Database and deep learning models for the prediction of total demand and maximum generation of electricity in Bangladesh

This work is based on the modification of the "BanE-16" dataset Salehin et al. (2023), obtained from a repository which integrates the dynamics of the Bangladesh power grid with meteorological variables. The dataset comprises peak power demand, environmental factors such as temperature, wind speed and atmospheric pressure, as well as electricity generation statistics. It is crucial to highlight that in order to prevent the introduction of bias into the analysis due to temporal patterns, we have chosen to exclude data capture dates from the model. This is because the focus of this work is on the underlying weather and energy relationships, regardless of time. Furthermore, a total of 52 rows in the original set were excluded due to the absence of data on total demand, maximum generation, or both variables. This was done in order to ensure a correct fit by the trained model. The repository consists of three folders, each corresponding to one of the models used: the Simple Artificial Neural Network (ANN), the Multilayer Artificial Neural Network (ANN), and the Long Short-Term Memory (LSTM) network. Each folder contains both the simulators and the database and best fits obtained for the respective models. Reference: Salehin, Imrus; Noman, S. M. (2023), “Peak Energy Demand in the Electricity Energy Dataset BanE-16”, Mendeley Data, V2, doi: 10.17632/3brbjpt39s.2