Journal of the Korean Solar Energy Society 한국태양에너지학회 논문집 ISSN:1598-6411(Print) 2508-3562(Online)

Adding various colors to photovoltaic (PV) modules is essential for the building-integrated photovoltaics (BIPV) market; however, it can also reduce PV performance and increase manufacturing costs. Therefore, a tool is required to predict PV performance based on color before production of PV modules. In this study, we demonstrate an approach for predicting the performance of colored PV modules before they are produced. First, we analyzed the optical properties of various colors of polyvinyl chloride (PVC) films available in the market and their correlation with power. A total of 15 prediction models were trained. Then, we manufactured colored films for BIPV, predicted their power using the trained models, and compared the predictions with actual power values. Additionally, we validated and optimized the prediction models using mean absolute error (MAE) and root mean squared error (RMSE). We achieved a low MAE of 3.8% and an RMSE of 4.3% using a prediction model formula that consisted of transmittance or a combination of transmittance and absorption. This approach offers several advantages, such as the ability to predict the performance based solely on the optical properties of front-colored films or glass used in colored BIPV without the need for sample production, which can save time and resources.