On-line control using the particle swarm optimisation algorithm

In the last two decades, evolutionary based algorithms have proved to be an important tool in solving optimisation problems in many disciplinary areas, namely in control system design. However one of its limitations for some type of applications is the usually high computational load required, which restricts its use for on-line control. This paper proposes the use of a stochastic search algorithm, known as particle swarm, as an optimisation tool for an on-line model predictive control of a custom made laboratory thermodynamic system. Preliminary results are presented