Page 122 - Mathematical Techniques of Fractional Order Systems
P. 122

110  Mathematical Techniques of Fractional Order Systems


               By rewriting (4.24) in the form

                             ð
                           W 2lnsÞ XsðÞ 2 x 0ðÞ=s 5 AX sðÞ 1 BU sðÞ   ð4:25Þ
            and taking the inverse Laplace transform afterwards reveals the fact that
            xtðÞ 5 x 1 tðÞ 1 x 2 tðÞ satisfies (4.17). Eq. (4.23) indicates that the solution of
            distributed order systems in the Laplace domain may be obtained by simply
            substituting the Laplace variable s in the solution of their corresponding inte-
                                                  α
            ger order systems with W 2lnsÞ 5  Ð  1  w αðÞs dα. In the following Lemma
                                   ð
                                           0
            the impulse response of distributed order integrator with a unitary weight
            function is presented. This result will be used to derive the analytical solu-
            tion of (4.17) later.
            Lemma 3: Impulse response of distributed order integrator with the unitary
            weight function w αðÞ 5 1 αA 0; 1ŠÞ is given by
                                     ½
                                 ð
                                       t
                                  itðÞ 5 e E 1 tðÞ;  t . 0            ð4:26Þ
            in which E 1 tðÞ is the exponential integral function as defined in (4.1).

               Proof: Noting the Laplace pair L t-s E 1 tðÞ 5  ln s 1 1Þ  which is presented
                                                       ð

                                                        s
            in Erdelyi et al. (1954, p. 178), and the fact that the Laplace transform of itðÞ
                                s dα 5
            is given by IsðÞ 5 1=  Ð 1 α  lns  in the case of a unitary weight function,
                              0       s 2 1
            the proof is easily obtained via the frequency shift property of the Laplace
            transform.
            Lemma 4: Let lW 21 :ðÞ denote the lower branch of Lambert W function
            defined by (4.2) and define

                                r 0 52 λlW 21 2e 2π21=λ =λ            ð4:27Þ
                                                    1
            for some λ .   21   .If s jj . r 0 then  j lnsj  , .
                           ð
                        lW 21 2e 2π21 Þ      j s 2 1j  λ
                                    2π2 1 λ
               Proof: Define g λðÞ 52  e  λ  . It can be easily verified that this function
            has the minimum of 2e 2π21  occurring at λ 5 1. Therefore, the inequality
                                             e 2π2 1 λ
                                     2π21
                                  2e     #2        , 0                ð4:28Þ
                                               λ
            holds for λ . 0. This, would result in

                         2N , lW 21 2e 2π21=λ =λ # lW 21 2e 2π21      ð4:29Þ

            as lW 21 :ðÞ is strictly decreasing. Therefore, considering the condition
            λ .    21 2π21 from (4.27) and (4.28) one could write
                lW 21 2eð  Þ
                                      1 , r 0 ,1 N                    ð4:30Þ
   117   118   119   120   121   122   123   124   125   126   127