Research On The Propagation Model And Evolution Law Of Mobile Source Emission Based On Complex Networks

With the continuous increase of motor vehicles ownership in China,pollutant emissions continue to rise,and vehicle emissions are one of the main causes of air pollution.According to the 2022 China Mobile Source Environmental Management Annual Report,the national motor vehicle ownership reached 395 million in 2021,with a total pollutant emission of15.577 million tons.Pollution emissions from mobile sources have a negative impact on the ecological environment,especially in urban areas.Therefore,establishing a comprehensive data monitoring system and dissemination process representation method for mobile source pollution emissions is of great significance for protecting the ecological environment and achieving the "dual carbon" strategic goal.However,due to the high construction and maintenance costs of mobile source pollution emission detection equipment,the monitoring capacity of urban mobile source pollution emissions cannot meet current needs.Additionally,various factors interact and propagate during mobile source pollution emissions,which cannot be expressed clearly,making intelligent analysis of propagation and evolution characteristics challenging.To address these challenges,this paper introduces the theory of complex networks into the field of mobile source pollution emissions and constructs a mobile source pollution emission propagation network to conduct multi-level and multi-perspective network analysis.The main contents are as follows:First,a mobile source pollution emission propagation model is constructed.Aiming at the sparsity of mobile source pollution emission data,a data-driven emission data estimation method is used to obtain high-resolution mobile source pollution emission spatiotemporal data.Taking Beijing as a research case,the GPS trajectory data of vehicles in Beijing is used as input,and the pollution emission model is used to output the time series of mobile source pollution emissions.A causal relationship representation method based on convergent cross-mapping is designed to represent the regional propagation and causal relationships of mobile source pollution emissions,and the propagation process of mobile source pollution emissions is constructed as a complex network.Through analysis,statistical characteristics and high-weight paths of the mobile source pollution emission network are obtained,which lays the foundation for subsequent research.Secondly,a key node mining algorithm based on adjacency information entropy is proposed.According to the structural characteristics of the mobile source pollution emission network,for the problem of extracting topological structure information and node propagation characteristics in complex networks,a combination weighting centrality(CWC)index that integrates global and local features is designed.Experimental results on simulated networks and mobile source pollution emission networks show that the algorithm can extract both global and local structural information and effectively combine them,demonstrating better key node mining capability than existing methods and higher differentiation between nodes.Finally,in order to study the evolutionary laws of mobile source pollution emission networks,a link prediction algorithm based on network representation learning(AFS-BERT)is proposed.Link prediction algorithm can discover missing or potential edges in the network,thereby predicting the possible propagation relationship of future pollution emissions.A random walk method is designed that integrates two sampling modes,namely neighbor and non-neighbor sampling,to solve the sampling limitation problem of random walks.To address the dynamic representation problem of nodes with different meanings in the network,a BERT-based embedding method represents nodes as low dimensional dynamic vectors,which solves the disadvantage of single static vector information.Finally,the experimental results demonstrate the effectiveness of the method,with AUC values higher than those of the comparison algorithms on three datasets.Through the above research,this paper combines complex networks with mobile source pollution emissions to achieve full link technology research on mobile source pollution emissions data acquisition,emission network construction,propagation characteristics analysis,and network evolution prediction.It provides effective support for strengthening mobile source pollution prevention and regional collaborative governance.