PURPOSE:To improve controllability by dividing an input/out space to a process into plural areas corresponding to the non-linear characteristic of the process, and re-applying an AR model on each divided space. CONSTITUTION:A fuzzy model processing part 3 identifies the input/output characteristic of the process in a closed loop state, and divides it to plural areas. An M system signal supply part 301 supplies a pseudo noise for identification to process input (x). An AR model function part 302 identifies the AR model based on an input signal (x) and output (y), and generates data D in an open loop based on the above identification. An area division function part 303 executes division so as to minimize an evaluation error function based on pseudo open loop data. A data separation function part 304 separates boundary data in a divided area into two areas, for example, at the crossing point of a fuzzy membership function. A data reconfiguration function part 305 reconfigures the AR model in each area based on divided data, and reconfigures pseudo data in each area based on the re-configured AR model.
展开▼
