Particle Swarm Algorithm, Its Application In Engineering

The particle swarm optimization (PSO) algorithm is a kind of global optimal technique based on swarm intelligence. Every particle position in the swarm denotes a candidate solution for the optimal problem. The optimal regions in the complicated searching space can be found, trough swarms acting each other. The PSO algorithm possesses the advantages of simplified, rather quick convergence speed, global optimization performance, and less controlling parameters, et al. A review focuses on the basic theory and study progress of the PSO algorithm. Developing trend of the theoretical research and engineering application of the PSO algorithm is presented.Starting off from engineering application, the method for solving nonlinear systems based on PSO is presented, aimed at the sensitive characteristic of initial value of the traditional method for solving nonlinear systems of equations. This method can avoid the weakness which is that convergence is largely dependent on the selecting initial value in Newton iteration method. Numerical results show that the high precision solution of nonlinear systems of equations can be obtained by this method.According to the influence coefficient method, the constrained minimax optimal model for solving optimum balance corrections of flexible rotors with the bound constraints is given out. Using the non-stationary multi-stage assignment penalty function method, the constrained optimal problem is transformed the unconstrained non-smoothing optimal problem. By means of PSO, the optimum balance corrections on balance planes, in which the bound constraints of balance corrections are included, are solvedThe unconstrained optimization model for the forward position analysis of in-parallel manipulators, which is based on the constrained length of the bars, is presented. And then, the PSO algorithm is used for solving the optimization problem. All assemble configuration of the robot are given out by PSO.Using the non-stationary multi-stage assignment penalty function method, the constrained optimization model for solving 0-1 knapsack problem is transformed to the unconstrained optimization model. Furthermore, the PSO algorithm is used for solving the optimization problem. Spreading application of PSO algorithm in discrete optimization problem is presented.Reasons that the basic PSO (BPSO) and simple PSO (SPSO) algorithm have easily trapped into local extremum are analyzed. The effective index by which we can evaluate premature convergence of population is given out. This index expresses the difference of the mean swarm fitness of a better excellent subpopulation and the optimal fitness. Based on the new index, the improved PSO, adaptive mutation PSO (AMPSO), in which the mutation operator is entranced into SPSO is proposed to avoid premature convergence disadvantages of BPSO and SPSO. AMPSO can enhance the performance of algorithm to escape region of local optimization. The experiment results of the classic benchmark functions show that the AMPSO improves extraordinarily the ability of global optimization, and can effectively refrain from the premature convergence.