Empirical Models For Prediction Of Daily Diffuse Solar Radiation On Horizontal Surface For Different Regions In China

Solar energy has been widely regarded as a proper alternative energy source for traditional energy due to the world energy consumption continues increasing and the renewable energy industry rapidly growing.Accessibility to accurate information on solar radiation is essential for the efficient conversion and comprehensive application of solar energy.Diffuse solar radiation is an important component of solar radiation and plays a significant role in the design and monitoring of solar energy projects,such as the thermal environment simulation of buildings,the design and application of photovoltaic systems,etc.However,diffuse solar radiation data are not easily accessible in most region in China since the number of stations recording diffuse solar radiation is still limited due to financial and technical limitations.Thus,it is necessary to estimate diffuse solar radiation by establishing mathematical models based on other readable available parameters.According to the continuously measured meteorological data and the solar radiation information from 1960 to 2016 in seven solar climatic zones in China,the characteristics of long-term variation trend of surface solar radiation were analyzed,the surface diffuse solar radiation predicting models were established and developed.The models were evaluated and compared based on statistical analysis and model comprehensive evaluation,and then the general diffuse solar radiation predicting models for each solar climatic zones were determined.Meanwhile,surface solar radiation has been severely weakened due to the air pollution.In this thesis,the models were modified based on the air quality data from 2014 to 2016.The main conclusions of this thesis are as follows:(1)The thesis investigated the characteristics of long-term variation trend of surface solar radiation in seven typical stations from 1960 to 2016 based on linear trend.Meanwhile,the tests of abrupt change were conducted by Mann-Kendall method.The results indicate that the surface global solar radiation generally decreased from 1960 to 2016,and Beijing shows the fastest decrease of–19.485MJ·m^-2·a^-1.Meanwhile,similar with the global process which referred as“Global dimming”and“Global brightening”,there was an significant rebound has occurred since 1990 in almost all stations.As for the interannual trend of surface diffuse solar dadiation,the results indicate that there were differences among different stations.(2)Based on the meteorological data and the solar radiation data from 1981 to2016 in China,several surface diffuse solar radiation predicting models were established and developed.The models were evaluated and compared using coefficient of determination(R²),mean absolute error(MAE),root mean square error(RMSE),mean absolute percentage error(MAPE)and Nash-Sutcliffe efficiency coefficient(NSE),and a global performance indicator.The results show that the diffuse fraction models performed better than diffuse coefficient models in estimating the diffuse solar radiation on daily basis in China.Moreover,the models achieved higher accuracy when considering both the clearness index and the relative sunshine duration as inputs,and the models that incorporated average air temperature and/or relative humidity showed higher accuracy.The proposed general models FN5,FN5,CN3,FN5 and the existing models FCS3,FCS3,FCS3 were the best-performing models in?–?s olar climatic zones,respectively,which could be used to predict the surface diffuse solar radiation in areas without diffuse solar radiation records.(3)The models were modified based on the air quality index(AQI)data from2014 to 2016,taking the representative cities in rapidly developing and heavily polluted regions as an example.The model performance of the modified models were compared with that of the original models.Validation of surface diffuse solar radiation measured in 6 representative cities indicated that values of R²,MAE,RMSE,MAPE and NSE of modified models improved by 0.272%,0.106%,0.617%,0.256%,and 0.159%on average,respectively.The results show that the accuracy of the diffuse solar radiation predicting models can be improved by modified based on AQI.