FUZZY LOGIC AND GENETIC ALGORITHMS SUPERVISORS FOR INTERNAL MODEL CONTROL STRATEGY

This paper presents two methods allowing the online adjustment of the filter gain in the internal model control (IMC) strategy. These methods are based on fuzzy logic and genetic algorithms. The IMC strategy needs the direct model and the inverse model of the process. These models can be estimated offline from input- output data. In this work, we have used feed forward Artificial Neural Networks to determine these models. The back propagation algorithm is used to train the neural networks. The neural network internal model control with the proposed supervisors is applied to numerical examples. The performances of the proposed controller are compared to a standard PI controller and to a PI controller with an anticipation action given by the inverse model of the process.