摘要
粒子群算法用于优化神经网络的权值和阈值,构成粒子群神经网络。将其应用于我国某地区中长期用电量预测建模,并采用滚动时间窗技术来处理用电量预测模型的输入输出数据。通过与实际数据对比,结果表明,采用滚动时间窗技术的粒子群神经网络用于该地区中长期用电量预测建模可行有效,模型预测结果满足要求。
Particle swarm optimization algorithm is used to optimize neural networks＇weights and thresholds in this paper, constructing particle swarm neural networks. Then it is used to model mid and long term electric load. The input/output data of the model are processed using the sliding time window technique. Compared with the real data, particle swarm neural network with sliding time window technique in modeling the mid and long term electric load is effective, and the model can meet the actual demands.
出处
《上海电机学院学报》
2009年第1期20-24,共5页
Journal of Shanghai Dianji University
基金
上海市教委第5期重点学科项目(J51901)
上海市教委项目(06VZ002)
关键词
粒子群算法
神经网络
滚动时间窗
用电量
模型
particle swarm optimization （PSO）
neural network
sliding time window
electricload
model
