Granular-Based Linguistic Models for Identification of Process System

This paper is concerned with a method for designing a Granular-based Linguistic Model (GLM) for identification of process system. For this purpose, a GLM is constructed by the use of fuzzy granulation realized via linguistic context-based fuzzy clustering. The proposed approach is comprised of three steps. In the first step, we use Linear Regression (LR) for global approximation. As a result, we obtain the approximation error. On the basis of the error, we construct the GLM based on the linguistic contexts produced by the error distribution for local approximation in the second step. Finally we use the Constrained Least Square Estimate (CLSE) method to generate the flexible linguistic context in the consequent part to improve the approximation performance. Finally, we use the proposed methods to identify a pH neutralization process in a continuous stirred-tank reactor (CSTR). The experimental results reveal that the proposed approach yields a better performance in comparison with linear model, Linguistic Models (LM), and GLM itself.