SOM-Based Multivariate Nonlinear Vector Time Series Model for Real-Time Electricity Price Forecasting

Electricity price is the precondition and foundation of making decision and plan for the stakeholders in modern electric power market. The deregulation of power markets makes the market environment competitive in recent years. For business opportunities, many aggregators and retailers sprung up, which make electricity price appear fluctuation coupled with changing market structure and some influence factors. It undoubtedly increases the difficulty to analyze electricity price. Therefore, our proposed a novel method called multi-variable nonlinear vector time series (MNVTS) model which considering multidimensional influence factors in order to forecast real-time electricity prices. This forecasting model firstly clusters RTEP fluctuation process into the peak prices, abnormal prices, low prices and stable prices, then calculates and extracts the most influence factors. Our experiments show that our method has good prediction precision.