亚小时级分解-转移模型对温带气候的误差分析

Feasibility software for photovoltaic (PV) systems leverage decomposition-transposition model pairs to approximate Plane-of-Array (POA) irradiance. This study analyses the accuracy of 15 optical model pairs, using minute input irradiance, to assess POA predictions in a temperate setting by comparing to measured POA for both a tracker and a 55° south-facing tilted system. Using the Mean Absolute error (MAE) of ≤5% as the benchmark, variations were revealed across diverse sky conditions. Model estimates showcased a range of errors, from 2.67% to 51.07%, influenced by condition and system type. For the tracking system, the evaluation showed that in clear conditions, 10 pairs maintained errors within the range. However, this success diminished under intermediate skies, with 5 models remaining within range, and further reduced to 2 models in overcast conditions. The fixed-tilt system demonstrated similar trends but with fewer models meeting the required thresholds; 4 in clear and 2 in intermediate conditions. Only the DISC-SO model pair met the threshold in overcast conditions, exhibiting an MAE of 2.67%. Thus, DISC-SO made it a preferred choice for transposing horizontal irradiance. However, R2 values highlighted challenges due to the high temporal resolution of input data and the hourly data-based SO transposition model. Moreover, the study also examined the impact of decomposition and transposition models on percentage errors. Decomposition changes caused up to 2.43% for tracking systems and 5.34% for fixed-tilt systems. Transposition errors were higher, at 8.53% and 11.51%. Using hourly data reduced errors to 2.35%, 1.44%, and −2.15% in clear, intermediate, and overcast conditions.