Analysis and development of iterative fast model control strategies for systems with constraints

Analysis and development of iterative fast model control strategies for systems with constraints

In this research, new fast model control strategies are developed and analysed. To avoid confusion with the competing directions taken by Predictive Control
they have been named ‘Iterative Fast Model Control (IFMC) Strategies’. It has been shown that new IFMC strategies deliver near time optimal performance and have finite settling time. IFMC strategies that include various state constraints are also developed. The system responses and the Lyapunov stability have been analysed.

The possibility of extending IFMC strategies for systems up to an nth order is supported by development of IFMC strategies for 6th to 11th order systems.
Application of IFMC in a real life situation of Aircraft Lateral Control has been studied. Successful implementation of IFMC in a third order ball and beam experiment demonstrates its effectiveness practice. A performance comparison with other contemporary strategies showed that IFMC delivers performance with almost 45% improvement.

The purpose of any real time controller is to determine the drive that should be applied to the plant at each instant. In IFMC, this is performed with the aid of a fast model of the system that can run at a speed that may be a thousand or more times that of the system. Then a decision on plant input is made, based on the fast model behaviour. The input is constrained at both extremes, full positive and full negative.

Fast Model Control strategies were developed from the mid 1950s, initially termed Predictive control although subsequently this term has been taken up by a different
research thread. There have been hardly any publications extending the original theme since 1990. It is possible that the slow and expensive computing resources of that time may have led to the methods being regarded as purely academic. However the more recent proliferation of high speed, yet small affordable microcontrollers may have renewed the relevance of fast model control strategies.