An assessment of energy demand using short-term load forecasting

Renewable energy reduces the problem of greenhouse effect which is a great problem to the planet. Thus, utilization of the solar/battery storage hybrid system is necessary, however there is need to eradicate problems such as energy insufficiency with demand, imprecise solar panels and battery storage sizing, and downtime which may cause energy and economical loses. To get around these limitations and improve load demand forecasting accuracy, a variety of methods have been developed, including statistical models like ARIMA and artificial intelligence approaches like SVR and artificial neural networks (ANN). All these three approaches are trained using the data in the training set. The trained intelligent techniques are then tested using the available data in the test data set as input. Furthermore, the three approaches are used to forecast the load demand for 24 hours using 646 observations/ hourly data set. Ultimately, the best method with the least MAPE, MAE, RSME is then selected as the best forecasting technique, which in this case is the ANN model. Thus, through load assessment and load demand prediction, accurate data is obtained, and it is possible to prevent installation of incorrect sized solar panel, and battery system. By so doing, greenhouse effect, energy and economical losses problems are eliminated.