Study On Evaluation Methodology Of Nuclear Safety Class I&C Software

Nuclear power plant I&C system has already entered the stage of fully digital I&C system.As China nuclear power plants imported fully digital I&C system of nuclear power technology from abroad,China is facing the re-innovation problem again.Therefore,to build a Nuclear Safety Class I&C Software reliability evaluation scheme which is suitable for our national situation and for solving series problems as what and how to evaluate is of great significance for revealing theoretically the rationale and methodology of Nuclear Safety Class I&C Software reliability evaluation and for enhancing safety level of nuclear power production.This study is unfolded according to the logic "evaluation structure study-content study-evaluation methodology.study-evaluation model and arithmetic study-empirical study".The seven main research sections are as follows:In the first section,the present study reviews literature domestically and abroad from four domains,namely the complexity of Nuclear Safety Class I&C Software system,the cause for the unreliability of Nuclear Safety Class I&C Software system,the reliability characteristics and development law of Nuclear Safety Class I&C Software,and the evaluation methodology of Nuclear Safety Class I&C Software.In the second section,with respect to the specialty of Nuclear Safety Class I&C Software and according to the system quality program logic,the present study helps to define the research line of Nuclear Safety Class I&C Software in requirement orientation,process drive and evaluation methodology.Through analyzing weak links in reliability evaluation of present Nuclear Safety Class I&C Software,this study researches corresponding evaluation methodology,hierarchical structure and content from the perspective of reliability requirement stage,development process stage and evaluation stage,and proposes the dimensional structure of Nuclear Safety Class I&C Software reliability evaluation.In the third section,the Nuclear Safety Class I&C standard system is taken as the carrier for requirement research of reliability evaluation as per the research line of requirement as well as the establishment of requirement model for Bayesian Belief Network Evaluation Methodology.Namely,this study confirms the configuration of Nuclear Safety Class I&C System Standard applicable in China from the comparison of several types of existing standard system,which can be utilized as the gist for establishing Nuclear Safety Class I&C Software reliability requirement.In the fourth section,allowing for the specialty of Nuclear Safety Software as well as summarizing the experience and weakness in previous evaluation,the present study proposes the evaluation principle of quantitative evaluation methodology,in which the main quantitative evaluation methodology shall be evaluated respectively in line with the principle,and chooses the Bayesian Belief Network Methodology,the one with higher conformity,as the main quantitative evaluation methodology.Then this study objectively summarizes the shortcomings of Bayesian Belief Network Methodology for research program,comes up with improvement approach and establishes evaluation procedure and evaluation strategy.Since the Bayesian Belief Network Methodology has its weakness in transferring qualitative information into quantitative information,the author puts forward qualitative expert guidance methodology to remedy the weak links of Bayesian Methodology.In the fifth section,the establishment of multi-layer Bayesian reliability network model shall be finalized to reduce complexity of single system and increase the flexibility of model and methodology.As per the modeling principle of Bayesian reliability network methodology,the bottom layer model of Bayesian reliability network shall be established according to the development phase of software reliability;the top layer model of Bayesian reliability network shall be established as per the crucial element of software.In the sixth section,the research of quantitative method and quantitative procedures of Bayesian reliability network will be conducted.With the function of network reasoning and diagnosis,the Bayesian Belief Network can be used to implement analysis of the research project.The influence of parameter on network can be analyzed through sensitiveness analysis method.The section to be strengthened and improved for enhancing network stability will be judged through probabilistic method.Finally,the proposed Nuclear Safety Class I&C Software reliability evaluation scheme will be applied in a new nuclear power plant test reactor for application evaluation,based on which,the improvement suggestions will be proposed.Through practical application,the practicability of the proposed methods can be verified,gaining more experience for improvement.