NONLINEAR CONTROL BASED ON T-S FUZZY MODEL Consider T-S fuzzy systems If z1 (t ) is Ai1 and z2 (t ) is Ai 2 and … and zq (t ) is Aiq , then x (t ) Ai x (t ) Bi u (t ) y (t ) Ci x (t ) For i 1, 2, , N , where Ai R nn , B R nm and Ci R pn (1) . N is the number of the subsystems and we call system (1) the ith subsystem. z (t )= z1 (t ) z2 (t ) zq (t ) is a vector of premise variables. Aij (i 1, N ; j 1, , q ) are fuzzy sets. Denote A () the membership function determining the membership of z j (t ) in the ij fuzzy set Aij . T A T-S fuzzy system actually represents a nonlinear system. Fuzzy system (1) can be represented by equivalently N x (t ) hi ( z (t ))[ Ai x (t ) Bi u (t )] i 1 N y (t ) h ( z (t ))C x(t ) i i i 1 where (2) q hi ( z (t )) j 1 q N Aij i 1 j 1 ( z j (t )) (3) Aij ( z j (t )) N We call hi ( z (t )) the fuzzy weights. Obviously, we have h ( z (t )) 1. i 1 i Consider affine nonlinear system x f ( x ) g ( x)u (3) n m where f ( x ) R n and g ( x ) R are two nonlinear function. Theorem 1: Suppose that f (x) is continuous and differentiable, then for any 0, there exists a T-S fuzzy model of system (1) or (2) such that N f ( x) g ( x )u hi ( Ai x Bi u ) i 1 for any x and u in a suitable neighbourhood. The computation of matrix Ai and Bi Step 1: Choose the operating points of x0i (i 1, , N ) . Usually, the equilibrium point x0 : f ( x0 ) 0 should be considered. Step 2: (1) For the equilibrium point x0, Ai where f ( x ) x and x0 f1 ( x ) x 1 f 2 ( x ) f ( x ) x1 x f n ( x ) x1 Bi g ( x0 ) f1 ( x) x2 f1 ( x ) xn f 2 ( x ) f 2 ( x ) x2 xn f n ( x ) f n ( x ) x2 xn (4) (2) for other operating points: al ,i f l ( x ) i 0 a1,Ti Ai anT,i where f l ( x0i ) x0iT f l ( x0i ) i 2 0 x and Bi g ( x0i ) x0i , (l 1, 2, , n) f l ( x ) x 1 f l ( x ) f l ( x ) x2 f l ( x ) xn (5) Example 1: Consider free-body of a cart with an inverted pendulum system Fig. 1: Free-bosy of a cart with an inverted pendulum system The dynamic system of the inverted pendulum system can be described in form x f ( x ) g ( x )(u ( x )) (6) with 0 x2 a cos x g sin x 1 1 2 2 4l / 3 mla cos x1 4l / 3 mla cos x1 f ( x) , g ( x) 0 x 4 (1 / 2) mag sin(2 x1 ) 4a / 3 2 2 4 / 3 ma cos x 4 / 3 ma cos x 1 1 ( x ) f c mlx22 sin x1 and For the nonlinear system (6), we propose the following two rules to construct a T-S fuzzy system: Rule1: If x1(t) is about 0, then x A1 x B1u Rule 2: If x1(t) is about / 4 , then x A2 x B2u That is we choose two operating points: x 0 0 0 0 1 0 T and x / 4 0 0 0 2 0 T Next, we compute the matrices of A1, A2, B1 and B2 based on (4) and (5) 2 For the operating point x0 / 4 0 0 0 , based on (5), we have T f 2 f 2 0 0 x x 1 0 1 1 f1 ( x ) , f 2 ( x ) 0 , f 3 ( x ) , f 4 ( x ) 0 0 0 0 0 1 0 0 0 Therefore, a1,2 f1 ( x ) 2 0 Since, f1 ( x02 ) x02T f1 ( x02 ) 2 2 0 x x 0 1 0 0 2 0 T a2,2 f 2 ( x ) 2 0 f 2 ( x02 ) x02T f 2 ( x02 ) 2 2 0 x x02 g sin x1 f 2 ( x) 4l / 3 mla cos 2 x1 The computation results are: 0 17.2941 A1 0 1.7249 1 0 0 0 14.3077 0 0 0 , A2 0 0 0 1 0 0 0 0.9723 1 0 0 0 0 0.1765 0.1147 0 0 0 , B2 , B1 0 0 0 0 1 0 0 0 0.1176 0.1081 We use the membership functions of the form 1 1 / (1 e 14( x1 /8) ) h1 ( x1 (t )) 1 e 14( x1 /8) and h2 ( x1 (t )) 1 h1 ( x1 (t )) Then the T-S fuzzy model consisting two subsystems for nonlinear system (6) can be represented as x h1 ( x1 )( A1 x B1 (u )) h2 ( x1 )( A2 x B2 (u )) 2 hi ( x1 )( Ai x Bi (u )) i 1 (A1) Nonlinear Control Scheme Based on T-S Fuzzy Model Consider a homogenous linear system x (t ) Ax (t ) (7) where x R n is sate vector and A R nn is a square constant matrix. Definition 1: If all the eigenvalues of matrix A have negative real part, then we call matrix A is a (asymptotical) stable matrix, and the system (7) is asymptotically stable at equilibrium state x = 0. Theorem 1: Square matrix A is stable if and only if for any symmetrical positive definite matrix Q, the following Lyapunov (matrix) equation AT P PA Q (8) has a symmetrical positive definite matrix solution of P. An equivalent statement of Theorem 1 is: Square matrix A is stable if and only if Lyapunov inequality AT P PA 0 has a symmetrical positive definite matrix solution of P. Now consider a T-S fuzzy homogenous model described by N x (t ) hi ( z (t )) Ai x (t ) i 1 (9) Theorem 2: A sufficient condition for the T-S fuzzy system (9) being asymptotically stable at 0 is that there exists a common symmetric positive definite matrix P such that AiT P PAi 0 (10) holds for i 1, 2, , N . Remark: A necessary condition for the existence of a common symmetrical positive definite P satisfying (10) is that each Ai is asymptotically stable. Theorem 3: There exists a common symmetric positive definite matrix P such that (10) holds, then the matrices s A k 1 ik are asymptotically stable, where ik {1, 2, , N } and s 1, 2, , N . Example: Given a homogenous T-S fuzzy system (9) with two subsystems: 1 4 1 0 A1 , A2 0 2 4 2 Obviously, both A1 and A2 are asymptotically stable, but there is no common positive definite matrix P such that (10) holds because 1 4 1 0 2 4 A1 A2 0 2 4 2 4 4 is unstable since its eigenvalues are {1.1231, -7.1231}. Now consider the T-S fuzzy system (2), we plan to design a state feedback control scheme such that the close-loop system is asymptotically stable. Assumption 1: The pair { Ai , Bi }(i 1, 2, , N ) is controllable. (i.e. every subsystem of the T-S fuzzy system is controllable.) Under Assumption1, the state feedback controller is designed as N u hi ( z (t )) K i x i 1 (11) where the control gains Ki are chosen such that Ai Bi K i (i 1, , N ) is stable. Substituting (11) into the state equation in (2) leads to the closed-loop system: N N N N N i 1 j 1 i 1 j 1 j 1 x hi ( z (t ))[ Ai x Bi h j ( z (t )) K j x ] hi ( z (t ))[ h j ( z (t )) Ai x h j ( z (t )) Bi K j x] N N N N i 1 j 1 i 1 j 1 hi ( z (t )) h j ( z (t ))( Ai Bi K j )x hi ( z (t ))h j ( z (t ))( Ai Bi K j ) x That is, the closed-loop system is N N x hi h j ( Ai Bi K j ) x i 1 j 1 (12) Theorem 4: A sufficient condition for the closed-loop system (12) to be asymptotically stable at x = 0 is that there exists a symmetric positive definite matrix P such that the following conditions are satisfied: ( Ai Bi K i )T P P ( Ai Bi K i ) 0, (i 1, 2, , N ) and GijT P PGij 0, (i j N ) (12) (13) where Gij ( Ai Bi K j ) ( Aj B j Ki ) Example 2: Consider the inverted pendulum nonlinear system (6) and the corresponding T-S fuzzy model (A1), and based on the T-S fuzzy model (A1), we design the state feedback as u h1 K1 h2 K 2 (14) The two gains K1 and K2 are chosen such that the eigenvalues of A1 B1 K1 and A2 B2 K 2 are placed to a1 1.0970 2.1263 2.5553 3.9090 T and a2 1.5794 1.6857 2.7908 3.6895 T respectively. Now by MATLAB code: a1=[-1.0970 -2.1263 -2.5553 -3.9090]’; a2=[-1.5794 -1.6857 -2.7908 -3.6895]’; K1=place(A1, B1, a1); K2=place(A2, B2, a2); We obtain K1 294.8755 73.1208 13.4726 27.3362 T and K 2 440.3915 118.5144 19.1611 35.5575 T Let G12 A1 B1 K 2 A2 B2 K1 and using LMI toolbox in MABLAB to solve the following LMIs ( A1 B1 K1 )T P P ( A1 B1 K1 ) 0 ( A2 B2 K 2 )T P P ( A2 B2 K 2 ) 0 G12T P PG12 0 provides a common solution of P: 54.9580 15.6219 6.9389 12.1165 15.6219 4.5429 2.1011 3.5488 P 6.9389 2.1011 1.3972 1.7978 12.1165 3.5488 1.7978 2.9375 So, the controller (14) can be applied to both the fuzzy model and the nonlinear system. Response of state x1 Response of state x2 Response of state x3 Fig. 2: The controller applied to the T-S fuzzy model Response of state x4 Response of state x1 Response of state x2 Response of state x3 Fig. 3: The controller applied to the nonlinear system Response of state x4