论文部分内容阅读
利用北京地区5a水平面辐射及气象环境要素资料,统计散射比与气象环境因子的关系,因子诊断分析表明,散射比与PM2.5存在正相关性.引入PM2.5作为输入变量,建立多种“直散分离”模型,并结合观测资料进行预测检验.结果说明,引入PM2.5可提高“直散分离”模型的预测精度.结合总云量、降水及能见度数据,将天气划分为晴天、多云、阴天和雨天4种类型,进行不同天气类型下各模型预测误差分析,得到不同天气类型下最优“直散分离”模型.综合比较后得出,以清晰度指数、日照百分率和PM2.5为输入层的BP神经网络模型预测北京散射比的效果最好.“,”Based on the data of 5 a horizontal radiation and meteorological environment elements data in Beijing,the relationship between the diffuse ratio and the meteorological environment factor is analyzed ,the factor analysis shows that the diffuse ratio is positively correlated with the PM2.5. PM2.5 was introduced as input variable,various beam-diffuse radiation separation models were established,and combined with observational data to predict testing. The results indicate that the introduction of PM2.5 can improve the prediction accuracy of the beam-diffuse separation model. According to the data of total cloud cover,precipitation and visibility,the weather types are classified as 4 types, including clear,cloudy,overcast and rain day,each model prediction error analysis under different weather types was carried out,and the optimal beam-diffuse radiation separation model for each weather type was obtained. After comprehensive comparison,it is concluded that the BP neural network model based on clearness index,sunshine hours and PM2.5 as the input layer has the best result on predicting diffuse ratio in Beijing.