Artificial Intelligence Models for Short Term PV Forecasting for Direct PV to EV System

The fast expansion of solar photovoltaic (PV) systems and the widespread adoption of electric vehicles (EV) in the future may lead to a mismatch between supply and demand, posing challenges for distribution operators. One potential solution to this issue is the direct integration of PV to EV, reducing the reliance on Battery Energy Storage Systems (BESS). However, a key challenge with solar PV technology is its dependence on intermittent factors such as cloud cover and weather conditions. To tackle this, Artificial Intelligence (AI) models and datasets can be used for short-term PV energy output forecasting. This paper presents an extensive review of recent research (2014–2024) in short-term solar energy forecasting and the case of direct PV to EV charging. The review analyzes content from reputable academic sources, assessing the performance of deep learning models, the availability of open-source datasets, and their implications for direct EV charging. Notably, M-CNN-L and GRU algorithms demonstrate superior accuracy in short-term solar energy forecasting. The integration of ground-based cloud images with time-series weather data greatly improves predictive accuracy. Direct PV to EV system is not yet widely implemented, due to challenge in DC protection system and complex infrastructure planning. Moreover, valuable online datasets, including meteorological data, historical solar irradiance, and cloud images, are available for research and practical use. Nevertheless, gaps exist in model performance and evaluation metrics. This paper provides recommendations for future research, such as refining methodologies, improving datasets, and developing customized evaluation metrics tailored to network management and future electricity with flexible market pricing. These suggestions aim to advance the field of short-term PV forecasting for direct PV to EV system development.