基于改进粒子群优化算法的BP预测模型  

BP Forecast Model Based on Improved PSO Algorithm

作　　者：王行甫[1] 陈宏亮[1] 

机构地区：[1]中国科技大学计算机科学与技术学院,合肥230027

出　　处：《计算机系统应用》2014年第4期135-137,143,共4页Computer Systems & Applications

基　　金：国家科技重大专项(2012ZX10004301-609)

摘　　要：该论文提出了基于改进粒子群优化的BP算法.在该算法中,通过对粒子群优化算法中的惯性权重的计算方法的改进,同时利用改进的PSO算法替代了BP算法中的梯度下降算法,使得改进后的算法具有不易与陷入局部极小等优点.并将该算法利用在预测气温上,实验证明:改进后的算法在预测模型上能够取得较好的预测效果,提高预测精度.A BP prediction model which based on improved PSO algorithm is proposed in this essay, in this algorithm, I modify the way to calculate the inertia weight, and use the improved PSO algorithm instead of the gradient descent algorithm, Make this changes it will not fall into local minimum. And use this way to forecast the weather, through the experiment result I find that the improved algorithm has high efficiency, and it also improve the prediction accuracy.

关 键 词：粒子群算法 BP算法 气温预测 惯性权值 梯度下降算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]