关键词: 平均计算时间复杂度，粒子群优化，动态，变异，多峰值函数优化

Abstract: Particle Swarm Optimization algorithm is widely used in many fields including lots of situations with high real　time requirements such as wide-band digital signal processing. A great number of particles should be updated in iteralion in traditional PSO algorithm. So the average computational time complexity is high. This property of traditional PSO algorithm leads to serious delay which makes it could not be used in system with high real-time requirements. So,the average computational time complexity of traditional PSO algorithm should be reduced with negligible performance penalty. We proposed a Dynamic PSO (DPSO) with adjustable particle quantity algorithm. The core of this algorithm is the criteria of discarding particles. During iterations, some of particles will be discarded dynamically to reduce the average computational time complexity. Furthermore, the mutation was used on local best position in iterations to avoid involving in local optimum solution. Simulation results and theoretical analysis show the average computational time complexity of DPSO algorithm is reduced by about 30 0 o under the condition of similar optimum solution compared with traditional PSO algorithm. In terms of algorithm performance, the performance of DPSO algorithm is same as that of traditional PSO algorithm for the single-modal optimization. On the other side,the performance of DPSO will be better than traditional PSO for multimodal optimization.

Key words: Average computational time complexity, Particle swarm optimization, Dynamic, Mutation, Multimodal optimization