基于CPSO-BP神经网络的PM2.5浓度预测模型

Prediction Model of PM2.5 Concentration Based on CPSO-BP Neural Network

为了提高大气中PM2.5浓度的预测精度,采用平均影响值(MIV)算法筛选出对大气中PM2.5浓度有影响的主要变量,并依次作为神经网络输入变量。利用混沌粒子学(CPSO)算法修正BP神经网络初始权值和阈值,优化BP神经网络机构,以达到提高预测模型精度的目的。以2017年西安市PM2.5日均浓度数据为样本建立预测模型,实验结果表明:相比于传统BP神经网络,基于CPSO-BP神经网络预测性能更优。

In order to improve the prediction accuracy of PM2.5 concentration in the air, the main variables that have an effect on the concentration of PM2.5 in the air were filtered out by means of Mean Impact Value(MIV) algorithm and as input variables for the neural network in turn. By using the Chaos Particle Swarm Optimization(CPSO) algorithm, the initial weight and threshold of BP neural network were corrected and the BP neural network mechanism was optimized to improve the accuracy of the predictive model. The prediction model was established based on the data of average PM2.5 concentration per day in Xi’an City in 2017. The experimental results show that the prediction performance of CPSO-BP neural network is better than that of the traditional BP neural network.