Real-Time Electricity Price Forecasting for Energy Management in Grid-Tied MTDC Microgrids

In power market, electricity price forecasting is one of the challenging issues and it plays a significant role in grid-tied microgrid's (MG's) energy management systems (EMSs). Accurate forecasting can make precise economically beneficial decision not only in efficient resource scheduling, but also in real-time energy arbitrage. To maximize the profit, the idea is to buy energy at low price and sell at high. However, it is very complex and volatile task since the electricity prices are highly fluctuating in nature as a result of bidding competition among the suppliers and consumers in the electricity market. In this context, an artificial neural network based approach with an adaptive learning algorithm called resilient backpropagation was used to forecast the real-time electricity price throughout the year 2016 and forecasting error was analyzed. Result showed that the yearly mean error was $3.705IMWh and the mean absolute percentage error was in acceptable range of 5%-10% most of the cases. Also, the economical impact of the real-time electricity price forecasting to the MG's EMSs was analyzed, and a model predictive control based EMS for grid-tied multi-terminal DC (MTDC) microgrids was proposed in this paper.