Analysis of variance using fuzzy logic models

A new algorithm to conduct the analysis of variance using fuzzy logic models (FLANOVA) has been proposed. The algorithm uses fuzzy logic models (FLMs) to rank the input variables of a process. The FLMs used are specially designed to be piecewise polynomials. Based on the data from the process, an FLM is constructed. After the adequacy of the model representing the process is checked, the algorithm calculates the F-number based on the established FLM. Because of the excellent correlation capabilities of the FLMs, the proposed algorithm can represent the process well and evaluate accurately the effects of the input variables. Two case studies with two test functions were conducted to compare the proposed algorithm with the existing methods for the analysis of variance using polynomial models. Results demonstrated the effectiveness of the proposed approach. Through the use of membership functions of the FLMs, the proposed approach can be implemented more easily than approaches using typical piecewise models for multidimensional applications