一种改进的雁群扩展粒子群算法研究

针对粒子群算法易陷入局部最优的问题,结合雁群启示粒子群算法和扩展粒子群算法提出了基于雁群启示的扩展粒子群(GeEPSO)算法。该算法在利用雁群飞行方向的多样性同时融合了所有粒子的个体极值信息,提高了种群多样性。为进一步提高改进算法的收敛速度,引入简化粒子群提出了 GeESPSO算法。基准函数的仿真表明:改进算法GeESPSO较好地平衡了收敛速度和局部最优两个矛盾,总体较优。为进一步验证算法在实际应用中的有效性,又分别用两种改进算法优化BP神经网络,并用相关气象数据对PM2.5的值进行预测。

Study on an Improved Extended PSO Algorithm Based on Geese Flight

To solve the local minimum problem of PSO, a new algorithm called GeEPSO is proposed, which hybrids the PSO based on the flight of geese and the extended PSO algorithm. The GeEPSO take the advantage of the various directions of geese and make full use of the position of each particle to assure the diversity of particle swarm. And to further improve the convergence performance, the simplified PSO is introduced to obtain another new algorithm GeESPSO. The simulation results of some benchmarks demonstrate the GeEPSO algorithm can deal well with the conflict between the convergence rate and the local minimum problem, which is better on the whole. And to further investigate the effect of the proposed algorithms in practical application, the new algorithms were used to optimize the BP neural network. Finally the prediction of PM2.5 is made with the improved BP neural network, using the meteorological data.