Fleet Load Charge Forecasting in Electric Vehicle Using Hybrid Deep Learning Model: LSTM-AQOA

When a high number of electric vehicles in a concentrated area are all charging at once from the same local transformer, the grid's performance, efficiency, and necessary capacity are all impacted. Thus, for a power system to function dependably and efficiently, precise load forecasting is essential. Charging loads from EV fleets are predicted to rise in importance to power grids. Therefore, anticipating the load from EV fleets is critical to ensuring the continued security and dependability of the electrical grid. In this study, we provide a novel hybrid EV fleet charging forecasting model based on multiple decomposition. Long short-term memory (LSTM) is one component of the artificial intelligence (AI) based forecasting model into which each deconstructed signal is input. Aquila Optimisation Algorithm (AQOA) is used to pick the best possible LSTM error threshold value in this case. The accuracy of the models is checked against actual charging data for EV fleets in the wild. Through a comparison of the deployed models, the prediction presentation of the projected model is measured using a variety of statistical measures. It is shown that, compared to models with only one decomposition, models with multiple decompositions perform better across a wide range of measures.