基于BP神经网络的空气质量预测与分析  

Prediction and analysis for air quality based on BP Neural Network

作　　者：田静毅[1] 范泽宣 孙丽华 

机构地区：[1]东北大学秦皇岛分校资源材料学院,河北秦皇岛066004 [2]东北大学环境污染控制研究所,辽宁沈阳110004 [3]秦皇岛市气象局,河北秦皇岛066004

出　　处：《辽宁科技大学学报》2015年第2期131-136,共6页Journal of University of Science and Technology Liaoning

基　　金：秦皇岛市科技局基金资助项目:秦皇岛空气重污染与气象因子分析(201301B030)

摘　　要：使用神经网络构造的算法,对秦皇岛市的空气质量进行预测。将秦皇岛市的气象监测数据与环境监测数据,按照季节关系分组,并进行相关性分析,从而确定出与空气质量呈强相关性的气象因子。并对其进行主成分分析,得出影响空气质量变化的主因子,以减少多种气象因子带来的数据处理难度。构建季节预测模型,并根据气象因子筛选结果,确定出神经网络模型需要输入的数据矩阵,从而通过气象参数变化对秦皇岛市的空气质量进行预测。并通过四个预测模型对季节空气质量的数值预测,得出了平均预测准确率,分别为81.18%,83.10%,81.72%,80.56%。结果表明,使用BP神经网络构建的秦皇岛市空气质量预测模型,可以成功预测该市四季的空气质量。Neural network was applied to the analysis of air pollution in Qinhuangdao city. In order to reduce the data complexity of modeling process,the correlation analysis between meteorological factors and environmental monitoring data of four seasons in Qinhuangdao city was completed to get the main meteorological factors affecting the air quality,according to the results of the data processing,get the neural inputing matrix,and the constructioning of the BP neural network model was completed after principal component analysis for the data,used to forecasting the air quality of the four seasons in Qinhuangdao city. The average prediction accuracy of the model was 85.18%,87.10%,85.72%,84.56%. The results show that the model of the four seasons can forecast the four seasons of air quality in Qinhuangdao city successfully.

关 键 词：空气质量 主成份分析 气象因子 神经网络模型 API预测 

分 类 号：X823[环境科学与工程—环境工程]