Computationally Efficient Nonlinear Predictive Control Based on RBF Neural Multi-models

Abstract

Listen

Translate article

This paper is concerned with RBF neural multi-models and a computationally efficient nonlinear Model Predictive Control (MPC) algorithm based on such models. The multi-model has an ability to calculate predictions over the whole prediction horizon without using previous predictions. Unlike the classical Nonlinear Auto Regressive with eXternal input (NARX) model, the multi-model is not used recursively in MPC, the prediction error is not propagated. The presented MPC algorithm needs solving on-line only a quadratic programming problem but in practice it gives closed-loop control performance similar to that obtained in nonlinear MPC, which hinges on on-line non-convex optimisation.