A Power Load Association Rules Mining Method Based on Improved FP-Growth Algorithm

Abstract

Some low-frequency information in the power load association analysis cannot be mined effectively by traditional mining algorithms. In this case, a comprehensive mining method based on FP-growth algorithm is proposed in this paper. Firstly, the data on meteorology, solar terms, holiday and total load in Pudong District of Shanghai for 546 days was clustered and generalized by K-means method. Then the original transaction sets was classified according to the counts of the clustering result. And they were mined by different methods comprehensively. Compared with association rules obtained by traditional algorithms, more association rules can be mined by the comprehensive mining method, with accuracy and robustness. The comprehensive mining method provides basis for load forecasting as well as distribution network load warning, which is crucial to the operation and management of smart grid.