A Comparative Analysis Of Artificial Neural Network-Based Load Forecasting Model For Chittagong Region

Abstract

This study is to observe a comparative analysis of electrical load consumption and forecasted load for different weather scenario of Chittagong, a region of Bangladesh. In Smart-grid techniques, electricity generation must follow electrical load demand for optimal power operation. In this study, ANN based some load forecasting algorithms are observed for several cases. A real time weather data set for different locations of Chittagong are recorded through an IoT device from April, 2021 to June, 2021 along with online weather data of 60 days to analyse the ANN model. The complete analysis classified into 5 cases in this work to find the better output model. The maximum MAPE value of 8.26% occurs in case 2 for twelve-hour interval data, while the minimum MAPE value of 0.59% occurs in case 5 for weekend days data. Using MATLAB built-in tool Neural Net Fitting, different Back-propagation algorithm such as Levenberg Marquardt, Bayesian Regularization and Scaled Conjugate Gradient algorithm are applied in this ANN training to get better performance.