Fuzzy Basis Function Networks of Hierarchical Structure for Modeling Manufacturing Processes

Abstract

Abstract A framework for modeling complex manufacturing processes using fuzzy neural networks is presented with a novel training algorithm. In this study, fuzzy basis function networks (FBFN) which are similar in structure to radial basis function networks are adopted. A hierarchical structure which consists of FBFNs is proposed to construct comprehensive models of the complex processes. The modeling scheme provides valuable advantages over the conventional approaches. A new adaptive least-squares (ALS) algorithm, based on the least-squares method and genetic algorithm (GA), is proposed for autonomous learning and construction of FBFNs. It adds significant fuzzy basis functions (FBF) at each iteration depending on error reduction measures while searching for the best nonlinear parameters of the networks using GA. The ALS algorithm realizes a hybrid structure-parameter learning without any human intervention. Simulation studies with grinding processes are shown to demonstrate that the proposed modeling framework with the training algorithm is a promising approach to modeling complex manufacturing processes.