| Railway external environment safety risk management and control is the main task of external environment governance at this stage.There are many kinds of hidden dangers in the external environment.How to effectively identify and assess the external environmental risk of railway,find out the key factors leading to the occurrence of risk events,determine the high-risk areas and locate the high-risk sources are important issues to be solved urgently to improve the efficiency of environmental governance and the level of risk management and control.From the perspective of risk identification and evaluation,based on the actual situation of the external environment,and based on the hidden danger investigation data and risk event data of China Railway Group in recent years,this paper analyzes the external environment risk profile.Then,from the three perspectives of risk event,risk area and risk description,we can make full use of data to recognize and mine environmental risks,help external environmental managers find the key of environmental governance,grasp the focus of governance work,realize the identification of high-risk line sections and the construction of section portraits,and put forward targeted risk management and control countermeasures,which can effectively improve risk management and control,so as to prevent the occurrence of risk events.The specific contents are as follows:(1)The environmental risk is recognized from the macro perspective by using the method of mathematical statistics.Analyze the number and type of hidden danger and risk event,and trace the distribution of hidden danger within 4km near the risk event point.It is concluded that floating objects are the main hidden dangers,and the main risk events are pedestrian access and light floating objects invasion.The tracing rate of floating object risk events is low,and the occurrence of risk events has greater uncertainty.(2)Through the complex network theory,the occurrence law of risk events is analyzed,and the key risks of external environment are identified and evaluated.By integrating the chain effect of 53 risk events,the event causation network is constructed,and the network topology is analyzed.It is concluded that risk events such as foreign body invasion and strong wind are important nodes in the network,and the impact of such risks can be cut off by increasing the height of the protective net and the closure rate of the fence.By analyzing the characteristics of the network,it is found that the average distance of the network is short,the risk is easy to spread,and the vulnerability of the network is strong when intentional events(such as typhoon and tornado)occur.(3)Through the construction of risk assessment system,combined with the line section division method,identify and evaluate the external environment risk area.The risk assessment system for single hidden danger and line section is constructed.The dynamic step advancing method and cumulative frequency method are used to determine the risk critical standard,and the bivariate interval filtering method is combined to realize the identification of high-risk section of Beijing Shanghai line.By comparing the division effect of high-risk sections under different parameters,it is concluded that the best identification result is 3 high-risk sections with a total length of19 km.(4)By marking the risk description text,the entity recognition model is trained,the risk entity is identified and the section portrait is constructed.After marking risk sources,risk sites and facilities as two types of risk entities,three feature templates are used to train conditional random field recognition model.The results show that the training effect of feature template is the best,which takes into account word and part of speech,and the window length is 3.The F value of two different entities is 98.6% and 94.58%respectively.The model has a good recognition effect on high-risk section of Beijing Shanghai line,which can effectively help to build the environmental portrait of the line section and intuitively show the environmental condition of the section.There are 27 figures,38 tables and 96 references. |
