Optimizing the Performance of Predictive Control Using Genetic Algorithm

Y.Y. Nazaruddin and F. Maulana (Indonesia)


predictive control, neuro-fuzzy modelling, nonlinear process, genetic algorithm


An alternative control strategy for nonlinear process which is an integration between Predictive Control technique and Genetic Algorithm (GA) is proposed in this paper. Generalized Predictive Control (GPC), which is considered as universal method for model based predictive control, is used based on neuro-fuzzy modelling. In this investigation, GA is implemented as an optimization method for parameter learning in neuro-fuzzy model instead of the common Backpropagation-error gradient-based algorithm. Further, the GA is also performed in determining the optimal control signal value to minimize the cost function of GPC. The overall performance of the proposed control strategy is evaluated either for off-line or on-line learning mode in order to investigate the capability of GA in modelling and the performance of GPC in the adaptation to the change in process parameter.

Important Links:

Go Back