An Adaptive Linear Element Based Method for Identification of Linear Parameter Varying Models

This paper is concerned with the identification of nonlinear processes which are parameterized as linear parameter varying (LPV) models. The LPV models are represented as blended linear models by using ADAptive LINear Element (ADALINE) structure. Firstly, ADALINE is employed to represent local models describing the system's behaviour at each fixed working point. Then, a novel method is proposed to estimate parameters of weight functions such that the global LPV model can be built by interpolating the local ones. The adaptation algorithm is discussed and extensive simulation studies are conducted to verify the effectiveness of this method.