Short-term Load Forecasting Based on Kprototypes Clustering and Random Forest

Short-term load forecasting is the basic of generation scheduling optimization, power grid dispatching, power market transaction and electricity pricing. The accuracy of prediction will significantly affect the optimal allocation of resources and the improvement of economic and social benefits. On the one hand, there is a high correlation between the short-term load and historical load, and the future load can be predicted through historical data. On the other hand, short-term load forecasting is easily affected by weather conditions and social factors. Therefore, by synthetically considering the historical load data, meteorological data and social factors, this paper proposes a short-term load forecasting method based on kprototypes clustering and random forest. Firstly, the natural and social attributes affecting short-term load forecasting are selected, the historical days are clustered by kprototypes algorithm, and the category of the days to be predicted is determined. Then, the data of the category of the day to be predicted is used as the training set, and the random forest algorithm is used to predict the load. Finally, an example is given to verify the effectiveness of the method.