Research On Air Quality Classification And Prediction Based On Machine Learning Methods

In recent years,the problem of air pollution caused by the rapid economic development has become increasingly prominent,and the impact on the health and daily life of the citizens has also become increasingly serious.Therefore,people who constantly improve their awareness of environmental protection are increasingly calling for the improvement of air quality.How to take good account of the dual elements of economic development and environmental protection,prevent serious air pollution incidents,and take scientific and effective air quality monitoring and prediction has become an important topic,and timely access to accurate and comprehensive information on future air pollution changes is the key to this topic.Effective air quality forecast information can help people to take preventive measures and plan reasonable travel arrangements,which is very conducive to the development of air pollution prevention and control work and urban environmental planning and construction,and has an important guiding significance for people's production and life and urban development.With the rapid development of technology,the emerging machine learning technology has greatly promoted the progress of artificial intelligence.Nowadays,machine learning has penetrated into all areas of the society,and machine learning has become an important means for researchers in many fields to solve difficult problems.This paper combines a swarm intelligence optimization algorithm with support vector machines,BP neural network,SARIMA and other models in machine learning to form a new model to study the air pollutant monitoring data of Taiyuan from 2014 to 2019,the main research content as follows:(1)The improved particle swarm optimization algorithm is combined with invasive weed optimization to optimize the initial weight and threshold of the BP neural network andconstruct the IWO-IPSO-BP classification model.Experimental results show that this model has a high classification accuracy for air quality and certain practical guiding significance.(2)In view of the influence of the parameters of support vector machine(SVM)on the performance of the model,Grey Wolf optimization algorithm and differential evolution algorithm are combined and optimized in this paper to construct GWO-DE-SVM classification model.The experimental results show that this model has a better classification effect.And compared with the IWO-IPSO-BP classification model in the previous chapter,this model has the characteristics of short time.Thus,this model has a higher feasibility.(3)A single prediction model has its own limitations.Therefore,when predicting a data sequence,the prediction result may be affected due to the inability to fully grasp the data information,that is,the accuracy of the prediction result of a single prediction model is not high.Therefore,the proposed SARIMA-SVR combined prediction model based on the combination of SARIMA model and SVR model is to comprehensively make use of the information provided by two single models and improve the prediction accuracy as much as possible.The experimental results show that the prediction accuracy of SARIMA-SVR combined prediction model is significantly better than that of single prediction model,which reduces the system error of prediction and improves the prediction effect significantly.Three models proposed in this paper,IWO-IPSO-BP,GWO-DE-SVM and SARIMA-SVR,are used in the classification and prediction of air quality of Taiyuan city.They provide a new method for the classification and prediction of air quality,and provide scientific and reasonable theoretical basis for air pollution prevention and control.