Performance Prediction of Low-Concentration Photovoltaic Photothermal Components

In this paper, there are two ways for the performance prediction of CPV/T components, heat transfer analysis and BP neural network, based on the nonlinearity of its performance fluctuation and the addition of heat transfer process, with the low-concentration photovoltaic photothermal complex utilization (LCPV/T) system as the research object. The comparison results show that both prediction methods show good prediction accuracy when the irradiance fluctuation is low, and the relative error is controlled within 10%; when the irradiation value fluctuates, the neural network prediction has higher prediction accuracy in some abnormal output determined by special experimental conditions, while the heat transfer analysis has higher overall prediction accuracy. Combining the advantages and disadvantages of both, the heat transfer analysis is suitable for component performance optimization, and the neural network is suitable for component performance prediction at longer time and space scales.