ANFIS-Hammerstein Model for Nonlinear Systems Identification Using GSA

Abstract

The nature of many real problems in the world is nonlinear type, and identifying their plants and processes symbolizes a challenging task. Nowadays, the block-structure systems, such as the Hammerstein model, are among the most current nonlinear systems. The main characteristic of a Hammerstein model is that its architecture is made up of two blocks; a linear dynamic model preceded by a static nonlinear. The adaptive neuro-fuzzy inference system (ANFIS) is a robust scheme that incorporates a two parts structural design; a nonlinear rule-based and a linear system. In this chapter, it is proposed a new scheme based on the Hammerstein block-structure model for nonlinear system identification. The methodology introduced takes benefit of the correspondence between the ANFIS and Hammerstein structure to couple them and model nonlinear systems. The Gravitational Search Algorithm (GSA) is incorporated to the methodology to identify the model system parameters. The GSA, compared to similar optimization algorithms, achieves more reliable performance in multimodal problems, avoiding being trapped in premature solutions that are not optimal. To test and validate the effectiveness of the methodology, it has been tested over several models and compared with related works in literature showing a higher accuracy in the results.