Daily Maximum Load and Its Occurrence Time Forecasting of Distribution Network Based on Hausdorff Distance and ElasticNet

摘要

The forecast of daily maximum load and its occurrence time is of great practical significance but lack attention in the field of short-term load forecasting currently. Based on the recent load, temperature and holiday information considered by traditional short-term load forecasting, this paper adds the holiday load data of the same period last year, for improve the forecast accuracy of the maximum load occurrence time on holiday. In addition, this paper analyzes the law between the daily load shape and the time when the maximum load occurs, and uses the Hausdorff distance algorithm to divide the original data into different categories according to the load shape. The ElasticNet is used to build a forecasting model of the daily maximum load and its occurrence time for each type of load separately. Experimental results show that the proposed model can effectively forecast the daily maximum load value and its occurrence time, and its accuracy is better than the model which only considers traditional factors.