Research On Railway Accident Analysis Based On FRAM

With the rapid development of the railway in China,both the domestic high-speed rail network and urban rail transit have been greatly improved in speed and capacity.However,the railway,as a socio-technical system,which is increasingly in contact with the public's life,needs to deal with the complex and ever-changing operating environment and the coupling of internal functions.Failure to deal with this complexity of the system can result in serious traffic accidents,which can cause a heavy impact on the society and the railway industry.The traditional accident analysis method is difficult to systematically analyse the cause of the accident,but the analysis method based on system theory can solve the nonlinearity of the system caused by the complexity of the accident.Functional Resonance Analysis Method(FRAM)is the representative one.For the accident analysis of railway systems,the classical FRAM analysis method has some shortcomings in function identification,variability description and resonance analysis.Therefore,this paper proposes an improved FRAM analysis method for the functional resonance analysis of railway system accidents and applies it.The main contents of the dissertation are as follows:(1)Starting from the Rasmussen's system hierarchy,according to the meaning of functions in the abstraction hierarchy model,five different levels of the railway system function are constructed,which are core functional purpose,organizational abstract function,human-machine generalized function,operational physical function and physical implementation form.And the author proposes the identification of accident-related functions at different levels by means of the correlation of the accident-related agent and the analysed functional level.(2)The causes and manifestations of variabilities in different types of functions have been fully clarified and the effects on functional coupling have also been explained.In addition to the traditional language definitions at the perspectives of timing and precision,semi-quantitative methods are used to characterize the source variability,influence of system function coupling and the scenario effect wherein the semi-quantitative description of the scenario effect is represented by different degrees of scenario performance conditions.(3)The author proposes a preliminary analysis of the functional distribution of FRAM model by combining graph theory,and using Failure Logic Modelling theory on FRAM to complete the modeling of functional variability aggregation in accidents.Under the guidance of the functional resonance analysis algorithm of depth-first principle,the aggregation modes and results of functional variability in functional resonance are analysed.The vulnerabilities of accident-related functions are disceussed and the corresponding barrier measures are put forward.(4)The passenger train derailment accident of Sandiego de Compostela in Spain is taken as the object of accident analysis case.The supervision and operation process of the railway line is analysed in the organizational abstract function layer,and the related functions of the train daily operation in the accident are identified in the human-machine generalized function layer with its functional network diagram being built.The variabilities of different types of functions in the system are analysed,and the semi-quantitative description is completed.The functional variability aggregation model is used to analyse the functional resonance phenomenon in the accident and corresponding barrier measures are proposed for the vulnerability of related functions.Comparing the FRAM analysis results in the case study with the causal analysis of the accident in the official accident report and other accident analysis methods including Fault Tree Analysis and Accimap,the advantages and disadvantages of the FRAM analysis are discovered,which provides suggestions for the application and development of the FRAM method in the future to continuously improve the functional resonance analysis method.