Kayode O: PhD, 2010 – 2014
The developments of advanced non-linear control strategies have attracted a considerable research interests over the past decades especially in the applications to process plants. Rather than an absolute reliance on mathematical models of process plants which often brings discrepancies especially owing to design errors and equipment degradation, non-linear models are however required because they provide better prediction capabilities but they are very difficult to derive. Further to this problem are the challenges of multi input multi output (MIMO) systems with interactions existing within their process variables, non-convex complex real time optimisation, and inherent strong plant non-linearities. These contribute to make the finding of the global optimal solution difficult tasks during non-linear modelling and control strategy.
Hence, the objectives of this research study is to explore and investigate the soft computing approaches in order to design, simulate, and implement a novel real time constrained MIMO non-linear model predictive controller (NMPC) with the aim to propose a new and enhanced approach. Approximation abilities, high predicting precision, and time-frequency localisation properties of wavelet neural network (WNN) were utilised for the non-linear models design using system identification approach from experimental data to improve upon the conventional artificial neural network (ANN) which is prone to low convergence rate and the difficulties in locating the global minimum point during training process. Salient features of particle swarm optimisation and a genetic algorithm (GA) combination were applied to optimise the network weights. Real time optimisation at every sampling instant is achieved using a GA to deliver results both in simulations and real time implementation on coupled tank systems with further extension to a more complex quadruple tank process in simulations. The results show the superiority of the novel WNN-NMPC approach in terms of the average controller energy and mean squared error over the conventional ANN-NMPC strategies.