A Photovoltaic Power Prediction Model for Solar Electric Vehicles Using Convolutional Neural Networks and Masked Multi-Head Attention Mechanism

Abstract

Accurate photovoltaic (PV) power forecasting is crucial for optimizing energy management in solar electric vehicles. To improve prediction accuracy, this study introduces a novel hybrid model that integrates Convolutional Neural Networks (CNN) with Masked Multi-Head Attention (MMHA). The dataset is categorized into four groups based on the correlation between weather conditions and PV power variations. CNNs are employed to extract essential features and patterns from these subsets, while MMHA captures long-term temporal and spatial dependencies in the data. The combined outputs are utilized to predict PV power for solar electric vehicles. Experimental results demonstrate that the proposed CNN-MMHA model delivers superior accuracy and robustness compared to other deep learning models, enabling efficient energy utilization and optimized route planning for solar electric vehicles.