Wavelet-based frequency band weighting approach to control-relevant process identification

Control-relevant identification (CRID), which emphasizes efficiently processing experimental data in terms of control design requirements, shows great difference from conventional identification. In this paper, a wavelet-based frequency band weighting (FBW) approach to CRID is presented. First, the frequency aliasing problem resulting from wavelet decomposition is discussed, and modified wavelet decomposition algorithms are given to guarantee correct decomposition of experimental data into different frequency bands. Then, the FBW approach is used to integrate the control design objective into the identification procedure by weighting these frequency bands in an appropriate way. Simulation examples are provided to illustrate the procedure of using the proposed approach to build suitable process models for feedback control design