Hourly load and price forecasting using ANN and fourier analysis

Forecasting is an important tool that enables the decision makers to understand the current market trends and plan the system accordingly. Load and Price Forecasting are two such important tools in the electricity market that help in performing essential tasks such as unit commitment of generating capacity, resource planning, cash flow analysis for stakeholders etc. This paper presents a new ANN model for hourly load and price forecasting. Since there are variations in demand throughout the day, a model needs to be devised to predict these variations. In this paper, Fourier analysis of the load and price signal has been done to find out the dominant features of these signals. These features are then fed to a Multi-layer Feed Forward Neural Network. The ANN model is trained on hourly data from ISO New England electricity market from the year 2007 to 2011 and tested on out-of-sample data from the year 2012. Results obtained from simulation show that the proposed model works well in both Load and Price Forecasting and the value of Mean Absolute Percentage Error (MAPE) is obtained is minimum compared to other methods.