基于神经网络模型的空气质量预测研究

Study on Air Quality Prediction Based on Neural Network Model

对空气质量的预测一直受到民众和研究者的关注。比较三种不同输入变量设置对神经网络模型预测效果的影响,得出AQI指标值与当日及滞后一日其他污染物浓度值相关系性较高,但与当日或滞后一日天气指标值的相关性较低,包含当日其他污染物指标的神经网络模型有着较高的预测精度,但应用意义较小。另一方面,仅依靠滞后一日污染物和天气数据不能实现对当日AQI指数的准确预测。根据研究结果,还对空气质量预测研究提出了相应的建议。

Prediction of air quality is focused by the public and researchers.The paper compares the effects of three different input variable settings on the prediction of neural network model.It is concluded that AQI index value is related to other pollutant concentration values of the same day and one day behind,but it is low related to the weather data of the day or one day behind.The neural network model,which contains other pollutant indexes of the day,has a high prediction accuracy,but it has little practical significance;on the other hand,it cannot accurately predict AQI index only by relying on the data of pollutants and weather ofter one day.According to the results,the paper also puts forward some suggestions on air quality prediction.