International Carpathian Control Conference ICCC’ 2002 MALENOVICE, CZECH REPUBLIC May 27-30, 2002 CLOSED LOOP CONTROL OF THE SYSTEM WITH THE MODES OF DIFFERENT DYNAMICS AND DAMPING Petr NOSKIEVIČ Department of Control Systems and Instrumentation, VSB - Technical University of Ostrava, Ostrava, Czech Republic, petr.noskievic@vsb.cz Abstract: The paper deals with the design and tuning of the PID controller for the closed loop control of the system with modes of different dynamics and different damping. The influence of the chosen controller on the properties of the controlled system is shown using the critical gain and step responses. The use of the presented rules is demonstrated on the tuning of the controller for the position control of the electrohydraulic drive. Key words: Controller tuning, damping, hydraulic drive. 1 System analysis Closed loop control of the five order system with two second order terms with the different dynamics and damping Figure 1 is the topic of this paper. This structure of the controlled system is typical for the actuator systems and position control especially for the closed loop position control using the hydraulic drive. But the presented results may be used also by the tuning of the controller of the closed loops of the similar structure and different kind. The controlled 5th order system is characterised by the two second order terms and one integrator. The first second order term is well damped, the relative damping ration is approximately 0.9, the second term is badly damped with the relative damping ratio from 0.05 to 0.2, the following derivations were made for the value 0.1. The dynamics of the is determined by two values – the time constant Tsv and the time constant TM, which can be calculated from the natural frequencies Tsv = 1 ω sv = 1 1 1 , TM = . = 2πf sv ω M 2πf M 235 (1) low good damping 1 2 2 Tsv s + 2ξ svTsv s + 1 x2 1 TM2 s 2 + 2ξ M TM s + 1 x1 KO s x Figure 1. The analysed and controlled system The dynamics of the open loop, the choice of the right controller and the step response of the closed loop controlled systems depend on the ratio of the natural frequencies κ= f sv . fM (2) This ratio is greater if the well damped term is faster than the second system with the low damping. In the practice the ratio κ could be also lower than 1. The Figure 2 shows the step responses of the closed loop for three different ratios κ=0.25, κ=0.7, κ=3 for the increasing gain of the proportional controller. In the case of κ=0.7 it is possible using only the proportional controller to make a necessary compromise between the swiftness of the response and the allowable overshot and to achieve a damped step response without the overshot. In the other cases the system starts to oscillate by the increasing gain or the step response is too slow [NOSKIEVIČ 1995, DIETER 1998]. 2 Controller structure To achieve a good behaviour of the closed loop system it is necessary to select suitable controller in dependence on the ratio κ - the proportional controller (P), the proportional controller with derivative part (PDT1) G ( s) = K R 236 Td s + 1 , Ts + 1 (3) the proportional controller with the time constant (PT1) G(s) = K R 1 Ts + 1 (4) or the state controller. Fig.2 Normalised step responses of the closed loop systems with proportional controller for κ=0.25, κ=0.7, κ=3 (from left). Let the critical gain – the marginal stability is the criteria for the choice of the controller. The suitable controller which enables as fast as possible but well damped behaviour with high amplification will be chosen for the closed loop system characterised by different ratios κ. The time constants of the controllers can be dimensioned using the frequency response and using the numerical calculation of the margin stability. Proportional controller allows to set the high gain only if the ratio κ is in the range from 0.5 to 1, Figure 3. Otherwise the step response is slow or the system starts to oscillate. For the ratio κ lower that 0.5 and greater than 1 it is necessary to find a different controller. If we set the time constant Td as a multiply of the time constant Tsv of the good damped term and parameter a Td = a * Tsv , Figure 3 Critical gain of the 5th order closed loop system with the proportional controller we can set the parameter a in dependence on the critical gain. Figure 4 shows the calculated normalised critical gain for the PDT1 controller and varying time constant Td and ratio κ in the range 0÷3. The highest influence of this type of controller on the margin stability can be observe for the systems characterised by the ratio κ in the range 0 ÷ 0.5 and for the value of parameter a = 1. In this range of κ and a it is the allowable critical gain 237 higher than the highest gain when we use the proportional controller. The time constant T is set T=0.1*Td. Figure 4. Normalised critical gain for the PDT1 controller (left), for the PT1 controller (right) Figure 5. Maximum critical gain surface for three controllers and different time constant setting Figure 4 (right) shows the normalised critical gain for the different setting of the time constant T of the PT1 controller. For the time constant T=2*TM is the critical gain nearly constant for κ ≥ 2. For higher time constant T is the critical gain higher but the step response is not so swift because the controller is working like the low pass filter with very low bandwidth. As a conclusion we can create the maximum value surface shown in Figure 5 and summarize the rules for the controller choice and tuning in the Table 1 and Figure 6. Table 1. Transfer functions of the controllers and recommended values of the ration κ. PDT1 controller P controller PT controller State controller G( s ) = K R TD = Tsv = TD s + 1 , Ts + 1 G( s ) = K R 1 , T = 0.1TD 2πf sv κ = 0 ÷ 0 .5 κ = 0 .5 ÷ 1 238 1 , Ts + 1 1 T = 2TM = πf M G( s ) = K R κ = 1÷ 3 u = sw − r T x κ ≥3 Figure 6 shows the curves of the critical gain normalized to the cylinder natural frequency for the three types of controller and different frequency ratios κ. The maximum value of the critical gain determines the type of the appropriate controller. After the choice of the right controller dependence on the ratio κ by the use of the plot in Figure 6, we can set the time constants according to the Table 1 and for the final tuning of the gain apply the root locus method supported by MATLAB designer for example. Figure 6. Normalized critical gain for controller of a different kind and different ratio κ. 3. Application examples of closed-loop position control Closed-loop position control is a very frequent industrial application of linear hydraulic servovalve controlled drives. The dynamic behaviour of the closed loop controlled linear hydraulic in Figure 7 depends on the ratio κ of the natural frequencies of the servovalve fsv and cylinder fM . F K F (TF s + 1) cylinder servovalve w u e G( s ) K sv K Q KM TM2 s 2 + 2ξ M TM s + 1 Tsv2 s 2 + 2ξ sv Tsv s + v Figure 7. Block diagram of the closed-loop position control system 239 1 s x The result of the controllers design for two valve/cylinder drives characterised by the parameters Ksn = 66.67 Vm-1, KM= 1.0575e+3 m-2, Ksv = 0.1 V-1, Kq = 5.0e-4 m3s-1 , ξm=0.1, ξsv =0.9, and the different dynamics given by the time constants Tsv = 0.008 s, Tm = 0.002 s, κ=0.25 and Tsv = 0.008 s, Tm = 0.02 s, κ = 2.5 is shown in Fig.8. The first system is characterised by the ratio κ = 0.25 and the appropriate controller is the PDT1- controller according to Table 1. The fastest step response 1 in Fig.8 without the overshot was obtained with the controller gain Kr = 13. The P controller with the same gain gives the step response 2 with the overshoot, and the slowest step response 3 without overshot is for the lower gain of the P-controller. To achieve the best response of the second system characterised by κ=2.5 it is necessary to use PT1 controller – step response 1 in Figure 8 right, response 2 is by the use of the P controller with the same gain, response 3 – P-controller with the lower gain. Fig.8 Step responses for the two drives and different controller Conclusions Controller tuning of the 5th order systems with second order subsystems characterised by different dynamics and different damping was described in the paper. The results have general importance and may be applied to the control of fluid power systems, mechatronic systems and other systems. The damping ratio of the both systems has influence on the shape of the curve of the critical gain. The change of the lower damping ratio causes the modification of the curve of the critical gain of the PDT1 controller, the change of the higher damping ratio has influence on the curve of the critical gain by the use of the PT1 controller. If the actual values of the analysed system are different from the supposed values, the critical gain curves can be recalculated in the same way. References NOSKIEVIČ, P., 1995. Auswahlkriterium der Reglerstruktur eines lagegeregelten elektrohydraulischen Antriebes. Ölhydraulik und Pneumatik, 39 ( 1995 ), č.1, str. 49 – 51.ISSN 0341-2660. DIETER,M., 1998. Ein Beitrag zur systematischen Reglerauslegung positionsgeregelter Ventil - Zylinder - Antriebe.1.Internationales Fluidtechnisches Kolloquium. Band1, S.331-344, 17.-18.3.1998, Aachen, ISBN 3-89653-242-1. 240