A Reusable software adaptive fuzzy controller architecture

This paper presents a method for designing a reusable software controller architecture. The proposed controller architecture is designed for a specific domain of controllers (e.g. zero order, first order, second order,.., nth), and it can be reused to develop a specific controller within this specific domain. The basic idea of this reusable controller architecture is to design a general adaptive control system in which knowing the plant parameters is not important, as long as they are within prespecified property constraints. The proposed controller architecture is composed of two submodel parts: (1) Model Identifier which is an on-line fuzzy model identifier represented as an adaptive fuzzy rule based system which takes a sample from the input and output signals, tries to approximate the plant model behavior, and, as long as it takes more samples, gives a more precise model for the plant; (2) Neurofuzzy Controller which is a fuzzy rule based system using a feedforward neural network at the representation level. This feedforward network is trained using the information coming from the fuzzy model identifier, from the desired output, and from the actual output.The 'experimental computing' aspect is based upon the investigation of several kinds of adaptibe algorithms to 'train' the membership functions representation layer, while portraying the user interface specifications layer by use of fuzzy logic. The 'applications' aspect is the important area of software architecture reuse, especially for the development of control systems software.