MATLAB Implementation of an Adaptive Neuro-Fuzzy Modeling Approach Applied on Nonlinear Dynamic Systems - a Case Study

In this paper one of the most accurate adaptive neuro-fuzzy modelling approach is investigated. It is suitable for modelling the nonlinear dynamics of any process or control systems. Basically, this new modelling approach is an improvement of a linear ARX polynomials models based on the least square errors estimation method that is preferred for its simplicity and faster implementation, since it uses typical functions from MAT LAB system identification toolbox. For simulation purpose, to prove its effectiveness in terms of modeling accuracy, an appropriate case study of a centrifugal chiller is considered. The reason for this selection is given by the fact that the centrifugal chiller control system is one of the most seen in a large variety of applications in HV AC control systems. Since its dynamic model is of high complexity in terms of dimension and encountered nonlinearities, a tight control in closed-loop requires a suitable modelling approach.