Neural generalized predictive controller stability analysis

This paper emphasizes on the stability analysis of the Neural Generalized Predictive Controller (NGPC) algorithm using Lyapunov methods. NGPC is a hybrid combination between the well known GPC algorithm and a Feed Forward Multi Layer Perceptron (FF MLP) neural network model identifier. This combination leads to a better stability characteristics in the closed loop systems in the presence of high nonlinearities in the process. In this paper, we prove the stability characteristics of NGPC and then present simulation results showing the efficiency of using NGPC over ordinary GPC.