Electricity price forecasting using neural network with parameter selection

Abstract

Price forecasting acts as an essential position in the current energy industry as to assist the independent generators in putting on a remarkable bidding system and scheming contracts, and helps with the selection of supply on the advance generation facility in the long term. These electricity prices are usually hard to predict as it always depends on the uncertainty factors which results in severe volatility or even spikes of price in the energy market. Therefore, determining the accuracy of electricity price forecasting had become an even more important task as there are often remains some crucial prices volatility in the electric power market. This approach focuses on the parameter selection (hidden neuron, learning rate, and momentum rate) and the selection of input data for three types of model. By using the appropriate parameters and inputs, the accuracy of the prediction can be improved. This approach is expected to provide market participants a better bidding strategy and will be used to boost profits in the energy markets using the artificial neural network.