摘要
电力调度和市场营销部门对电力负荷数据的走势形态和预测方法十分重视。在实际应用中,电力市场对提前预测未来连续多天的日最大负荷提出了新的要求。本文根据电力系统中日最大负荷的历史数据,分法定假日与非假日两部分单独研究其特性。对于假日最大负荷的预测,设定假日因子;对于非假日,通过小波分解提取日最大负荷变化的周期特征,再分别建立相应的BP神经网络模型进行预测。通过对某市电力负荷数据的预测及结果表明:采用这种组合方法可行有效、预测精度满足行业要求。有较强的理论意义和广泛地应用前景。
Dispatch and marketing departments of electric power put great emphasis on trend and forecasting method of power load data.In practical applications,the electricity market raises new requirement for forecasting the daily peak load in continuous multi-days ahead of time.In this paper,according to the history data of daily peak load,the characteristics of holiday and non-holiday daily peak load are investigated respectively.For the forecasting of holiday daily peak load,a holiday factor is added; for the non-holidays,the period characteristic of daily peak load is obtained through the wavelet decomposition.And then,the forecasting is respectively carried out by establishing BP neural network model.Simulation results show that this method is feasible and effective,which can meet the industry requirement for prediction accuracy and has a strong theoretical significance and wide application prospects.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2014年第10期31-34,共4页
Proceedings of the CSU-EPSA
基金
国家科技部政府间科技合作项目(2009014)
上海市创新基金项目(jwcxsl1302)
关键词
特征提取
连续多天负荷预测
日最大负荷
假日负荷预测
神经网络
characteristics distilling
continuous multi-days load forecast
daily peak load
holiday load forecast
neural network
作者简介
马立新（1960-）,男,博士,教授,研究方向为电力系统分析与优化运行、智能电网与智能科学、电气设备状态监测与诊断方法、电能质量监控与能效测评技术.Email：malx_ai-i@sina.com
李渊（1989-）,女,硕士研究生,研究方向为电力系统负荷预测.Email：sunflower.123@163.com
