Enhancing photovoltaic-efficiency using thermoelectric generators and machine learning optimized installation
摘要
Solar cells generate significant power from sunlight, but they efficiently convert only a portion of the solar spectrum into electricity. The remaining energy is lost as heat, which reduces both performance and lifespan. Integrating a thermoelectric generator (TEG) with a photovoltaic (PV) system can improve efficiency by converting this excess heat into additional electricity. In this study, an 85-W PV panel and eight thermoelectric modules were used to assess the performance of a combined PV-TEG system. Hotspots were identified using a laser thermometer, and thermoelectric modules were installed at these high-temperature locations, resulting in a 4.07% increase in power output. However, heat accumulation negatively affected PV performance due to the relatively low efficiency of the TEGs. To mitigate this issue, finned heat sinks were introduced to enhance the temperature gradient across the TEGs, leading to an overall output improvement of 5.6%. Further enhancement was achieved by employing a multilayer perceptron algorithm to predict optimal installation points for the thermoelectric modules. The heat sinks had a significant impact on increasing the output power of the thermoelectric modules. This approach reduced installation errors to below 3% and improved accuracy, maintaining a temperature margin of error within 2°C. By minimizing human error, this method streamlined the research process and enhanced measurement precision.