Research On Cascading Failure Assessment Methods And Application For Ship Fire-fighting Systems Research On Cascading Failure Assessment Methods And Application For Ship Fire-fighting Systems

With the development of science technology and increasing demands of system functions,complex systems have some characteristics such as modularization,network,hierarchy,nonlinear etc.The mutual coupling between components makes system functions more powerful.At the same time,many accidents show cascading failure problems of complex systems.Once the components or subsystems fail,the domino failure effect is caused by correlations between components,which makes system performances degrade and results in enormous losses.Therefore,it is a important theoretical significance and practical value that cascading failure of complex system is researched.On the basis of mechanism analysis of cascading failure for complex systems,this dissertation applies system engineering,brittleness and complex network theory to study cascading failure assement methods from some aspects,such as cascading failure model,failure correlation and cascading failure indexs etc.The research is applied to ship fire-fighting systems.Firstly,cascading failure mechanism is analyzed by combining cascading failure features with dissipation and brittle castrophe theory.With bifurcation characteristics producing folding catastrophe and cusp catastrophe,this dissertation puts forward folding catastrophe entropy model and cusp catastrophe entropy model of cascading failure,and gives the conditions which results in brittleness catastrophe.Applying cusp catastrophe entropy model,a catastrophe condition of automatic fire-alarm sysem is discussed which arouses brittleness and causes cascading failure,and it shows cascading failure mechanism.A method is given that measures positive entropy and negative entropy.Secondly,cascading failure paths and brittle sources are identified and analyzed.Failure dynamicity and time sequence between components are described by markov chain,and triangular fuzzy numbers are used to describe failure rate of components.The dissertation puts forward a dynamic fault tree analysis method based on fuzzy markov chain,and assesses effectively cascading failure of complex systems.The model and analysis of ship automatic wet sprinkler system are done using dynamic fuzzy markov model.Further,identification of cascading failure paths and brittleness source is shown.In addition,a probability model of dual-power sources and hot-standby redundancy system is established with generalized markov process,in order to indicate the influence of brittle source on system reliability.Main power supply and standby power supply of automatic wet sprinkler system constitue a power supply unit which uses dual-power sources and hot-standby redundancy structure.The calculation and numerical simulation are done to verify the effectiveness of this method.Thirdly,node failure correlation degrees of complex systems are determine by combing multiplication weight method,Particle Swarm Optimization algorithm and grey theory.Node failure correlation degrees are presented which depicts coupling effect between subsystems by improving gray correlation analysis.A fire-alarm system is equivalent to a host and five subsystems which is a master-slave structure.Different subsystems failure attacked is seen pointcuts.Combining Adaptice Particle Swarm Optimization algorithm optimizes discrimination coefficient,it gets failure correlation degree between subsystems by combing multiplication weight method and grey correlation analysis.This dissertation applies these node correlation degrees to power grid disaster evolution model.By simulation,it shows that failure correlation infuences evolution speed and scale of disaster of systems.Finally,assessment of cascading failure due to load redistribution is carried out.Applying complex networks theory,subsystems are set as nodes.These are dynamic node correlation degrees,and the definition of load is also given.This dissertation constructs load-redistributed principle.After that,an improved cascading failure load-capacity model is proposed.The degree of overload and network transmission efficiency are used as failure evaluation indexs.The method is used to assessment cascading failure of automatic-fire alarm system whose loads are redistributed,and the failure simulations caused by random attacks and attempted attacks are done?The numerical simulation shows that cascading failure of automatic fire-alarm system not only is affected by the capacity and the structure,but also restricted by operation control management strategy of the system.