Page 167 - Adaptive Identification and Control of Uncertain Systems with Nonsmooth Dynamics
P. 167

APPC of Strict-Feedback Systems With Non-linear Dead-Zone  163


                                            m 1r M  1    2      2  2  2
                                               2
                                  ≤− k 1 −       −                                     ˙
                                                        z + c 12r g z + r[g 1 (x 1 )N(ξ 1 ) + 1]ξ 1
                                                         1     M 11 2
                                             2c 11  4c 12
                                       rˆε 1 |z 1 |σ 11  σ 12r M θ ˜ 2 1  σ 13r M ˜ε 1 2  σ 12r M θ 1 ∗2  σ 13r M ε ∗2
                                                                                      1
                                    +            −        −         +         +
                                       ˆ ε 1 |z 1 |+ σ 11  2   2         2         2
                                                           m 1 e  à 1m

                                      c 11         2  z 1            2
                                    +     1 − 2tanh (  )            k (x 1 ) −  V d1
                                                                     1j
                                       2             ω 1    j=1 1 −¯τ 1
                                                     2 2  2
                                                                            ˙
                                  ≤− γ 1V 1 + ϑ 1 + c 12 γ g z + r[g 1 (x 1 )N(ξ 1 ) + 1]ξ 1
                                                    m 11 2
                                                         m
                                      c 11        2  z 1     e  ˜ ωà 1m  2
                                    +    (1 − 2tanh (  ))        k (x 1 )             (10.23)
                                                                  1j
                                       2            ω 1    1 −Èτ 1
                                                        j=1
                            where g i1 is the upper bounds of g i (¯x i ),and γ 1 and ϑ 1 are positive constants
                                              2

                            γ 1 = min 2(k 1 − m 1r /2c 11 − 1/4c 12 ),  1r M σ 12 ,  a1r M σ 13 ,  , ϑ 1 = σ 11r M +
                                              M
                                             ∗2
                                  ∗2
                            σ 12r M θ /2+σ 13r M ε /2. Since k ij (¯x i ) are bounded on arbitrarily large com-
                                  1
                                             1
                                                              2
                            pact set C i and the fact −1 ≤ 1 − 2tanh (z i /ω i ) ≤ 1 holds, the last term of
                                                          2
                                                               2
                                                            2
                            (10.23) is bounded. The term c 12r g z is also bounded as long as z 2 is
                                                          M 11 2
                            bounded (will be guaranteed in next step). Then for small ϑ 1, c 11, c 12,or
                                                 ˜
                            large γ 1,the errors z 1, θ 1, ˜ε 1 can be proved to be bounded according to
                            Lyapunov’s Theorem and Lemma 10.2 [12].
                            Step i (2 ≤ i < n). Consider z i = x i − α i−1,then
                                ˙ z i =˙x i −¨α i−1 = f i (¯x i ) + g i (¯x i )(z i+1 + α i ) + h i (t, ¯x i (t − τ i (t))) −¨α i−1
                                                                                      (10.24)
                            Then the following control laws are developed
                                                                           2
                                                 ˆ
                                                 θ isgn(z i )  T          ˆ ε z i
                                                                           i
                                 α i = N(ξ i ) k iz i +    (Z i )  i (Z i ) +         (10.25)
                                                          i
                                                   2η 2                ˆ ε i |z i |+ σ i1
                                                     i
                                                                         2 2
                                                2
                                                                         i
                                                         T
                                          ˙
                                         ξ i = k iz +  θ i ˆ |z i |   (Z i )  i (Z i ) +  ˆ ε z i  (10.26)
                                                i     2  i
                                                    2η
                                                      i               ˆ ε i |z i |+ σ i1

                                              ˙      |z i |  T
                                                                         ˆ
                                              ˆ
                                              θ i =   i  2    (Z i )  i (Z i ) − σ i2 θ i  (10.27)
                                                          i
                                                     2η
                                                       i
                                                    ˙                                 (10.28)
                                                    ˆ ε i =   ai |z i | − σ i3 ˆε i
                            where   i > 0,   ai > 0, k i > 0,η i > 0and σ i1 ,σ i2 ,σ i3 > 0 are design parame-
                            ters.
   162   163   164   165   166   167   168   169   170   171   172