APPLICATION OF THE RECURRENT MULTILAYER PERCEPTRON IN MODELING COMPLEX PROCESS DYNAMICS

A nonlinear dynamic model is developed for a process system, namely a heat exchanger, using the recurrent multilayer perceptron network as t he underlying model structure. The recurrent multilayer perceptron is a dynamic neural network, which appears effective in the input-output modeling of complex process systems. A dynamic gradient descent learni ng algorithm is used to train the recurrent multilayer perceptron, res ulting in an order of magnitude improvement in convergence speed over a static learning algorithm used to train the same network. In develop ing the empirical process model the effects of actuator, process, and sensor noise on the training and testing sets are investigated. Learni ng and prediction both appear very effective, despite the presence of training and testing set noise, respectively. The recurrent multilayer perceptron appears to learn the deterministic part of a stochastic tr aining set, and it predicts approximately a moving average response of various testing sets. Extensive model validation studies with signals that are encountered in the operation of the process system modeled, that is steps and ramps, indicate that the empirical model can substan tially generalize operational transients, including accurate predictio n of process system instabilities not included in the training set. Ho wever, the accuracy of the model beyond these operational transients h as not been investigated. Furthermore, on-line learning becomes necess ary during some transients and for tracking slowly varying process dyn amics. In view of the satisfactory modeling accuracy and the associate d short development time, neural networks based empirical models in so me cases appear to provide a serious alternative to first principles m odels.