Research Of A Safety-related Software Quantitative Assessment Methodology

The introduction of the digital technology enable the Distributed Control System(DCS)based on computers and microprocessors to provide higher reliability,better performance and more diagnostic functions.Compared with Analog Control Systems,the main difference is that DCS consists of software.One of the important steps of developing high reliability Digital Instrument Control System is to ensure the reliability of its software.Because DCS is composed of software,hardware and firmware,so that the failure mechanism is complex.It is extremely difficult to evaluate the reliability of software,especially the quantitative evaluation of the reliability of the security-level software.The existing methods of quantitative evaluation of software reliability include Software Reliability Growth Model,Bias Belief Net,Statistical Test,and Net Model Method.Software testing is an important means to ensure software reliability.It plays a crucial role in developing high quality software products.Even though a life cycle verification and validation(V&V)method is proposed to achieve the high reliability of software,the study of reliability quantitative evaluation method for high reliability software is still in the exploratory stage.Based-on probabilistic risk assessment and the physical simulation model,a software test case generation method fit for nuclear power plant safety digital control system is provided in this paper.The test cases generated by this method include risk-informed information and could qualitatively describe the software operation scenarios.After combining with Bayesian inference theory,the concept of fault-free testing is put forward,and A Safety-related Software Quantitative Assessment Methodology based on probabilistic risk analysis and physical model simulation is developed.Using a NPP protection subsystem software as an example,by analyzing the system failure mode and the subsystem software input space,the operational profile is built.Combining with a RELAP5 simulation model,the test cases for safety-related digital system software reliability testing are obtained.The test cases generated with the method could be the base for software reliability quantitative assessment of NPP safety-related digital control system.Finally,we realized the quantitative evaluation of software reliability by building an automation test platform.