Application of a general multi-model approach for identification of highly nonlinear processes-a case study

Abstract

An identification method for highly nonlinear processes is proposed based on a multi-model approach and Kolmogorov-Gabor polynomials. Owing to the large number of possible terms in this general model structure, the significant terms are selected by several statistical test procedures leading automatically to a minimal-order model realization. The performance of this method is evaluated in an in-depth case study using a simulated pH neutralization process. The effects of important variables such as range of operating conditions, signal-/noise-ratio, and data length are discussed.