The Prediction Of Hammerstein Model For PM2.5 Based On PSO

With the continuous improvement of living standards and the progress of science and technology,the air pollution caused by industrial production and life become more and more serious.In recent years,haze weather produced a huge threat to human health,and cause the primary pollutant haze is PM2.5.Because of this,research on monitoring and forecasting of PM2.5 is particularly important.By observing the curve of the annual PM2.5 average value in Ningbo from 2013 to 2014,we find that the change trend is roughly periodic for the period of one year.Therefore,this paper will observe the air quality index value except for PM2.5 in 2013 as the sample data,and establish the mathematical model to predict the average concentration value of PM2.5 for one day.Firstly,this paper analyze the sample data normalization and principal component from the 13 dimensional data observed to 6 dimensional data.The complexity of the system is greatly simplified.Secondly,the 6 dimensional preprocessed data are as input.The predicted value of PM2.5 is as output.A multi input single output ARMA model is established to preliminary forecast the average concentration of PM2.5 in the air.The predicted residual sum of squares,the absolute error and the relative error are as the objective function to test the accuracy of prediction.Because of problems with nonlinear characteristics,we try to construct the Hammerstein model with nonlinear the input of the ARMA model and find that the prediction accuracy is improved by about 0.3028 of the ARMA model to about 0.1910.The prediction effect is significantly improved.Swarm intelligence optimization algorithm is an advanced and efficient processing method to solve extremal problem or optimization problem in recent years.The particle swarm optimization(PSO)algorithm is simple and easy to implement,and it is preferred estimation algorithm of model identification.The determine for the order of model and parameter estimation are the number of the complex calculation process,which can use PSO to solve these problems effectively.But in the practical application,the traditional PSO algorithm has slow convergence speed and the prediction precision is not very high.The PSO algorithm can be effectively improved the convergence speed and prediction accuracy.The specific implementation framework and idea of PM2.5 modeling is the new idea of this paper.