Research On Air Quality Prediction Based On Influencing Factor Optimization And Confidence Interval Modification

Air pollution is harmful to human health,and has attracted more and more attention from the government and people.As an important step in air pollution control,air quality prediction has become a hot topic of current research.Most of the traditional air quality prediction models only use the acquired air quality data.This method lacks the analysis and research of the internal variation law of air quality index.Besides,it lacks the consideration of meteorological factors,human activities and other air quality influencing factors.The prediction accuracy of the model is not good.In addition,the existing prediction algorithms do not constrain the prediction results by using threshold in predicting future air quality,which leads to the problem that the predicted values are too high or too low when predicting air quality.It affects the prediction effect of the algorithm.In view of the above problems,this paper carries out the research on air quality prediction based on influencing factor optimization and confidence interval modification.The main research contents of this paper are summarized as follows:(1)Establish an air quality index prediction model based on the optimization of influencing factors.The law of air quality change is studied in this paper.Based on the traditional air quality prediction model,an air quality index prediction model based on the optimization of influencing factors is established.In this paper,meteorological factors,cyclical factors and human activities factors are added to the traditional air quality prediction model.Besides,linear regression and neural network algorithm are used to compare the predictive ability of the traditional air quality prediction model with that of the air quality prediction model based on the optimization of influencing factors proposed in this paper.This paper compiles a web crawler to crawl the historical air quality data of Xi'an from 2014 to 2018,and takes it as the research data.The experimental results show that the air quality prediction model based on the optimization of influencing factors has better prediction ability under the same experimental algorithm.(2)Propose an air quality prediction modification strategy based on confidence interval.Aiming at the problem that the predicted value of air quality prediction algorithm is too high or too low when predicting air quality index,this paper proposes an air quality prediction modification strategy based on confidence interval.Firstly,this method resamples the historical data of air quality and establishes the interval prediction model of air quality in the coming week.Then,the interval of air quality index in the coming week is obtained by using the air quality data of 7 days before the forecast day as the input of the weekly prediction model.Besides,the method uses this interval as the confidence interval of air quality index of the forecast day.Finally,when the predicted value of air quality index exceeds the confidence interval,the predicted value of the algorithm is modified by using the modification coefficient.The prediction values of linear regression and neural network algorithm on the air quality prediction model proposed in this paper are modified by using the air quality prediction modification strategy based on confidence interval.The experimental results show that the proposed air quality prediction modification strategy based on confidence interval effectively solves the problem that the prediction value of the algorithm is too high or too low when predicting air quality,and has high application value.In summary,aiming at the shortcomings of traditional air quality prediction,this paper establishes an air quality index prediction model based on optimization of influencing factors and proposes an air quality prediction value modification strategy based on confidence interval.They effectively solve the problems of incomplete consideration of influencing factors and too high or too low prediction value in air quality prediction.