Modelling Electricity Consumption Forecasting Using the Markov Process and Hybrid Features Selection

Given the problem of electrical energy storage, it is critical to predict amount load required in order have a reliable and stable power distribution network. Predicting electricity consumption subscribers analyzing their behavior under influence various factors time variables important. large volume subscriber data effective factors, only possible analyze using new information technology tools such as mining. In this paper, feature selection, clustering Markov process techniques are used model subscribers. First, selection subset features based on combined PCA approach Firefly algorithm. Subscribers then clustered selected by K-means. Finally, patterns modeled high-risk clusters. This study simulated Bushehr-Iran Power Distribution Company. The simulation results show superiority proposed over other similar algorithms LASSO-QRNN HyFIS. accuracy prediction method about 1% compared 0.5%