摘要
为了进行有效而高质量的售电量预测,根据实际电力系统电价和影响因素的不同,将售电量分为五类,运用人工神经网络中的误差反向传播算法,利用重庆某地区历年来的月温度和月售电量等数据建立起最佳预测模型进行学习训练,并用重庆某供电局2006年的售电量实际值和预测值进行校验。最后利用MATLAB平台开发出了实用化的售电量分类预测软件。实际算例的分析表明,售电量分类预测比售电量总体预测具有更好的预测精度和实用价值。
Load forecasting is vitally important for power system administrative departments. In terms of the sale prices and influencing factors, power consumption is classified into five sections to implement load forecasting with high quality and efficiency. Based on historical power comsumption data of Chongqing and the theory of back propagation artificicl neural network, the optimal input model is developed. Comparison between the actual sales and forecasing value of one power supply bureau of Chongqing in 2006 is analyzed. A practical software package has also been developed. Simulation results show that the classified power consumption forecasting model has higher forecasting precision and practical value than that of total power consumption forecarting model.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2008年第6期51-55,共5页
Proceedings of the CSU-EPSA
基金
2005重庆市城区供电局自控科技项目3#
关键词
售电量
BP神经网络
预测模型
power consumption
BP neural network
forecasting model
