Mining of electricity prices in energy markets using a computationally efficient neural network

This paper presents a computationally efficient neural network for electricity price forecasting in an Energy market. The proposed neural network is somewhat similar to the conventional functional link neural network (CEFLANN), but differs in the trigonometric expansion block. Unlike the FLANN the input layer comprises the inputs and functions of all the inputs known as the basis functions. The weights in the input layer are obtained using a training algorithm with a sliding mode strategy. The studies on a Ontario energy market and California Energy market exhibit excellent forecasting results over different time horizons for one day ahead of time.