Data-Driven Forecasting Method of Multiple Day-Type Load Curves for Active Distribution Networks

The traditional distribution network load forecasting usually can only get the total load result in a certain period, which is difficult to reflect the multiple influence of social and meteorological factors on the load. In recent years, with the gradual integrations of distributed generations and flexible loads, the randomness and uncertainty of distribution network load have been increasing significantly, and the traditional forecasting methods can no longer meet the requirements of planning and operation. Therefore, a data-driven load curve forecasting method for distribution network is put forward in this paper. Firstly, a data-driven model is established for different day types such as holidays, working days and rest days. By combing and applying data-driven techniques, the extraction of potential data law is realized. Then, the valuable information extracted by the data-driven techniques is combined with load forecasting models such as least squares support vector machine and discrete grey model to obtain the load curve of the day to be predicted. Finally, the standard data set provided by the 9th National Student Electrician Mathematical Modeling Competition is used as an example to compare and analyze the forecasting effect. The results show that the proposed method can provide valuable support for distribution network planning and operation.