A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Energy

Thesis (M.Sc.) - Asian Institute of Technology, 2011

This study analyzes and forecasts monthly electricity demand for Provincial Electricity Authority in Chingmai with two approaches. They are Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Network (ANN). The study focuses on monthly historical data of electricity consumption from 2003 to 2010. ARIMA method used in the study will use SPSS software. Artificial Neural Network (ANN) approach applied the multi - layer perceptron and backpropagation algorithm and MATLAB program has been used for the study. This study involves not only past electricity consumption pattern, but also local socio - economic and climatic factors influencing electricity demand in the model such as monthly average temperature, monthly average maximum temperature, monthly average minimum temperature, number of customers and Ft charge. According to the study, various error measures (MPE, MAPE, MAE and RMSE) are used to evaluate the model performance. All of the error measures confirm that forecasted consumption results from ANN are closer to the actual data than forecasted demand from ARIMA. The MPE, MAPE, MAE and RMSE of testing data from ARIMA model are - 2.42%, 4.53%, 9.16 and 12.28, respectively. The MPE, MAPE, MAE and RMSE of testing data from ANN model are - 0.21%, 1.91%, 3.58 and 4.68, respectively. Therefore, ANN is an attractive technique applied for electricity demand forecast in Chiangmai.