Page 171 - Adaptive Identification and Control of Uncertain Systems with Nonsmooth Dynamics
P. 171

APPC of Strict-Feedback Systems With Non-linear Dead-Zone  167


                                                           m n e  à nm
                                      c n1        2  z n             2
                                    +     1 − 2tanh (  )            k (x)
                                                                     nj
                                      2             ω n     j=1 1 −¯τ n
                                                                ˙
                                  ≤− γ nV n + ϑ n +[g n (x)dN(ξ n ) + 1]ξ n
                                                           m n e  à nm

                                      c n1        2  z n             2
                                    +     1 − 2tanh (  )            k (x)             (10.46)
                                                                     nj
                                      2             ω n     j=1 1 −¯τ n
                            where γ n and ϑ n are positive constants, which is given as
                                                                               ∗2        ∗2
                            γ n = min 2(k n − m n /2c n1 ),  n σ n2 ,  an σ n3 ,  ,ϑ n = σ n1 +σ n2 θ /2+σ n3 ε /2.
                                                                               n        n
                            Similar to the analysis in the previous steps, the last term of (10.46)is
                            bounded since the functions k nj (x) are bounded on any compact set C n and
                                         2
                            −1 ≤ 1 − 2tanh (z n /ω n ) ≤ 1 holds, which can guarantee the boundedness
                            of z n, θ n, ˜ε n for small enough ϑ n, c n1,orlarge γ n.
                                  ˜
                            10.3.2 Stability Analysis
                            In above analysis, Nussbaum functions N(ξ i ) are included in the Lyapunov
                            functions. Therefore extra efforts should be made to prove the system sta-
                            bility and guarantee the prescribed tracking control performance (10.4).
                            The following theorem states the main results of this chapter:


                            Theorem 10.1. Consider system (10.1) with unknown non-linear dead-zone
                            (10.2), the control is given by (10.37)–(10.40), then for any bounded initial condi-
                                θ i (0) ≥ 0, ˆε i (0) ≥ 0 and −δμ(0)< e(0)< δμ(0), there exist control feedback
                            tion ˆ                                ¯
                            gains k i fulfilling (10.47)suchthat

                                   ⎧        2
                                          m 1 r  1
                                   ⎪ k 1 ≥  M  +
                                          2c 11  4c 12
                                   ⎨
                                          m i  1        2
                                     k i ≥  +    + c i−1,2g i−1,1 ,  i = 2,··· ,n − 1  (10.47)
                                          2c i1  4c i2
                                   ⎪
                                   ⎩      m n       2
                                     k n ≥   + c n−1,2g n−1,1
                                          2c n1
                             i) All signals in the closed-loop system remain semi-globally bounded;
                            ii) The tracking control with prescribed performance condition (10.4)ispreserved.

                                                                                   ˜
                            Proof. For any given initial condition compact set 
 0 = z i (0),θ i (0), ˜ε i (0),
                                    !
                            i ≤ 1,···n , we can always construct a larger compact set 
 than 
 0 com-
                                        ,i = 1,··· ,n, in which the NN approximation is valid and
                            prising C i ,
 z i
                            the functions k ij (¯x i ) are bounded. We denote G i (x) = g i (¯x i ),i = 1,··· ,n−1,
                            G n (x) = g n (x)d,and α = max{r M ,1}, which are also bounded functions
   166   167   168   169   170   171   172   173   174   175   176