摘要
电力系统超短期负荷预测的准确性直接影响到电力系统发电与用电量平衡的问题。目前大多数的电力系统超短期负荷预测都只利用了电荷负载变化本身的时间序列特效。而事实上,除电荷本身的时序特征,温度、湿度、降雨量和人口等因素也对电荷变化产生了明显的影响。利用某地2009-01-01至2015-01-09的数据建立了结合温度、湿度、降雨量等因素的多维度XGBoost模型,和只考虑时序特征的XGBoost模型进行了多角度比较。从平均绝对误差(MAE)、均方误差(MSE)和均方根误差(RMSE)等多角度预测数据分析可以发现,只考虑时序特征的XGBoost模型可以很好地预测超短期负荷的整体变化趋势,而结合温度、湿度、降雨量等因素的多维度模型在整体趋势的基础上更好地预测了电网负荷变化的细节。同时也证实了机器学习中特征因素并不是越多越好,当特征项存在相关性过高的冗余因素时模型预测精度会降低。
The accuracy of ultra-short term load forecasting of power system directly affects the balancebetween the power generation and the power consumption. Currently,most ultra-short term load forecasting of power system only uses the time series effects of the charge load change itself. In fact,in addition to the time series characteristics of the charge itself,the temperature,humidity,rainfall,and population and other factors also have a significant influence on the change of charge. Therefore,in this article,a multi-dimensional XGBoost model combined with factors such as temperature,humidity and rainfall is established based on data from 2009-01-01 to2015-01-09 in a certain place,and from many angles compared with XGBoost model considering the time series characteristics only.From many angles ofthe average absolute error( MAE),the mean square error( MSE) and the root mean square error( RMSE) etc.In the prediction data analysis,we had found that the XGBoost model could well predict theoverall trend of the ultra-short term load,and themulti-dimensional modelcombined withtemperature,humidity,rainfall and other factorscould better predict thepower load change details based on the overall trend. Furthermore,this paper confirmed that the prediction accuracy of machine learning was not lwayspositively related with the amounts of machine learning characteristics,and it would be reduced when the feature item had the redundant factor of high correlation.
作者
杨修德
王金梅
张丽娜
杨国华
李冰轩
Yang Xiude;Wang Jinmei;Zhang Lina;Yang Guohua;Li Bingxuan(Electronic and Communication Engineering School of Physics and Electronic-Electrical Engineering, Ningxia University, Yinchuan Ningxia 750000, China;Ningxia Electric Power Energy Security Autonomous Region Key Laboratory, Yinchuan Ningxia 750000, China)
出处
《电气自动化》
2019年第1期32-34,共3页
Electrical Automation
基金
国家自然科学基金项目(NO.51167015)
宁夏自然科学基金项目(NE.17022)