一种基于支持向量机回归的旋转遮光带日射表散射辐照度修正新算法

A New Correction Algorithm for Diffuse Irradiance Measured with Rotating Shadow-Band Pyranometer Based on Support Vector Regression

通过比较旋转遮光带日射表(RSP)和参考标准表所测散射辐照度之间的数值差异,分析了太阳总辐照度、环境温度、相对湿度和太阳光谱等气象要素对RSP散射辐照度测量误差的影响关系,提出了一种修正RSP散射辐照度测量值的新算法。该算法从支持向量机回归预测角度,建立了对RSP散射误差修正值的一次预测模型,然后根据误差修正值最优预测模型推导出RSP散射辐照度修正算法模型。利用该算法对美国国家太阳辐射研究实验室和劳里观测站采集的RSP散射辐照度数据进行修正,修正后两观测站数据的平均偏差和均方根误差分别降低到-0.2W/m2,3.3W/m2和1.9W/m2,8.5W/m2,显示算法具有良好的修正性能和适用性。该算法能够有效避免Vignola算法中存在的欠修正和Vignola and Augustyn(VA)算法中存在的过修正现象。

Through comparson of the numeric difference between diffuse irradiance measured with rotating shadow- band pyranometer （RSP） and standard reference pyranometer, the influence of meteorological elements like global irradiance, ambient temperature, relative humidity and solar spectrum, etc. on measurement errors of diffuse irradiance measured with RSP is analyzed, based on the analysis, a new algorithm of correcting diffuse irradiance measured with RSP is proposed. The algorithm constructs an error modifier prediction model based on support vector regression （SVR） theory first, and then diffuse correction algorithm of RSP can be derived from mentioned prediction model. Applying the correction algorithm to the diffuse irradiance data measured by Solar Radiation Research Laboratory and Lowry Range Solar Station, mean deviaiton and root mean square error of the two stations＇ data after correction decrease to --0.2 W/m^2 , 3.3 W/m^2 and 1.9 W/m^2 , 8.5 W/m^2 , which indicates that the algorithm is effective and applicative. Meanwhile, compared with Vignola correction algorithm and Vignola and Augustyn （VA） correction algorithm, the algorithm presented can avoid respective undercorrection and overcorrection of similar correction algorithm effectively.