Correction Algorithm Study On Solar Irradiance Measured With Rotating Shadowband Radiometer

Rotating shadow-band radiometer (RSR) is a photovoltaic based radiometer which can be used to measure global horizontal and diffuse horizontal irradiance, and then calculate the beam irradiance. RSR has many features like good persistence, fast response and low cost, however, the photovoltaic sensor in the RSR has a limitation of uncertainty because of its sensitivity to the temperature, relative humidity and solar spectrum, which can lead to measurement errors, so it is very necessary to develop an algorithm model used for correcting the irradiance measured with RSR.In order to correct the measurement errors of RSR, in this paper the data measured in the national solar radiation research laboratory are selected as the research object, the least square polynomial regression, support vector machine theory and statistical evaluation indicators are used to research for global irradiance correction algorithm and diffuse irradiance correction algorithm model respectively, and then the corrected beam is calculated using the corrected global and the corrected diffuse, thus, the correction algorithm model for all irradiance data measured with RSR is obtained. The main research contents include:Irradiance measurements and quality control method. The structure, performance, accuracy and price of the irradiance observation instruments are researched first, then the geography and climate condition of the irradiance observation station are analyzed, and the irradiance quality control method is studied, a systematic quality control procedure is proposed and the data used for research are obtained.Study on global irradiance correction algorithm model. Based on the physical source of the measurement errors of photovoltaic sensor, using the least square polynomial regression, step fitting relating the correction factor residual of global irradiance to temperature, solar zenith angle and air mass is conducted creatively, then the global irradiance correction algorithm model is developed. Correction results show that, both in short and long term, the MBE of the corrected global is lower while the RMSE is a little higher than the one of the uncorrected, but the comprehensive t-statistic is over50%and23.9%lower than the uncorrected, so the correction algorithm is effective.Study on diffuse irradiance correction algorithm model. The influences of the environment factors on RSR measurement errors are analyzed, based on support vector regression, a method of compensating the diffuse measurement errors by the optimum prediction of the diffuse measurement error corrections is creatively developed, then the algorithm model of correcting diffuse irradiance is proposed. Correction results show that the MBE and RMSE of the corrected diffuse are much lower than the one of the uncorrected. Meanwhile, compared with Vignola correction algorithm and VA correction algorithm, the algorithm presented in the paper can avoid respective underestimation and overestimation of the similar correction algorithm effectively.Comprehensive performance evaluation of the whole correction algorithm model. Using the randomly selected irradiance data, according to global correction algorithm and diffuse correction algorithm, obtain the corrected global and diffuse, then calculate the corrected beam. Comparison results show that the MBE and RMSE of the corrected beam are84.4%and29.1%lower than the one of the uncorrected, and the t-statistic is86.6%lower. Obviously, the accuracy of the corrected beam derived from the corrected global and the corrected diffuse is apparently improved, and the comprehensive performance of the correction algorithm is effective and applicative.