Application of interval type-2 subsethood neural fuzzy inference system in control and function approximation

Abstract

This paper presents the application of Interval Type-2 Subsethood Neural Fuzzy Inference System (IT2SuNFIS) [1] in the area of control of a chemical plant and function approximation. In this model, a subsethood method between the inputs and hidden rule layer neurons determines the similarity between interval type-2 fuzzy set (IT2 FS) inputs and IT2 FS antecedents. The inputs to the system are fuzzified using IT2 FS with Gaussian primary membership function (GPMF) having identical mean but different variance. The signal aggregation of type-2 based activation is performed using product operator. This neuro-fuzzy system trains in differential evolution (DE) framework. Different DE learning strategies have been used for this purpose. During the training, different networks are generated and trained using DE methodology. The system is tested on the control of a chemical plant. Comparisons with other type-1 and type-2 neuro-fuzzy models verify the excellent control of the proposed methodology for the control of the chemical plant. The system is also tested on Hang function approximation problem. It is observed that the system performs better than other models reported in literature in terms of lesser number of free parameters; the result accuracy is similar to other models.