Model Construction And Application Research For Air Pollution Early Warning And Economic Loss Assessment

For the air pollution problem,digging out the change pattern from the historical data.understanding and mastering the evolution pattern and mechanism of the air pollution problem,can provide accurate air pollution warning and economic loss assessment information and establish reliable decision basis for people's daily travel and environmental management departments to formulate air pollution prevention and control measures.However.due to the characteristics of air pollution data such as nonlinearity,uncertainty and stochastic,and the complex relationship between air pollution and economic development,traditional models and analytical tools can hardly meet the current demand for complex data analysis and forecasting.In recent years,scholars at home and abroad have conducted some research and achieved some results on air pollution early warning and economic loss assessment.However,in air pollution early warning,most studies ignore the importance of data pre-processing,feature selection,multi-objective optimization of model parameters and fuzzy time series forecasting models,which largely limit the improvement of forecasting accuracy.In air pollution economic loss assessment,most of the existing studies focus on health economic loss assessment,and there are fewer studies on air pollution transportation economic loss assessment.How to discover new methods to establish a scientific and efficient air quality early warning system and to reasonably assess the economic loss caused by air pollution has become a key concern in the field of statistical forecasting.In order to improve the current research status,in the air pollution early warning research,this paper is based on fuzzy intelligent computing models such as fuzzy time series prediction model and machine learning prediction model,combined with data mining methods such as data pre-processing methods,feature selection algorithms and non-linear error correction strategies to carry out the establishment and application of air pollution early warning system research from different perspectives.The single-factor,multi-factor and interval air pollution early warning systems are established respectively,which provide effective and comprehensive air pollution early warning information and have important scientific significance and practical value.In the study of air pollution economic loss assessment,this paper quantitatively assesses the health economic loss and transportation economic loss caused by air pollution,which lays the foundation for the environment-related departments to carry out cost-benefit analysis studies and effectively promote the prevention and control of air pollution.The research content of this paper is divided into seven parts.Chapter 1 introduces the research background and research significance.On the basis of combing,summarizing and reviewing the current situation of domestic and foreign research,putting forward the research ideas and main research contents,emphasizing the main innovation points and shortcomings of this paper.Chapter 2 takes the construction of an air pollution early warning system as a starting point,discusses the base model and the evaluation framework,and finally introduces the theory underlying the air pollution economic loss assessment method proposed in this paper.Chapter 3 establishes a single-factor air pollution analysis and early warning system based on fuzzy theory and machine learning,and air pollution forecasting models based on machine learning and fuzzy time series forecasting models are developed for different forecasting objectives according to local conditions respectively.Chapter 4 establishes a multi-factor air pollution early warning system based on feature selection and fuzzy intelligent computation,proposes for the first time a multi-objective chaotic bonobo optimization algorithm,innovatively combines type-? fuzzy sets with artificial intelligence models and applies them to air pollution prediction,focusing on feature selection and fuzzy intelligent forecasting methods for multivariate data sets.Chapter 5 establishes an interval air pollution early warning system based on hesitant fuzzy sets and nonlinear error correction strategies,and for the first time combines hesitant fuzzy time series prediction models with multi-objective optimization algorithms and neural networks for pollutant concentration prediction research,with the research focusing on the processing and analysis of uncertain information.Chapter 6 provides a quantitative assessment and analysis of the health economic losses and transportation economic losses caused by air pollution,and provides information to support the cost-benefit analysis of environmental protection.Chapter 7 summarizes the full text of the research work and provides an outlook for future research.The research work in this paper focuses on the construction of an air pollution early warning system and the research and application of the economic loss assessment of air pollution,and the specific ideas and conclusions drawn mainly include the following aspects:(1)To address the problems that existing studies do not analyze and mine the complex characteristics of air pollution series deeply enough and ignore the importance of feature selection,this paper establishes two feature selection models,which successfully make up for the shortcomings of a single data pre-processing method.For one,the three-stage feature selection algorithm based on Hampel filtering technique,decomposition method and optimization algorithm can effectively weaken the influence of outliers and random noise on the forecasting performance and construct the optimal input structure.Second,the two-stage feature selection algorithm based on the linear Pearson correlation coefficient and Relief-F algorithm,in terms of multivariate feature selection.can dig deeper into the important features of the variables and further improve the generalization and forecasting ability of the model.In addition.to address the problems that most of the existing studies only consider the forecasting accuracy and ignore the forecasting stability,this paper uses a variety of advanced multi-objective optimization algorithms and proposes an improved multi-objective bonobo optimization algorithm to optimize the forecasting performance of the model.The experimental results found that the multi-objective optimization strategy can significantly improve the forecasting accuracy and stability.(2)To address the problems that the existing research is weak in dealing with air pollution uncertainty and the research on air pollution prediction models based on fuzzy theory is still at a preliminary stage,this paper improves and extends the traditional fuzzy time series forecasting model and establishes air pollution early warning systems based on type-? fuzzy set.type-? fuzzy set and hesitant fuzzy set,respectively.Chapter 3 adaptively divides the universe based on the information entropy discrete method,which improves the disadvantage of poor data adaptability of the traditional universe division method,and the experimental results prove that the forecasting model based on fuzzy theory has certain advantages in the field of air pollution prediction.Chapter 4 establishes an improved evolutionary interval type ? quantum fuzzy neural network and assigns weights to the input feature variables using the proposed multi-objective chaotic bonobo algorithm.The empirical study shows that combining interval type ? fuzzy sets and neural networks can better handle the uncertainty of multivariate data sets,which makes the established early warning system have the advantages of compact structure,high forecasting accuracy and high forecasting efficiency.Chapter 5 establishes a fuzzy time series forecasting method based on hesitant fuzzy sets,which extends the one-dimensional universe partitioning method to multidimensional.The experimental results show that the forecasting model based on hesitant fuzzy sets can divide and aggregate uncertain information,and can construct prediction intervals with higher integrated quality.In addition,this paper establishes a nonlinear error correction model for the hesitant fuzzy forecasting model,and the experimental results prove that the model can further improve the forecasting performance of the model and has a high feasibility.(3)Considering the existing air pollution research mostly focuses on single-factor point prediction and ignores the necessity of multi-factor and interval prediction,this paper conducts an in-depth study on multi-factor and interval air pollution warning systems.Firstly,a multi-factor air pollution early warning system based on feature selection and fuzzy intelligent computational methods is established,considering both pollutant concentration information and meteorological variables influence,and the potential relationships and interactions between variables are characterized and mined.The improved evolutionary interval type-? quantum fuzzy neural network can improve the forecasting accuracy and efficiency of multivariate data sets,which can help in air pollution prevention and control management decision making.Then,an interval air pollution warning system is established based on hesitant fuzzy sets,nonlinear error correction strategy and optimal distribution of error series,and the prediction interval is used to quantify and analyze the uncertainty information in air pollution data,which helps to improve the rationality of air pollution warning.(4)In response to the shortcomings of existing air pollution economic loss assessment studies that focus on health economic loss assessment and less on transportation economic loss assessment,this paper integrates the exposure-response model,the cost-of-illness method and the direct loss assessment method to conduct a comprehensive assessment of economic losses caused by air pollution from both health and transportation aspects.The experimental results show that the economic damage caused by air pollution to the transportation industry cannot be ignored.The main innovations of this paper include:(1)In terms of the overall structure,single-factor,multi-factor,interval air pollution early warning systems and air pollution economic loss assessment models are established to comprehensively study air pollution characteristics and economic loss impacts,and air pollution information is predicted and analyzed from different perspectives,with high theoretical research significance and practical value.(2)For data pre-processing,multi-stage feature selection models are constructed separately in different prediction dimensions to make up for the deficiencies of single data pre-processing methods.In this paper,a three-stage feature selection method is proposed in single-factor prediction,which significantly reduces the nonlinearity and complexity characteristics of air pollution series;a two-stage feature selection method is proposed in multi-factor forecasting to select valid input variables while eliminating redundant information to improve model efficiency.(3)With regard to model construction,the research in each section is cascading.The depth of application of fuzzy theory is gradually deepened,and the fuzzy sets cited in the forecasting models range from type-I fuzzy set,type-? fuzzy set to hesitant fuzzy set,enriching the theoretical study of fuzzy time series forecasting models in air pollution early warning.The universe partitioning method is gradually expanded from one-dimensional to multi-dimensional,which makes up for the shortcomings of the existing universe partitioning method and improves the model's ability to handle data uncertainty.In addition,this paper innovatively combines and extends fuzzy time series forecasting models with machine learning forecasting models and combines them with various performance enhancement strategies to establish an air pollution early warning system with different prediction dimensions that includes the processes of data pre-processing,model building and optimization,and system evaluation,which improves the feasibility and rationality of the air pollution early warning system.(4)In relation to air pollution economic loss assessment,this paper improves the current situation that most existing studies ignore the economic loss caused by air pollution to the transportation industry,and quantifies and assesses the macroeconomic loss caused by air pollution from both health and transportation aspects comprehensively,which makes up for the deficiency of less research on macroeconomic loss assessment of air pollution transportation industry,and helps to improve the early warning,prevention and control of urban air pollution.It is conducive to improving the scientific and rational cost-benefit analysis of urban air pollution warning,prevention and control.The shortcomings of this paper include:(1)The developed air pollution early warning system combines a variety of auxiliary models to improve the forecasting performance,for more expansion of research methods and tools to be continued to explore.(2)This paper only considers the influence of pollutant concentration and climate factors,and it is a direction for future research to introduce more influencing factors into the model under the condition of ensuring model calculation efficiency and prediction accuracy.(3)Due to the wide range of indirect economic losses and the lack of obvious boundaries in the calculation scope,the assessment of indirect economic losses of air pollution in this paper is not comprehensive and deserves further exploration in the future.