Research On Failure Mechanism And Spread Prevention Of Accident

Safety production accidents are incidents that cause harm or loss to people or things during production and operation activities.They always accompany production and operation activities and are difficult to completely avoid.Improving the ability to identify,predict,and control production safety risks has become a key means to prevent accidents,control their spread,and reduce social impact.Production accidents are affected by the coupling of multiple factors and have the characteristics of conductivity and derivatives.Therefore,studying the evolution process of accidents and constructing the conduction path of accidents is an effective method for accident prevention and control.The author takes the evolutionary process of work safety accidents as the main research object,establishes a dynamic evolution model of work safety accidents,and proposes corresponding prevention and control methods around the role of key nodes in the formation and transmission of accidents from the perspective of blocking or delaying the evolution of accidents.To realize the failure mechanism of safety production accidents.At present,methods such as failure mode and impact analysis(FMEA),fault tree(FTA),event tree(ETA),and probability model are often used to conduct qualitative and quantitative studies of accidents.They regard accidents as occurring in a certain specific sequence.The independent accident chain has formed some research results in many application fields.However,there are some problems with these methods:? The manufacturing system is considered static;? The subjective evaluation part of the expert is used as a part of the model parameters,which affects the objectivity of the diagnosis result;?cannot flexibly adapt to new information,or process includes Uncertain data including event dependencies;?In the face of large and complex systems,inferential diagnosis has low efficiency and poor interpretation of diagnostic results;?There are certain requirements for the magnitude of input data,and analysis and calculation are more complicated.Due to the use of backup components or backup production lines to respond to sudden conditions in the manufacturing process,modern production manufacturing systems have evolved from static systems to dynamic systems with certain self-healing functions,and the coupling relationship within the manufacturing system is becoming more and more complex..The effectiveness of prevention and control methods relying on existing accident models is inhibited when facing modern continuous manufacturing systems.That is,the uncertain topological structure,large number of production nodes,and complex node relationships in modern large-scale and complex continuous production and manufacturing systems have become difficult points for safety production accident prevention and control.Therefore,the establishment of an accident model adapted to the modern dynamic continuous production system,revealing the nature of the formation and evolution process of production accidents,and then proposed corresponding prevention and control methods to invalidate production accidents,becoming an urgent need for production safety guarantee.The failure of a safety production accident refers to the termination of a production accident or to make it work in a smaller area to reduce its area of influence.The main means to achieve the failure of safety production accidents is to strengthen the monitoring of key risk factors and cut off the transmission path of accidents.Therefore,on the basis of studying the factors of the production accident system and the impact of each part on the accident,the article proposes to establish a production accident evolution model from the internal and external parts of the production accident system,describe the transmission process of the production accident,and propose corresponding prevention measures.Control method,and finally achieve the purpose of safety production accident failure.Based on theories of complex networks,formal concept analysis,disaster science,and neuron learning mechanisms,this paper uses literature analysis,qualitative analysis,computer simulation and other methods to divide the safety production accident system,environmental data mobile acquisition network modeling,and internal/external accidents.System modeling,accident model dynamics feature simulation,accident network static feature analysis,and accident prevention and control methods focus on discussing the evolutionary prevention and control and failure of safety production accidents in modern dynamic continuous manufacturing systems.Its main work is summarized as follows.1.Based on system science,the theory of disasters was introduced into the field of work safety accidents,and a method for dividing the structure of work safety accident systems was defined.The method makes over the problem that the dimensions of the influencing factors are too high,and increases the complexity and calculation cost of the safety production accident system modeling;?people,things,objects,organizations,and the environment,except for the effects of production accidents,also on each other affect each other.Safety production accidents are the result of multiple factors,and a series of chain reactions occur in the process,which is similar to the formation mechanism and evolution process of natural disasters.Therefore,the theory of disaster science can be applied across disciplines in the field of safety production.In disaster science,the natural disaster system is divided into a disaster-preventing environment,a hazard-causing factor,and a disaster-bearing body based on whether the factor is the initiator or the receiver of the disaster.This paper refers to the natural disaster system and divides the production accident system into two subsystems according to the participation of risk factors in the manufacturing process:the induced environment(external influence factors)and the production system(internal influence factors).Induced environment refers to the natural environment(non-artificial environment such as geology,hydrology,etc.around the production and operation unit)and social environment(artificial environment around the production and operation unit),which are not part of the manufacturing system itself,which can adversely affect safety production and reach accidents.Rare or extreme events of the program.The production system is the main body of the manufacturing system,which is composed of the source components of the production accident,the participating components,and the directly affected components.Document analysis and text mining technology are used in this paper.The environmental system elements in the basic geographic information element data dictionary GB/T 20258.2-2006 and the accident description are used to extract the environmental system elements,and the production safety accident reports and safety production related regulations in Beijing are used to extract production.System elements,and finally establish the form of the basic elements of the incentive environment,the form of the basic elements of the production system.2.Based on the data collection time series triggering rules,a limited-time consistent mobile sensor acquisition collaborative control scheme(CCS)for environmental data is established to solve the problem of accurate collection and effective transmission of natural environmental data.Natural environmental data is one of the data foundations of the induced environmental system.Therefore,the acquisition of natural environment data has become one of the research contents in this paper.CCS consists of a wireless communication module,a direction decision module,and a motion control module,which solve the problems of stable transmission of environmental data acquisition,directional mobile network,automatic update of environmental model,and rapid change of sensor node orientation.The wireless communication module uses the data acquisition sequence as a trigger for the data transmission of the acquisition network,that is,the mobile sensor network starts communication only when the environmental data information changes,which greatly reduces the energy loss of the mobile wireless sensor network.The direction decision module uses a radial basis function to establish an environmental model,and maps the environmental attribute data transmitted by the wireless communication module to the environmental model in real time.The motion control module is based on continuous-time dual-integral dynamics,and designs a finite-time consistency controller.The gradient information of the environmental model is used to guide the sensor node's movement direction,so that the sensor node always follows the field information enrichment area,and solves the problem of invalid data collection.Compared with other data acquisition methods,this method is effective in both directional and non-directional communication networks.3.Based on the image edge detection technology,set theory and FCA theory,an incremental conceptual lattice construction method(SSIMAddExtent)is proposed to solve the mathematical expression problem of the dynamic evolution of the incentive environment system.The advantage of the SSIMAddExtent algorithm is that it reduces the number of iterations when constructing or updating the network,and can obtain obvious advantages at almost all test points.The state of the natural environment and the social environment around the production and operation unit is changing,that is,the incentive environmental system has dynamic characteristics.Therefore,modeling the evolution process of incentive environmental system is one of the research contents in this paper.Based on the theory of formal concept analysis,the incentive environment system is depicted as a hierarchical structure diagram(Hasse diagram)from top to bottom.Each node on the graph represents a class of features with the same attributes,and each edge represents the parent-child relationship between the nodes.In this paper,the cohesion of the center point of the feature class to the edge point of the feature class is transformed into the contour change degree of the class map.Based on this idea,the image edge detection technology and set theory were introduced into the FCA theory,and the problem of dynamic changes in the environmental system was transformed into the problem of structural similarity of the image.The SSIMAddExtent algorithm was established.Based on the Dirichlet function,the algorithm maps the classes represented by the nodes on the incentive environment hierarchy diagram to two-dimensional images,and the problem of the content changes of the classes is transformed into the structural similarity of the images.When the structural similarity of the map before and after the content of the class changes is greater than the set threshold,the corresponding node of the class is updated,otherwise it is unchanged.After that,it is only necessary to traverse the child nodes of the update node,adopt the set theory method to establish the final positioning of the new object,and implement the topology update of the incentive system network or the update of the node content.4.A reasonable fusion of dynamic disaster spreading model and neuron learning mechanism(STDP)is proposed.Based on causal reasoning theory,a dynamic model of accident spreading with edge weight evolution(MDAE-STDP)is proposed.It solves the weight invariance defect of the existing accident network model.While describing the dynamic spread of the accident,it also enables the production system structure to have autonomous learning capabilities driven by data,and more objectively reflects the elements of the real-world accident system.The strength of the interrelationship.As another component of the accident system structure,the production system is the main medium for the formation,evolution and conduction of accidents.Therefore,the evolution of accidents in production systems is one of the research contents in this paper.In this paper,the basic elements of the production system are used as nodes,the causality between the elements is connected,and the strength of the causality is used as the weight of the edges to describe the topological structure of the production system.Taking causal reasoning as the theoretical basis,the time-dynamics-based disaster spreading model and the neuron learning mechanism(STDP)were rationally combined to establish a dynamic model of accident spreading with edge weight evolution(MDAE-STDP).The causality of node pairs during the spreading process is used as the basis for edge weight modification.When there is a causal relationship between the node pairs,the connection weights increase in an exponential fashion.Otherwise,it decreases exponentially.5.Based on the complex network theory and the dynamic evolution characteristics of the production system,a method for preventing and controlling production safety accidents is proposed.On the basis of understanding the mechanism of accident formation and evolution,accident prevention and control achieves the purpose of prevention and control by weakening or blocking the path of accident formation and conduction.Because the inducement environment does not directly affect the manufacturing system itself,the accident prevention and control method is proposed based on the evolution characteristics of accidents in the production system.In this paper,we take the subway construction accident data in China and foreign railway accident data as an object,and based on the static characteristics of complex networks and important parameters of MDAE-STDP,we analyze the regional center node,overall center node,and sub-connected areas of the accident network in detail.Propose corresponding accident prevention and control methods.Such as:distinguishing different types of central nodes and defining their scope of impact in the accident;weakening the funnel effect in the production system and reducing the importance of key nodes;weakening the coupling of internal factors in the production system subsystem and reducing the coupling of network factors;isolation The coupling effect of the production subsystem blocks the connectivity of the production network;isolates the internal and external influence factor systems of the manufacturing system to block the internal and external factors' communication paths;improves the self-repair ability of the production system and reduces the risk of transitioning to accidents;Reduce the internal noise of the production system,reduce the role of internal interference in the production system;increase the time delay factor of the production system,delay the rapid transmission of the accident;weaken the causal relationship of the nodes in the production system,and destroy the accident's conduction path.