Study On A Reliability Evaluation Technology Of Nuclear Safety Critical Digital Instrumentation & Control Systems

The Instrumentation & Control(I&C)System is called the nerve center of the nuclear power plant(NPP)which determines the the safe,reliable and economical operation of nuclear power plant.Nowadays the Digital Instrumentation & Control Systems(DI&C)including safety critical DI&C systems are becoming more and more widely applied in NPP.Compared with the analogue Instrumentation & Control(AI&C)systems,the DI&C systems have incomparable superiorities in both measurement and control precision,logic processing,and economy.However,because of the distinguished characters of the DI&C systems,such as multiple-state and non monotone,the conventional methods are no longer suitable for the tasks of reliability analysis of DI&C systems.New reliability analysis technologies are required by nuclear industry to assess the reliability of DI&C systems.A DI&C system is composed of hardware and software.Both failures of hardware and software can lead to system failures,where a hardware component is a physical product,while a software code is a kind of abstract logic product.The failure mechanisms of hardware and software are essentially different.It requires different methodologies for the reliability analysis of hardware and software parts of a DI&C system.This thesis first analyzes the advantages of reliability modeling by using MFM.The system knowledge on both hardware and software of DI&C systems is organized in decision tables of MFM and can describe the dynamic characteristics of DI&C systems.A DI&C system is a typical system with multiple-state and its minimum failure mode is no longer mini cut sets but prime implicant sets(PIS).Obtaining PIS of DI&C systems is very difficult and there is not commercial reliability analysis software tool that can deal with the multiple-state system problems.This thesis presents a method of obtaining PIS by two major steps.The first step is to divide the process of obtaining PIS into two layers of set operation.The first layer of set operation uses Isograph Reliability Workbench(RWB),a mature and commercial reliability analysis software tool,to get implicant sets.The second layer of set operation handles the relationship between the multiple-states of a component.This thesis puts forward a algorithm suitable for the DI&C systems to obtain the PIS based on the cube representation method.The disjoint operation of implicant sets is discussed and a disjoint algorithm is given based on the cube representation.The proposed method is demonstrated bytaking the digital reactor protection system as an example.It shows that the algorithm is suitable for the two sate non monotonic system,and is also suitable for the the monotone and non monotone systems with multiple-state.Lots of software is applied in DI&C systems,and evaluation of software reliability is an issue that must be considered in assessing the reliability of DI&C systems.A Software Reliability Auto-Modeling and Analysis Platform(SRAMAP)is developed based on Flow Network Model(FNM)and Bayesian Belief Networks(BBN).SRAMAP can automatically establish Flow Network Model of the software source code and analyze the software reliability.SRAMAP can also provide sensitivity analysis of the code and identify the key testing paths which will save the test cost significantly especially when the code is complex.