Research On Quantitative Risk Analysis Of The Safety-critical Systems

Safety-critical systems(SCS) are extensively employed in the fields like avionics, spacecraft, energy, defense and transportation, which have high requirements for resource, response time, fault tolerant and special hardware, especially for dependability and safety. So, the major issues have become how to assure the dependability and safety of the SCS as well as control the cost of the development. Currently, part of the work of the analysis of the SCS is carried out in risk analysis. However, traditional risk analysis is still largely manual and informal activities, tool support is limited and artifacts are lack of being integrated with architectural descriptions, system interfaces, high-level behavioral descriptions or code. Architecture Analysis & Design Language(AADL) is good at describing the runtime architecture of the SCS, and its Error Model Annex(EMA) focuses on the behavior of error propagation in the SCS, but the direct risk quantitative analysis based on AADL model is not allowed. Addtionally, Risk-based Failure Mode & Effect Analysis(RFMEA) is more realistic to consider cost and severity which meet requirements for risk analysis in safety-critical area, yet it lacks the ability of uniformly describing the softwares and hardwares of system.In this thesis, combining AADL and RFMEA, a framework of component-based model which is used for risk quantitative analysis is proposed. To describe the RFMEA method, we extend the AADL EMA with RFMEA property. Then a method of risk analysis based on Error Propagation Graph(EPG) is put foward to realize risk quantitative analysis of the system. This paper presents our work as follows:(1) Combining AADL and RFMEA, a framework of component-based model which is used for risk quantitative analysis is proposed.(2) Modeling risk model with AADL: extending the EMA with the RFMEA property to make it possible to realize RFMEA based on the AADL model. Associating the extended EMA with the components.(3) Risk quantitative analysis based on the EPG: defining EPG based on the Dual-Graph Error Propagation Model, designing the mapping list between AADL model and EPG, completing the transformation from AADL model to EPG; A method is proposed to realize the risk quantitative analysis based on the EPG.Finally, designing a RFMEA OSATE plug-in and exemplifying the approach with the case study to illustrate the efficiency of this method. And our approach provides a new idea for the SCS risk analysis.