An effective hybrid PSOSA strategy for optimization and its application to parameter estimation

摘要

As a novel evolutionary technique, particle swarm optimization (PSO) has received increasing attention and wide applications in a variety of fields. In this paper, an effective hybrid optimization strategy by incorporating the jumping property of simulated annealing (SA) into PSO, namely PSOSA, is proposed for estimating parameters of non-linear systems, which is an important issue in control fields and essentially is a hard multi-dimensional numerical optimization problem. By employing the SA-based selection for the best position when updating the velocity in PSO, the hybrid strategy is of more effective global exploration ability over pure PSO at the beginning searching stage (when temperature is high) so as to avoid premature convergence. As the temperature decreases, the hybrid strategy transforms to PSO smoothly to stress the exploitation. Simulation results based on three different kinds of models as well as a DTS200 three-tank system demonstrate the effectiveness and efficiency of the proposed PSOSA hybrid strategy, whose estimating quality is much better than that resulted by GA and is competitive to that resulted by GASA and SMSA hybrid strategies. Moreover, it is also demonstrated by comparative simulation results that, the ability to avoid being trapped in local optimum of PSOSA is much superior to pure PSO.