FORECASTING ELECTRIC ENERGY CONSUMPTION IN A HOUSE USING ARTIFICIAL NEURAL NETWORK

Electricity plays a significant role in human life. Human beings rely on electrical appliances in day-to-day activities. Electrical appliances includes electric stove, washing machine, refrigerator, air conditioner, fan, lights, electric bi-cycle, electric car, motors, buses, train, flights, etc. Electricity generation process relies on the amount of energy consumed by the occupants of a house. Electric energy can be generated by various renewable forms such as wind energy, solar energy, etc. In the scenario the prediction process is needed for generating required amount of energy for future. Artificial neural network provides different techniques to predict various time-series data like energy consumption, weather data, etc. The accuracy of different techniques for the prediction process can be varying based on its specifications. Predicting energy consumption can also contain some error rate. It differs from algorithms to algorithms. Accuracy and Performance metrics of prediction algorithms can conclude the best and worst algorithms to predict energy consumption. In this paper, three different algorithms such as Levenberg Marquardt, Bayesian regularization and scaled conjugate gradient are compared to predict energy data. In these algorithms, The Bayesian regularization algorithm provides the best results with minimum error rate which performs metrics such as root mean square error (rmse), normalized root mean square error (nrmse) and mean absolute percentage error (mape).