Page 768 - The Mechatronics Handbook
P. 768

0066_Frame_C24  Page 16  Thursday, January 10, 2002  3:44 PM









                         If we consider the continuous, time-invariant, and linear state space model defined by equations (24.42)
                       and (24.43), with initial state x(k 0 ∆) = x 0 , we can use Eq. (24.44) to calculate the next value of the state:

                                                                   k ∆+∆
                                                                         (
                                                   (
                                                                    0
                                      (
                                     x k 0 ∆ + ∆) =  e A k ∆+∆−k ∆) x k 0 ∆) + ∫ k ∆  e A k ∆+∆−t) B u t() t  (24.90)
                                                            (
                                                                          0
                                                         0
                                                    0
                                                                                     d
                                                                    0
                         Furthermore, if a zero order hold is used, i.e., u(t) = u(k 0 ∆) for k 0 ∆ ≤ t < k 0 ∆ + ∆, we obtain
                                                        A∆
                                                                            (
                                                                     Ah
                                            x k 0 ∆ +(  ∆) =  e x k 0 ∆) + ∫ ∆ e dhB u k 0 ∆)   (24.91)
                                                           (
                                                                   0
                         And, if we know the state and the input at time k 0 ∆, the output is defined by Eq. (24.43):
                                                 y k 0 ∆(  ) =  Cx k 0 ∆) +  D u k 0 ∆)         (24.92)
                                                                      (
                                                             (
                         We can now conclude that given a continuous-time model with state space matrices {A, B, C, D}, and
                       we sample inputs and outputs every ∆ seconds then, the equivalent sampled data systems will be described
                       by the discrete-time state space model:
                                                x k∆ + ∆) =  A d x k∆(  ) + B d u k∆)           (24.93)
                                                                        (
                                                 (
                                                               (
                                                                        (
                                                     (
                                                    y k∆) =  C d x k∆) + D d u k∆)              (24.94)
                       where
                                                           Ah
                                         A d =  e ,  B d =  ∫ ∆  e dhB,  C d =  C,  D d =  D    (24.95)
                                               A∆
                                                         0
                         There are different methods to obtain A d  defined in (24.95), but a simple way to calculate this matrix
                       is to use Laplace transformation. This yields

                                                 A d =  e A∆  =  L { (  sIA) } t =∆             (24.96)
                                                                       1
                                                              1
                                                             –
                                                                      –
                                                                  –
                       Example 24.6
                       Consider the mechanical system of Example 24.1 on the page 4, that was described by the state space
                       model:

                                               x ˙ 1 t()  0  1  x 1 t()  0
                                                    =                 +    ft()                 (24.97)
                                                        K
                                                                         1
                                               x ˙ 2 t()  – ----- – D  x 2 t()  – -----
                                                             -----
                                                        M
                                                             M
                                                                         M
                       where f(t) is the external force, and where we can choose either the mass position, x 1 (t), or the mass
                       velocity, x 2 (t), of the mass, as the system output.
                         For the purpose of a numerical illustration, we set M = 1 kg, D = 1.2 N s/m, and K = 0.32 N/m.
                         The matrix A d  is obtained from (24.96), applying inverse Laplace transformation
                                       1            – 1         – 0.4∆  – 0.8∆  (  – 0.4∆  – 0.8∆ )
                                A d =  L   s   – 1       =     2e  –  e     2.5 e   –  e      (24.98)
                                      –
                                          0.32 s +  1.2        (  – 0.4∆  – 0.8∆  – 0.4∆ +  – 0.8∆
                                                                              –
                                                        t =∆  0.8 e   –  e  )   e    2e
                       ©2002 CRC Press LLC
   763   764   765   766   767   768   769   770   771   772   773