Real-time electricity pricing trend forecasting based on multi-density clustering and sequence pattern mining

Abstract

The implementation of real-time electricity price has become an essential point in the electricity market reform. It reflects the balance between the real-time market price and the electricity price. However, due to the non-linear, non-stationary, time variant and other uncertainties factors in power market, prediction accuracy is difficult to guarantee. Therefore, we proposed a Multi-density Clustering (MD Clustering) algorithm use different radius to classify the electricity price data, and automatically generated multi-levels clusters by different price ranges. Then, we forecast the trend of electricity price based on the association analysis and pattern recognition of different level catagories. The experimental results show that our MD clustering algorithm has fast performance and high accuracy in dealing with the data of density attributes nonuniformity condition, and ensure the accuracy of real-time electricity price forecasting.