An Adaptive Resource Allocating Neural Fuzzy Inference System

Adaptivity to non-stationary contexts is a very important property for intelligent systems in general, as well as to a variety of applications of knowledge based systems in the area of Electric Power Systems. In this paper we present an innovative Neural-Fuzzy architecture that exhibits three important properties: online adaptation, knowledge (rule) modeling, and knowledge extraction from numerical data. The ARANFIS (Adaptive Resource Allocating Neural Fuzzy Inference System) has an adaptive structure, which is formed during the training process. We show that the resource allocating methodology enables both online adaptation and rule extraction; the latter differentiates it from the majority of Neurofuzzy systems with fixed structure which perform mainly rule modification/ adaptation rather that rule extraction. The efficiency of the system has been tested on both publicly available data, as well as on a real generated dataset of a 120 MW power plant.