关键词: 配电网;负荷预测, Hausdorff 距离, ElasticNet 回归

Abstract: The accurate prediction of daily maximum load and its occurrence time are the focus of economic dispatch and safe operation of distribution network. This paper studies the internal law of daily maximum value and its occurrence time of various types of load,and proposes a forecasting method of daily maximum load and its occurrence time combined with Hausdorff load shape classification and holiday correction in the same period of last year. Firstly， the shape characteristics of daily load are analyzed, and the load types are classified by Hausdorff distance algorithm. Then, the corection effect of holidays in the same period of last year in the prediction of daily maximum load and its occurrence time are analyzed, and it is used as the forecast input together with the recent daily load, temperature and other data. Finally, based on ElasticNet regression algorithm, the daily maximum load and its occurrence time prediction model are constructed for each type of load. Taking the load data of a certain station area in Hunan province as an example, the daily maximum load and its occurrence time in the Spring Festival of the station area are predicted and the accuracy and effectiveness of the method are verified in an example.

Key words: distribution network , load forecasting , Hausdorff distance , Elasticnet regression