H∞ CONTROLLER DESIGN FOR NETWORKED CONTROL SYSTEM WITH SHORT UNCERTAIN TIMEDELAY Author*a, Authora, Authorb aDepartment of Mathematics and Physics North China Electric Power University, Beijing, 102206, PRC bDepartment of Automation North China Electric Power University, Beijing, 102206, PRC ABSTRACT: For the NCS with disturbance, we point out a sufficient condition ensuring the system meeting prescribed H∞ performance γ and a state feedback controller ensuring the system meeting optimal performance index. Numerical examples are given to illustrate the effectiveness of our results via TrueTime toolbox. KEY WORDS: networked control system (NCS); time-delay; asymptotical stability; state feedback; H∞ performance 1. INTRODUCTION In modern industrial control systems, sensors, controllers and plants are often connected over a network medium, which are called networked control systems (NCSs). NCS has many advantages, such as low cost, and simple installation, thus NCS has been becoming increasingly important in practical applications (Zhang W. (2001)). However, there are several drawbacks in NCS such as packet dropout and network-induced delay (Dang X.D. (2008), Walsh G.C. (2002)) and so on. The delay can degrade the performance of NCS and even cause the system unstable. Therefore, it is necessary to study the stability of NCS with network-induced delay (Richard J.P. (2003)). 2. NETWORKED CONTROL SYSTEMS MODELING Consider a continuous time linear time-invariant plant described by: { 𝑥̇ (𝑡) = 𝐴𝑥(𝑡) + 𝐵1 𝑢(𝑡) + 𝐵2 𝑤(𝑡) 𝑦(𝑡) = 𝐶1 𝑥(𝑡) + 𝐶2 𝑤(𝑡) ( 1) where x(t), u(t), w(t), y(t) are the state vector of plant, input vector of plant, disturbance of plant and the output of plant respectively. A, B1 , B2 , C1 , C2 in equation (1) are known constant matrices with compatible dimensions. Discretize the continuous plant into the following model during the interval [kh, kh+h], where k is a non-negative integer. 3. STATE FEEDBACK CONTROLLER DESIGN FOR NCS WITHOUT DISTURBANCE Consider the asymptotical stability of NCS without disturbance. Theorem 3.1 Consider a continuous time linear time-invariant plant described by (2.1) with state feedback controller u(t) = K x(t), the sampling period h is fixed and known, the overall delay in the system satisfies dk ∈ [0, h]. The closed-loop system is asymptotically stable if there exist symmetric positive definite matrices P, Q, feedback gain matrix K , and a scalar α > 0, such that the following matrix inequality holds. 4. STATE FEEDBACK CONTROLLER DESIGN FOR NCS WITH DISTURBANCE Figure 1 shows a schematic of the feedback controller design while the results are found in Table 1. Figure 1 Diagram of feedback controller design. Description Case 1 Initial guesses n tr C10 C01 1000 1 -0.5 g C10 C01 Final values n tr 0.5 0.2 118 0.41 -0.38 0.12 g 0.23 10 1 -0.5 0.5 0.2 75 0.17 3 100 10 -0.5 0.5 0.2 97 1.78 -0.49 0.66 0.09 4 100 0.1 -0.5 0.5 0.2 100 1.00 -0.50 0.50 0.20 5 100 1 -1 0.5 0.2 99 0.90 -0.60 0.64 0.25 6 100 1 -0.25 0.5 0.2 102 1.09 -0.42 0.37 0.15 7 100 1 -0.5 5 0.2 97 1.30 -0.76 1.06 0.22 8 100 1 -0.5 0.05 0.2 109 0.61 -0.39 0.20 0.21 9 100 1 -0.5 0.5 0.4 102 0.54 -0.60 0.48 0.30 10 100 1 -0.5 0.5 0.1 99 1.30 -0.47 0.53 0.15 all high 11 1000 10 -1 5 0.4 101 0.05 -1.38 1.00 0.46 all low Table 12 10 0.1 -0.25 0.05 0.1 0.98 0.00 13 90 the 10 10 -1 5 0.4 91 15.67 -1.69 2.94 0.00 14 1000 0.1 -1 5 0.4 80 26.51 -2.18 2.20 0.03 15 1000 10 -0.25 5 0.4 100 1.00 -0.50 0.50 0.20 10 -1 0.05 0.4 241 0.00 -5.E+09 2.E+07 0.27 10 -1 5 0.1 97 1.34 -1.15 1.87 0.26 5. 1 low, 4 highs 1 low or high at a time 2 1 Comparison between initial guesses and CONCLUSION 16 1000 17 1000 -8.11E+09 1.36E+11 0.01solutions. -0.30 final 0.00 2 lows, 3 highs 18 10 0.1 0.4 76 0.03 -6565 been 5.E+05 0.00 as a discrete time system with The networked control system-1 with 5uncertain time-delay has modelled 19 10 10 -0.25 5 0.4 96 10.34 -0.95 1.15 0.00 without disturbance which make uncertain parameters. A state feedback controller has been designed for NCS 20 10 10 -1 0.05 0.4 96 7.56 -1.61 3.92 0.01 the system 21asympotically stable. For NCS with disturbance, a controller has been presented which stabilizes 10 10 -1 5 0.1 89 16.88 -1.89 2.45 0.00 the closed-loop system with prescribed and optimal performance index. The numerical examples in the end of 22 1000 0.1 -0.25 5 0.4 100 1.03 -0.62 0.67 0.24 23 1000 0.1 -1 0.05 0.4 216 0.00 -1999 92.24 0.27 this paper have illustrated the effectiveness of our results. 24 1000 0.1 -1 5 0.1 80 26.00 -1.74 1.50 0.03 25 1000 10 -0.25 0.05 0.4 103 1.12 -0.39 0.33 0.13 26 1000 REFERENCES 10 -0.25 5 0.1 100 1.00 -0.50 0.50 0.20 0.27 27 1000 10 -1 0.05 0.1 117 0.32 -0.42 0.13 28 10 0.1 -0.25 5 0.4 101 0.59 -0.01 76.39 31 10 10 -0.25 0.05 0.4 76 0.26 3 lows, 2 highs 0.84 Zhang W, Branicky MS, Phillips SM (2001) Stability of networked control system. 29 10 0.1 -1 0.05 0.4 97 1.03 -12.40 35.53 0.30 Magazine,30 21, 84-99. 10 0.1 -1 5 0.1 91 14.46 -241.24 5.E+02 0.00 -4997.55 73931.58 IEEE Control Systems 0.00 Gahinet P, Nemirovski A10(1995) LMI5 Control Toolbox Users-0.72 Guide. 1.06 USA:MathWorks. 32 10 -0.25 0.1 97 5.69 0.00 33 4 lows, 1 high 10 0.05 0.1 114 11.08 -0.55 0.30 0.00 Dang XD (2008) 10 Stability and-1 controller design of networked control system with time-varying delays and data 34 1000 0.1 -0.25 0.05 0.4 216 0.00 -0.49 0.02 0.17 packet dropout. Control and Decision Conference, Shandong, 1376-1379. 35 1000 0.1 -0.25 5 0.1 86 25.85 -1.24 0.97 0.00 36 1000 0.1 -1 0.05 0.1 129 0.19 -0.44 0.08 0.27 37 1000 10 -0.25 0.05 0.1 225 0.00 -0.24 0.00 0.05 38 10 0.1 -0.25 0.05 0.4 76 0.50 -1949 2.E+04 0.00 39 10 0.1 -0.25 5 0.1 77 0.29 -5819.84 9.E+04 0.00 40 10 0.1 -1 0.05 0.1 0 905.02 -15.35 0.59 0.00 41 10 10 -0.25 0.05 0.1 101 1.85 -0.36 0.43 0.00 42 1000 0.1 -0.25 0.05 0.1 248 0.00 -0.22 0.00 0.09