基于BP神经网络的大气污染物浓度预测模型

A prediction model for atmospheric pollutant concentration based on BP neural network

针对目前大气污染物浓度预测精度不高的问题,本文提出了一种大气污染物浓度预测模型。对大气污染物浓度时间序列进行相空间重构,采用C-C法计算延迟时间和嵌入维数,在此基础上构建了网络结构为5-11-1的BP神经网络大气污染物浓度预测模型。采用某市SO2浓度监测数据进行仿真分析,并将BP神经网络的污染物浓度预测模型结果与SVM模型和ELM模型进行对比,BP神经网络对测试集的预测结果的均方根误差和平均相对误差分别为0.298和4.35%,预测精度更高,验证了本文所提污染物浓度预测模型的正确性和实用性。

In response to the current problem of low accuracy in predicting the concentration of atmospheric pollutants,this paper proposes a model for predicting the concentration of atmospheric pollutants.A phase space reconstruction was performed on the time series of atmospheric pollutant concentration,and the C-C method was used to calculate the delay time and embedding dimension.Based on this,a BP neural network prediction model for atmospheric pollutant concentration with a network structure of 5-11-1 was constructed.Using SO2 concentration monitoring data from a certain city for simulation analysis,the results of the pollutant concentration prediction model of the BP neural network were compared with the SVM model and ELM model.The root mean square error and average relative error of the prediction results of the BP neural network on the test set were 0.298 and 4.35%,respectively,with higher prediction accuracy,verifying the correctness and practicality of the pollutant concentration prediction model proposed in this paper.