Grouped neural network model-predictive control

This work provides experimental demonstration for a previously proposed parallel model structure for general nonlinear model-predictive control (NMPC). The model comprises of a group of sub-models, each providing prediction of one process output at one selected future point in time. The sub-models are mutually independent and therefore can run in parallel. This work uses neural networks (NNs) for each sub-model, and terms the prediction model as a grouped neural network (GNN). NMPC based on the GNN model is referred to as GNNMPC. This work demonstrates implementation of GNNMPC on a nonlinear, multivariable, constrained pilot-scale distillation unit. (C) 2002 Elsevier Science Ltd. All rights reserved.