Optimal input for autoregressive model discrimination based on the Kullback’s discrimination information

Experiment design for dynamic system identification has attracted considerable attention to obtain the maximal information from the observed input/output data. Most studies in this area has been devoted to accurate parameter estimation within a specified model structure. However, the problem of primary importance in system identification might be to determine the model structure itself. From this point of view, the optimal input design problem is discussed in this paper for discriminating two rival autoregressive models with a controllable input efficiently. An optimal input is derived, which maximizes the time increment of the Kullback’s discrimination information to make the distance between two models as large as possible, i.e., to make the difference of the models more clearly. The applicability of the proposed input for efficient autoregressive model discrimination is exemplified by simulation studies. [In Japanese.]