BP网络应用于大气颗粒物的源解析

Application of BP neural network to sources apportionment of atmospheric particulates

应用BP网络对大气颗粒物进行源解析,将大气采集样本中的元素含量和大气颗粒物源成分谱构成训练样本集,用BP网络进行训练,由训练好的网络的权值可以计算出大气颗粒物的污染排放源的权重贡献率。将BP源解析法的计算结果与其它源解析法得到的结果比较,表明BP网络应用于大气颗粒物的源解析是可行的。

With the data of element contents in atmospheric samples and the source profiles of atmospheric particulates as training samples, BP network was applied to sources apportionment of atmospheric particulates. The weight contribution rates of sources of atmospheric particulates were calculated with the weights of trained BP network. Compared with the results from other methods, it shows that the application of BP neural network to sources apportionment of atmospheric particulates is practicable.