Neural-network-based predictive control for nonlinear processes

This paper presents a design methodology for generalized predictive control (GPC) using recurrent neural networks (RNNs) for a class of nonlinear processes. The discrete-time nonlinear system model using RNN is constructed with an appropriate learning rate, in order to identify the weights in the recurrent neural network model (RNNM). The proposed neural-network-based predictive controller is derived via a generalized predictive performance criterion and an appropriate learning rate for guaranteeing the convergence of the GPC controller. Two examples, including the control of a nonlinear process and the control of a physical variable-frequency oil-cooling machine, are exemplified to demonstrate the effectiveness of the proposed control approach. Both results from numerical simulations and experiments show that the proposed method is capable of controlling a class of nonlinear processes with satisfactory performance under setpoint and load changes.