Solar power, an essential form of renewable energy, is vital for reducing fossil fuel use and curbing environmental pollution. Its economic benefits and efficiency have spurred the global adoption of photovoltaic (PV) systems, particularly in industries like aquaculture. By incorporating solar panels, aquaculture farms can reduce energy costs, increase profitability, and enhance sustainability. This paper focuses on a deep learning (DL) method to identify faults in solar panels, which is vital for asserting the effectiveness of solar energy systems used in aquaculture. We employed VGG16 and VGG19 pre-trained models to analyze solar panel data. The test results show that the VGG19 model outperforms VGG16, achieving higher training accuracy (98% compared to 97%) and lower training loss (0.02 to 0.03). VGG19 consistently demonstrates better accuracy from epoch 7 onwards, although VGG16 is more accurate up to epoch 15. The higher performance of VGG19 can be recognized to its more profound network design, which captures more intricate features, leading to enhanced predictive accuracy and reduced loss. Utilizing advanced deep learning techniques to monitor and maintain solar panels supports the broader goal of sustainable energy use in aquaculture, resulting in reduced costs, improved operational efficiency, and a lower environmental impact.