基于B-P神经网络优化算法的城市环境空气中PM_(10)浓度预测模型

Prediction Model of PM_(10) Concentration in Urban Atmosphere Based on Modifying Algorithm of B-P Neural Network

将B—P神经网络应用于西安市环境空气中P10浓度预测，对网络结构和算法进行了优化，建立了PM10浓度预测模型。经验证模型精确度比较高，PM10日平均浓度绝对误差0．015～0．020mg／m^3，相对误差在-20％～20％范围内。

This paper used B--P neural network to predict PM10 concentration of Xi ＂an city and modified structure and algorithm of network. The authors established a prediction model of the atmospheric pollutant concentration. It is proved that the accuracy of the model is high and the absolute error as well as relative error ranges form 0. 015mg/m^3 to 0. 020mg/ m^3 and -20% to 20%, respectively.