WAVELET-BASED MODULATION IN CONTROL-RELEVANT PROCESS IDENTIFICATION

The identification of models from operating data for process controlle r design requires that the information from the process be extracted i n pieces that are localized in both time and frequency. Such an extrac tion process would allow the separation of valuable signal information from the effects of nonstationary disturbances and noise. The wavelet transform provides an efficient approach for such a decomposition, wh ich is organized in a multiscale, hierarchical fashion. By using the m ethod of modulating functions in conjunction with the wavelet decompos ition, it is demonstrated that recursive state-space models, which are multiscale in character and suitable for the design of model-predicti ve controllers, may he readily constructed with lower levels of modeli ng error than yielded by traditional techniques. The method is especia lly suitable for the identification of time-varying and nonlinear mode ls, where the nonlinear process is represented by a set of linear mode ls. The multiscale character of the wavelet basis makes it particularl y suitable for multirate, multivariable processes. A series of example s illustrates various aspects of the proposed approach and its inheren t advantages.