Intelligent Analysis Research Of Air Quality Propagation Relationship And Behavior Characteristics

Air quality is affected by multiple factors.Integrating the factors that affect air quality in a reasonable way and analysing system structure is of fundamental and key significance for the accurate analysis and improvement of air quality.Based on the propagation and changing characteristics of air quality system,the characteristic analysis of the whole,region and individual levels are carried out,predicting the future network structure.It intelligently excavates the hidden laws behind massive monitoring data.The research of air quality modeling,characteristic analysis and structure prediction is helpful to analyze the development trend of pollutants,divide the functional areas of environment,and formulate pollution control measures.The main contents are as follows:Firstly,the feasibility of applying complex network to air quality research is analyzed based on the network structure.This dissertation analyzes the current research status of air quality models and propagation network,finding out the research direction through the problems and short boards exposed in the existing research.Taking the construction of nodes and edges as the basic problem in model building.Then summarize various factors that affect air quality analyzing the characterization methods of air quality node attributes and interaction relations.Secondly,an air quality model based on propagation network is proposed.The rationality of the model is verified according to the overall characteristics of the network,which provides a new direction for air quality research.The model takes local station as node and constructs the transmission path between nodes after fully analyzing the pollutant transmission mechanism.This paper summarizes the correlation between main factors and monitoring data,putting forward the concept of communication cost,and takes it as the parameter of path construction.A statistical analysis method is proposed to generate the air quality network structure,which effectively improves the robustness of network structure.Based on the basic concept of propagation network,the rationality of network structure is verified from the small world and scale-free characteristics of the whole network.Comparing the high weight path in the network with the reality,the validity of the model is proved.Thirdly,the regional characteristics and key node mining methods of air quality system on propagation network model are studied.Based on community detection,the network is divided and the characteristics of the area are analyzed.In order to solve the problem of inaccurate results caused by the random spread of labels,new algorithms are proposed to improve the traditional label propagation algorithm under the heterogeneous structure of local node information.According to different types of air quality network,different strategies are applied to divide communities,analyzing the evolution mechanism and synchronous law of pollutants.The hidden law under the community structure are explored.Two key node mining algorithms are proposed,which mine key nodes based on node centrality and bidiretional transmission conduction.Among them,CC algorithm synthesizes three kinds of indexes: weight centrality,betweenness centrality and closeness centrality to comprehensively measure node importance.The BiTs algorithm is based on bidiretional transimission considers both in-link and out-link,and sorts node importance under the effect of attenuation factor.In order to evaluate the results of importance mining,an ITP model is proposed based on IC and LT model,which is suitable for the directed weighted network.Taking the total number of activated nodes in the model as the measurement index,it is proved that the key nodes mined by CC and BiTs algorithm have a greater influence on the network.Finally,based on the analysis of node similarity,the prediction method of network structure evolution is studied.Similarity is the full expression of node correlation characteristics.Nodes with high similarity tend to establish association relationship,which can predict the future trend of network structure.Based on Markov process,a new algorithm of node similarity representation is proposed,which is widely used in different types of networks.Walker in the network is used to simulate the spread process of pollutants,to get the high similarity node pairs.The method can predict the future generation trend of edges in the network,and analyse the evolution process of network structure.The accuracy of the algorithm is proved by the method of missing edge prediction,and the future path of pollutants in air quality network is predicted reasonably.In this dissertation,the propagation network and air quality research are combined together.The characteristics of propagation network are used to study air quality intelligently.We mine the law of pollutant transmission and the individual nature of monitoring stations based on the characteristics of the network.The future evolution trend of the network is predicted reasonably.The modeling analysis and characteristic research of air quality propagation network have a good hint for air quality improvement,function area division,source analysis,monitoring and prediction.