Page 636 - Mathematical Techniques of Fractional Order Systems
P. 636

Enhanced Fractional Order Chapter | 20  607


                Note that the control law Eq. (20.28) is realizable only while fðX; tÞ and
             gðX; tÞ are well known.
                However, fðX; tÞ and gðX; tÞ are unknown and external disturbance
             dðtÞ 6¼ 0, the ideal control effort Eq. (20.28) cannot be implemented. We

             replace fðX; tÞ; gðX; tÞ by the fuzzy logic system fðX θ f Þ, gðX θ g Þ in a speci-


             fied form as Eqs. (20.14, 20.15), i.e.,
                                      T               T
                             fðX θ f Þ 5 ξ ðXÞθ f ; gðX θ g Þ 5 ξ ðXÞθ g ;  ð20:32Þ


                                           T
                Here the fuzzy basis function ξ ðXÞ depends on the fuzzy membership
             functions and is supposed to be fixed, while θ f ; θ g and are adjusted by adap-
             tive laws based on a Lyapunov stability criterion (Aguila-Camacho et al.,
             2014; Duarte-Mermoud et al., 2015; Sastry and Bodson, 1989).

                The optimal parameter estimations θ ; θ and are defined as:
                                             f   g

                                         sup      fðX θ f Þ 2 fðX; tÞ

                           f                xAΩ x
                          θ 5 arg min θ f AΩ f
                                                                      ð20:33Þ
                                         sup
                           g                xAΩ x
                          θ 5 arg min θ g AΩ g  gðX θ g Þ 2 gðX; tÞ
             where Ω f ; Ω g ; Ω p , and Ω x are constraint sets of suitable bounds on θ f ; θ g ; θ p ,


                                                                   θ f # M f ,
             and x respectively, and they are defined as Ω f 5 θ f




                                          θ p # M p , and Ω x 5 x jj # M x gf
             Ω g 5 θ g                                       j  x      where
                       θ g # M g , Ω p 5 θ p
             M f ; M g ; M p and M x are positive constants.
                Assuming that the fuzzy parameters θ f ; θ g , and θ p never reach the
             boundaries.
                Let us define the minimum approximation error,
                              h             i  h             i

                          ω 5 fðX; tÞ 2 fðX θ Þ 1 gðX; tÞ 2 gðX θ Þ u  ð20:34Þ


                                          f                  g

                               ~
                                          ~
                                                         ~



             and define the errors: θ f 5 θ f 2 θ ; θ g 5 θ g 2 θ , and θ p 5 θ p 2 θ .
                                        f          g              p
                Then, the equation of the sliding surface Eq. (20.17) can be rewritten as:
                              h             i  h              i

                   s ðqÞ  5 ω 1 fðX θ Þ 2 fðXθ f Þ 1 gðX θ Þ 2 gðXθ g Þ u


                                   f                 g

                          2 pðs θ p Þ 1 pðs θ Þ 2 pðs θ Þ 1 dðtÞ



                                        p       p
                               T
                                              T
                                      T
                              ~
                                             ~
                                     ~
                        5 ω 2 θ ξðsÞ 2 θ ξðXÞ 2 θ ξðXÞu               ð20:35Þ
                               p      f       g



                                      2 pðs θ Þ 1 dðtÞ
                                            p

             20.5 STABILITY ANALYSIS
             Consider the commensurate fractional order SISO nonlinear dynamic system
             Eq. (20.17) with control input Eq. (20.28), if the robust compensator u a and
             the fuzzy-based adaptive laws are chosen as:
                                             1
                                               T
                                      u a 52 B Pe                     ð20:36Þ
                                             r
   631   632   633   634   635   636   637   638   639   640   641