Neural fuzzy controller in a temperature control system

The paper describes a control algorithm which linking advantages of the fuzzy logic (FL) method and the artificial neural network (ANN) method. The proposed structure can be treated as a neural fuzzy controller, whose inferential unit, in contrast to classical fuzzy controller, was substituted with an artificial neural network structure. In this work an attempt has been made to work out an algorithm which would combine the advantages of FL and ANN methods. The design of an algorithm for neural fuzzy controller (NFC) is illustrated with an example of thermal object control. The example is given of NFC training set acquired through selection of such configuration for the rules and experimental data that the both forms of information, i.e. qualitative and quantitative, would complement each other. It was tested whether would be possible the practical implementation of the computing power available in the digital control systems utilised in practice.