摘要
在研究近几年西安市PM10污染的现状的基础上，初步选取8类20个气象因子，再采用主成分分析法进行精简，得到11个与PM10相关的主要因子，在此基础上，采用人工神经网络模型对西安市PM10污染状况进行预测，确定了网络模型结构。预测结果表明：预测值与实际值的相关系数达到0．801，在265个测试样本中，预测结果与实际完全吻合的为212天，占80％；相差不超过一级的天数为262天，占98．87％，与实际情况基本一致，效果理想。
The PM10 pollution of Xi＇ an city in recent years was introduced in this paper. Eight kinds, twenty s were selected, These factors were chosen by and eleven s were. The BP neural network model was built up by means of software MATLAB and was used to forecast Xi＇an PM10 pollution. The results indicated that the correlation coefficient of forecast value and actual value was 0. 801, of the 265 testing samples, the number of days when forecast value completely conformed to the actual value was 212, which accounted for 80%, no more than one grade, 262 days met the requirements, which accounted for 98.87%. The conclusion was intuitive and the effects were i-deal.
