基于门限回归的气温对光伏发电量的影响分析

光伏发电与气象要素密切相关，正确认识气象因子对于光伏发电量及其发电效率的影响是进行发电量预测和维持光伏电站稳定运行的重要基础。采用江苏省淮安市某小型光伏电站2018—2020年发电量数据、淮安市太阳辐射观测站数据及光伏电站附近气象站观测数据，分析气温、太阳总辐射等气象因子对光伏发电量的影响。相关性分析结果表明，光伏发电量与太阳总辐射呈极强的正相关，基本随太阳总辐射的变化而变化；与气温、降水分别呈弱的正相关和弱的负相关；与风速几乎无相关。然而，灰色关联度分析显示，光伏发电量还与气温呈现出较高的灰色关联度。进一步分析表明，光伏发电量随太阳总辐射的变化受到气温的明显调制，不同气温下光伏发电效率不同。门限回归分析显示，当气温达到一定阈值时光伏发电效率会出现下降，具体表现为：当气温达到12.1 ℃，光伏发电效率下降26.7%；当气温超过22.8 ℃，光伏发电效率下降43.7%。使用这两个温度阈值建立的TR（Threshold Regression）模型的准确率要比线性回归模型提高约10%。

Impact analysis of temperature on photovoltaic power generation based on threshold regression

Photovoltaic power generation is closely related to meteorological factors, and understanding the impact of meteorological factors on the photovoltaic power generation and its efficiency is an important foundation for power generation prediction and maintaining the stable operation of photovoltaic power plants. In this study, the influence of meteorological factors such as temperature and solar radiation on the photovoltaic power generation was analyzed using the data of a small-scale photovoltaic power plant from 2018 to 2020 in Huai’an City，Jiangsu Province，as well as the data from the solar radiation observation station of Huai’an and the nearby meteorological station. The results of correlation analysis showed that the photovoltaic power generation was strongly positively correlated with solar radiation and varied with the changes in solar radiation. It presented weak positive correlation with temperature and weak negative correlation with precipitation，while it was almost unrelated to wind speed. However, the grey relational analysis showed a relatively high grey correlation between photovoltaic power generation and temperature. Further analysis indicated that the variation of photovoltaic power generation with solar radiation was significantly modulated by temperature，and the photovoltaic power generation efficiency varied at different temperatures. Threshold regression analysis revealed that the photovoltaic power generation efficiency would decrease when the temperature reached a certain threshold. Specifically，when the temperature reached 12.1 ℃，the photovoltaic power generation efficiency decreased by 26.7%. The photovoltaic power generation efficiency decreased by 43.7% as the temperature exceeded 22.8 ℃. The accuracy of the threshold regressionmodel established by these two temperature thresholds was approximately 10% higher than that of linear regression.