Solar Power Generation Prediction Using Advanced Deep Learning Approach for EV Applications

Accurate solar power prediction plays a pivotal
role in the effective management of Electric Vehicle (EV)
charging systems, especially in smart energy networks. This
paper presents an advanced Deep Learning (DL) based
approach for solar power generation forecasting. Historical and
real-time solar power generation datasets are collected and
undergo comprehensive preprocessing, including data cleaning
and normalization, to ensure quality and consistency. The pre
processed dataset is then split into training and testing sets to
develop a reliable predictive model. A hybrid DL architecture
combining Convolutional Neural Networks (CNN), Gated
Recurrent Units (GRU), and an attention mechanism is utilized
to extract spatial-temporal features and focus on critical input
parameters, thereby improving forecasting accuracy.
Furthermore, model optimization is carried out using the
Chimp Optimization Algorithm (ChOA), which efficiently tunes
models hyperparameter by simulating intelligent search
behavior. The proposed work implemented using Python and
demonstrates strong forecasting capability, achieves MAE of
0.03 and RMSE of 0.09.