| Spectroradiometer is a precise electro-optic instrument system with high data dimension and multi-parameter dimension.The evaluation and optimization of instrument accuracy are based on high-throughput data statistics and matrix data analysis.The repeated iterations of various analytical testing process in the instrument system,which include of differential models reducing dimension and integral models adding dimension,make the characterization model complex and data computation heavy.The development of spectroradiometer system characterization model based on multi-dimensional parameters and high-dimensional data lags behind the development of core devices and systems,which is the fundamental reason that restricts the improvement of spectroradiometer system accuracy.With the development of computer application in measurement,control and data analysis,instrument modeling and statistical metrological analysis related to spectroradiometer are increasing.However,there are still some key problems to be studied urgently in order to accurately evaluate and optimize the calibration accuracy of spectroradiometer.In order to solve the problems of multi-dimensional parameter error characteristics,such as stray light quantization characterization modeling,two-parameter correlation influence characteristic modeling of CCD non-linear model,and error characteristics of key parameters in uncertainty evaluation of metrological calibration,the following research has been carried out:(1)In order to obtain complete system parameter information for parameter control,a precision spectroradiometer is designed,which provides an instrument foundation for error characterization and calibration.The key influence links,such as receiver function,stray light and slit coupling,are analyzed and optimized.In order to eliminate the effect of stray light enhanced by volume reduction,linear variable filter and stray light trap are separately used to suppress in-band high-order stray light and out-of-band diffuse stray light.(2)On the basis of acquiring the system parameters of spectroradiometer,the stray light characteristics in the error characteristics of spectroradiometer are quantitatively characterized and modeled,and the uncertainty is evaluated.Based on the measured data of stray radiation matrix and the key geometric feature points,the out-of-band suppression and background stray radiation characterization model are constructed.The evaluation results of the stray light characteristics of the designed spectroradiometer are given quantitatively.The out-of-band suppression is less than 0.1%,the upper limit of background stray radiation is less than 0.01%,and the statistical mean is about 0.001%.The results show that the design of stray light suppression in the spectroradiometer is effective and achieves the stray light level of the monochromator system.Furthermore,combining the spectral range and spectral resolution bandwidth,the evaluation value of stray light for measurement uncertainty is calculated to be 2%,which provides a basis for quantitative calculation of composite uncertainty.(3)In order to accurately evaluate and suppress the influence of non-linear errors of spectroradiometer,quantitative characterization of non-linear characteristics of CCD,modeling correction and evaluation value of uncertainty are carried out.The nonlinear parameter dependence characteristics of integration time and intensity gray scale of CCD detector are experimental analyzed and modeled.The non-linear quantitative characterization function of CCD is given.On this basis,the multi-scale non-linear correction of CCD in the measurement process is realized.The model is compared with the monitoring power meter data after non-linear correction,the results show that it can reduce the non-linear relative deviation of CCD from 18% to less than 2%,effectively reduce the non-linear error of spectroradiometer,and get the evaluation value of uncertainty of 2%.It provides a basis for quantitative calculation of composite uncertainty.(4)By building stray light and non-linear error models,the problem of quantifying the systematic errors of spectroradiometer is solved,and the spectroradiometer can be calibrated accurately.In the wavelength calibration,the parameters of the data processing combination algorithm are optimized,and the wavelength calibration accuracy reaches 1/10 pixel wavelength interval.In the evaluation of the uncertainty of radiation calibration,the integration time control experiment proves that the accuracy's wavelength dependence is actually signal-to-noise ratio dependence,and the signal-to-noise ratio is proposed as an indicator parameter.Pixel-by-pixel calibration method solves the problem of wavelength dependence of calibration accuracy.The data show that the uncertainty of halogen tungsten lamp source calibration system in 2 hours is less than 0.5%.Then,on the basis of the uncertainty test and evaluation results of calibration system,combined with stray light and non-linear error characterization,the composite uncertainty is calculated.Finally,the composite uncertainty 3.8% is regard as the instrument accuracy,which meets the design requirements of the system and can be used for outdoor test and evaluation.(5)In order to validate the effectivity of model evaluation and accuracy optimization of error characteristics,application experiments and comparison test analysis are carried out.Firstly,the experiment of solar radiation observation is carried out.The analysis results show that the precise spectroradiometer can be effectively used for operational observation of ground-based meteorological spectral radiation and provide fine data for solar spectral analysis.After that,the outdoor comparison test analysis is carried out with the MS700 solar spectroradiometer of EKO Company in Japan(MS700 is the standard instrument for global meteorological spectral radiometry at present).The test results show that the calibration accuracy of the SVG spectroradiometer(3.8%)is better than that of the MS700(4%).Compared with the ISO9845 standard solar spectrum,the standard deviation of the wavelength distribution error of the SVG spectroradiometer is about 1%,which is lower than that of the MS700.The calibration accuracy and resolution bandwidth error of SVG spectroadiometer are better than MS700.The effectivity of error characterization model and calibration parameters optimization results is validated.By the study of error characterization and optimization control methods of spectroradiometer,the measurement accuracy(3.8%)of spectroradiometer is better than the standard accuracy requirement(<5%).The study of modeling and analysis evaluation methods provides an effective means for further improving the accuracy of spectroradiometer. |
PDF Full Text Request