Page 105 - Control Theory in Biomedical Engineering
P. 105

92    Control theory in biomedical engineering



                                                                     ρT  ∗
                                    u ∗
                                               ∗
            0
           λ ðtÞ¼ 1+ λ 1 ðtÞ δ + a 1 ð1 e  Þ + λ 2 ðtÞnT + λ 1 ðt + τÞχ  μT    ,
                                                                 ∗
            1                                            ½0, t f  τŠ
                                                                    η +T  ∗
                                                     u ∗
            0
                                     ∗
                                           ∗
                                ∗
           λ ðtÞ¼ 1+ λ 2  r 2 +2r 2 β T + nE + c 1 N + a 2 ð1 e  Þ
            2                 1
                                      "                   #
                                        ρE T  ∗  ρE ∗
                                          ∗
                 + λ 3 c 2 N + χ ½0, t f  τŠ 1 ðt + τÞ     + μE ∗  ,
                               λ
                       ∗
                                            ∗ 2
                                       ðη +T Þ  η +T  ∗
                          
                        u ∗
            0
                                         ∗
           λ ðtÞ¼ λ 2 c 1 T  λ 3 r 3  2r 3 β N  c 2 T  a 3 ð1 e  Þ  γ,
                                    ∗
                     ∗
            3                    2
                                        ∗
                                                     ∗
            0
                            ∗
           λ ðtÞ¼ λ 1 ðtÞa 1 e  u ∗ E + λ 2 ðtÞa 2 e  u ∗ T + λ 3 ðtÞa 3 e  u ∗ N + λ 4 ðtÞd 1 ,  ð9Þ
            4
          with transversality conditions
                                                 (
                                                   1 if t ½0, t f  τŠ,
               λ i ðt f Þ¼ 0, i ¼f1,2,3,4g and χ  ¼                    (10)
                                          ½0, t f  τŠ
                                                   0 otherwise:
          Furthermore, the following properties hold

                                    λ 4                   λ 1 s 1
                                            ∗
                       ∗
                      v ¼ min v max ,   ,w ¼ min w max ,       :       (11)
                                    B v                   B w
             Proof. The adjoint equations and transversality conditions can be
          obtained by using Pontryagin’s minimum principle with delay in state
          (Pontryagin et al., 1962) such that
                             ∂H              ∂H
                     0          ðtÞ χ
                    λ ðtÞ¼                ðtÞ   ðt+ τÞ, λ 1 ðt f Þ¼ 0,
                     1               ½0, t f  τŠ
                             ∂E              ∂E τ
                            ∂H               ∂H
                     0         ðtÞ χ
                    λ ðtÞ¼                ðtÞ   ðt+ τÞ, λ 2 ðt f Þ¼ 0,
                     2               ½0, t f  τŠ
                             ∂T              ∂T τ
                             ∂H                                        (12)
                     0
                    λ ðtÞ¼      , λ 3 ðt f Þ¼ 0,
                     3
                            ∂N
                            ∂H
                     0
                    λ ðtÞ¼     , λ 4 ðt f Þ¼ 0:
                     4
                             ∂u
             The optimality system consists of the state system (3) coupled with the
          adjoint system (9) with initial and transversality conditions together with the
   100   101   102   103   104   105   106   107   108   109   110