Study On Prediction Of Meteorological Potential Index Of Atmospheric Pollutants

In recent years,with the continuous development of industrial technology and the acceleration of urbanization,air pollution caused by human activities has become increas-ingly serious.Air pollution has many adverse effects on the environment,human health and social economy,such as destruction of the ozone layer,deterioration of the living environment,suspended particulate matter,carbon monoxide and other pollutants will cause people's lung inflammation,dermatitis,conjunctivitis,poisoning and so on.China is one of the countries with serious air pollution in the world,urban air pollution is more prominent.Therefore,the prevention and control of air pollution is imminent.How to grasp the opportunity of prevention and control,target the object of prevention and control,and determine the scope of prevention and control are the key points for local governments.Accurate control of the opportunity of prevention and control is the premise of efficient prevention and control.Air quality index(AQI)is a standard to measure the comprehensive air quality,but it can not accurately express the concentration of single pollutant,and its prediction model has a short prediction period,which may lead to the loss of the best prevention and control opportunity.In view of the limitation of AQI,this paper puts forward the meteorological potential index.Firstly,correlation analysis is used to find out the main meteorological factors affecting each pollutant.Secondly,BP neural network and improved BP neural network based on genetic algorithm and multi-objective optimization model are used to establish the prediction model of meteorological potential index,and then the model training is realized by Python tensorflow framework program-ming.Finally,the case is forecasted and the results of the two models are compared and analyzed.The first chapter mainly introduces the background and significance of the research on the prediction of meteorological potential index,as well as the research status of air quality mdex,the relationship between atmospheric pollutants and meteorological factors,BP neural network and genetic algorithm in recent years.The second chapter mainly studies the definition of meteorological potential index,the acquisition method of air pollutant data and meteorological data,the method of screening the main meteorological factors affecting air pollutants,the general steps and Realization of BP neural network and genetic algorithm,and the concept of multi-objective programming model solution.The third chapter,aiming at the limitations of the traditional BP neural network in the application of meteorological potential index prediction model,the BP neural network is improved by genetic algorithm,and the output of the improved BP neural network is quadratically optimized by combining the multi-objective optimization model.The fourth chapter mainly carries on the experiment analysis to the case,examines the feasibility of the method.On the basis of obtaining the concentration data of O3 and PM10 in Chongqing from 2016 to 2017 and meteorological data,the main meteorological factors affecting O3 and PM10 were screened by correlation analysis,and then the forecast models of O3 and PM10 were established by combining BP neural network with improved BP neural network based on genetic algorithm and multi-objective programming model,respectively.The results were compared and analyzed.