SHORT TERM ELECTRICITY PRICE FORECASTING USING NEURAL NETWORK

This paper presents neural networks applied for short term electricity price forecasting in Ontario energy market. The accuracy in electricity price forecasting is very crucial for the power producer and consumer. With the accurate price forecasting, power producer can maximize their profit and manage short term operation and long term planning. Meanwhile, consumer can maximize their utilities efficiently. The objective of this research is to develop models for day ahead price forecasting using back-propagation neural network during summer. Six models were developed representing six types of inputs. The result shows that 24 models representing 24 hours ahead price forecasting with price and demand inputs gives better result compared to other five models due to unique model developed for each hour rather than a model for a day with mean absolute percentage error (MAPE) of 18.74%.