本文已被：浏览 1273次 下载 740次
投稿时间：2011-07-04
投稿时间：2011-07-04
中文摘要: 分析了BP神经网络的特点,从学习速率的角度讨论了BP算法的改进方法,并用加州负荷数据进行24 h负荷预测及算例分析.仿真结果表明,改进BP神经网络算法预测的平均误差比常规算法降低了0.445%,并且克服了当接近最优解时易产生波动和振荡现象的问题,训练速度也有所提高.
Abstract:The principles of BP network are firstly analyzed，and then several means of improving BP network are discussed with emphasis on its learning rate． Finally，the proposed BP natural network algorithm is tested by load forecasting with historical data of California． Simulation results show that the forecasting precision of the proposed method is enhanced by 0． 445%，while the time is reduced，indicating that the improved algorithm could gain better forecasting effect ．
文章编号：20120421 中图分类号： 文献标志码：
基金项目:
| 作者 | 单位 | |
| 潘雪涛 | 上海电力学院电力与自动化工程学院, 上海 200090 | panxuetao@ shiep．edu．cn． |
| Author Name | Affiliation | |
| PAN Xue-tao | School of Electric Power and Automation Engineering,Shanghai University of Electric Power, Shanghai 200090,China | panxuetao@ shiep．edu．cn． |
引用文本：
潘雪涛.基于改进BP神经网络算法的短期负荷预测[J].上海电力大学学报,2012,28(4):388-391.
PAN Xue-tao.Short-term Load Forecasting Based on Improved BPNeural Network Algorithm[J].Journal of Shanghai University of Electric Power,2012,28(4):388-391.
潘雪涛.基于改进BP神经网络算法的短期负荷预测[J].上海电力大学学报,2012,28(4):388-391.
PAN Xue-tao.Short-term Load Forecasting Based on Improved BPNeural Network Algorithm[J].Journal of Shanghai University of Electric Power,2012,28(4):388-391.