Hybrid Model for Very Short-Term Electricity Price Forecasting

摘要

It is expected that more and more grid connected renewable plants, coupled with energy storage, will be added to the Australian grid to meet the grid integration challenges of renewables. With increased penetration of such systems in the grid, optimal daily scheduling of hybrid renewable-storage generation systems has become a topic of interest. The very short-term (VST) electricity price forecasting is one of the key inputs of such optimal scheduling models. This paper presents a hybrid Support Vector Regression (SVR) and Feedforward Artificial Neural Network (FANN) based approach for VST forecasting of electricity prices. The forecast accuracy of this proposed model is demonstrated with real data from the National Electricity Market (NEM) of Australia.