) Valve Respone(V) 1 -3 0.8 0.6 -4 0.4 -5 0.2 Measured Response Simulated response -6 0 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 0 Comparison of Measured & Simulated Respnse for Input Command =5V 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 Comparison of Measured & Simulated Respnse for Input Command =5V 0 1.8 Measured Response Simulated Response 1.6 Measured Response Simulated Response -1 1.4 1.2 Sled Position(V) Valve Response(V) -2 -3 -4 1 0.8 0.6 -5 0.4 -6 -7 0.2 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 0 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 63 Comparison of Measured & Simulated Respnse for Input Command =6V Comparison of Measured & Simulated Respnse for Input Command =6V 2 2 Measured Response Simulated Response 1.8 0 1.6 -1 1.4 Sled Position(in) Valve Response(V) 1 -2 -3 -4 1.2 1 0.8 -5 0.6 -6 0.4 -7 0.2 -8 0 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 Measured Response Simulated Response 0 Comparison of Measured & Simulated Respnse for Input Command =7V 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 Comparison of Measured & Simulated Respnse for Input Command =7V 0 2.5 Measured Response Simulated Response -1 2 -2 Sled Position(in) Valve Response -3 -4 -5 1.5 1 -6 -7 0.5 Measured Response Simulated Response -8 -9 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0 0.5 0 Comparison of Measured & Simulated Respnse for Input Command =8V 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 Comparison of Measured & Simulated Respnse for Input Command =8V 0 2.5 Measured Response Simulated Response -1 -2 2 Sled Position(in) Valve Response -3 -4 -5 -6 1.5 1 -7 -8 0.5 Measured Response Simulated Response -9 -10 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 0 0 0.05 0.1 0.15 0.2 0.25 Time(s) 0.3 0.35 0.4 0.45 0.5 64 A.2: Closed Loop Response of Loaded Sled System When Compensator pole is at 155 rad/s and Compensator zero is at 26 rad/s at Different Gains Sled Position Vs Time Gain=10 Tolerance=0.01" Input Valve Command Vs Time Gain=10 Tolerance=0.01" 10 10 9 8 Input Valve Command(V) 8 7 Sled Position(in) 6 4 2 6 5 4 3 0 2 1 -2 0 0.5 1 1.5 2 Time(s) 2.5 3 0 3.5 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 Sled Position Vs Time Gain=15 Tolerance=0.01" Input Valve Command Vs Time Gain=15 Tolerance=0.01" 10 10 9 8 8 6 Input Valve Command(V) 7 Sled Position(in) 4 2 0 -2 6 5 4 3 -4 -6 2 -8 1 -10 0 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 4 Sled Position Vs Time Gain=20 Tolerance=0.01" Input Valve Command Vs Time Gain=20 Tolerance=0.01" 10 10 9 8 8 7 4 Sled Position(in) Input Valve Command(V) 6 2 0 -2 -4 6 5 4 3 -6 2 -8 1 -10 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 0 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 65 A.3: Closed Loop Response of Loaded Sled System When Compensator pole is at 65 rad/s and Compensator zero is at 21 rad/s for Different Gains Sled Position Vs Time Gain =10 Tolerance=0.01" Input Valve Command Vs Time Gain =10 Tolerance=0.01" 10 9 10 8 7 Sled Position(in) Input Valve Command(V) 8 6 4 6 5 4 3 2 2 1 0 0 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 0 0.5 Input Valve Command Gain=20 Tolerance=0.01" 1 1.5 2 Time(s) 2.5 3 3.5 Sled Position Vs Time Gain=20 Tolerance=0.01" 10 10 9 8 7 6 Sled Position(in) Input Valve Command(V) 8 4 2 6 5 4 3 0 2 -2 1 -4 0 0.5 1 1.5 2 Time(s) 2.5 3 0 3.5 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 Sled Position Vs Time Gain =30 Tolerance =0.01" Input Valve Command Vs Time Gain=30 Tolerance=0.01" 10 10 9 8 8 7 4 Sled Position(in) Input Valve Command(V) 6 2 0 -2 -4 6 5 4 3 -6 2 -8 1 -10 0 0.5 1 1.5 Time(s) 2 2.5 3 0 0 0.5 1 1.5 Time(s) 2 2.5 3 66 A.4: Closed Loop Response of Loaded Sled System When Compensator pole is at 50 rad/s and Compensator zero is at 18 rad/s for Different Gains Sled Position Vs Time Gain =10 Tolerance=0.01" Input Valve Command Vs Time Gain =10 Tolerance=0.01" 10 9 10 8 7 Sled Position(in) Input Valve Command(V) 8 6 4 6 5 4 3 2 2 1 0 0 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 0 0.5 Input Valve Command Gain=20 Tolerance=0.01" 1 1.5 2 Time(s) 2.5 3 3.5 Sled Position Vs Time Gain=20 Tolerance=0.01" 10 10 9 8 7 6 Sled Position(in) Input Valve Command(V) 8 4 2 6 5 4 3 0 2 -2 1 -4 0 0.5 1 1.5 2 Time(s) 2.5 3 0 3.5 0 0.5 1 1.5 2 Time(s) 2.5 3 3.5 Sled Position Vs Time Gain =30 Tolerance =0.01" Input Valve Command Vs Time Gain=30 Tolerance=0.01" 10 10 9 8 8 7 4 Sled Position(in) Input Valve Command(V) 6 2 0 -2 -4 6 5 4 3 -6 2 -8 1 -10 0 0.5 1 1.5 Time(s) 2 2.5 3 0 0 0.5 1 1.5 Time(s) 2 2.5 3 67 A.5: LabVIEW Code for Phase Lead Closed Loop Control of the Loaded Sled System Figure A.5.1 LabVIEW Code for Creating DAQmx Tasks ,Initialize Analog Input Channels and Create Analog Output Channel 68 Figure A.5.2 LabVIEW code for Initializing Output Channels and Synchronizing with input channels. 69 Figure A.5.3 LabVIEW Code for Closed Loop Control 70 Figure A.5.4 LabVIEW Code for Recording Data