An Empirical Approach to Solar Photovoltaic Cell Temperature Prediction

Solar cell temperature is critical in the determination of solar energy generated by a solar photovoltaic power plant. High temperatures are associated with a reduction in the energy generated and hence prediction of photovoltaic cell temperature is essential in temperature mitigation and solar energy forecasting, especially in commercial power plants. The present study focused on the development of a hybrid machine learning based predictive model for solar photovoltaic cell temperature prediction in solar photovoltaic arrays. A physical experimental set up was developed to measure solar cell temperature under different weather and other related parameters. Satellite data were also collated for these parameters and were used to compliment experimental data used in this study. Satellite data used in the study were statistically transformed to mimic experimentally measured data. Feature selection and dimensionality reduction were performed to reduce the input variables and maintain relevant data in the modelling process. A solar cell temperature predictive model based on selected weather parameters was developed using a machine learning approach (Random Forests), and parameters used were selected from the statistical analysis. The prediction accuracy of the developed model was analysed using the coefficient of determination (R²) and the Mean Absolute Percentage Error (MAPE). The results indicated a higher model performance compared to generic models used in cell temperature prediction. The prediction MAPE for the developed model was 0.08 % while an R² value of 0.99 was obtained which was indicative of a good model. The developed model was also comparable to other contemporary models developed to predict solar photovoltaic cell temperature. Simulations were also done to determine the annual energy generated with the incorporation of the solar cell temperature prediction model. The results revealed an average of 25.52 % daily energy difference between a simulation which considered solar cell temperature and that which ignored solar cell temperature.