Optimization Of Performance Shaping Factor Of Nuclear Power Plant Operators Based On SEM-BN Model

In order to ensure the safe operation of nuclear power plants,it is extremely urgent to explore ways to reduce human error.Behavior forming factor is the cornerstone of human reliability analysis method.Different human reliability analysis methods have their own advantages and disadvantages,so optimizing the behavior forming factor of nuclear power plant operators has certain research significance.In this paper,the CPC system of cognitive reliability and human error analysis method(CREAM)is taken as the optimization object,a new factor classification system is proposed by combining structural equation model with Bayesian network,and an SEM-BN model for calculating the probability of human error in nuclear power plants is established.The main work of this paper is as follows.(1)Summarize the performance shaping factors of different human error analysis methods at home and abroad,and analyze their advantages and disadvantages and their improvement methods.To improve the prediction accuracy of human error probability,we can start with improving the correlation of behavior forming factors.(2)CREAM method has great advantages compared with other HRA methods in nuclear power plant application,but CPC definition is relatively vague and coincident,and it mainly depends on subjective judgment of experts when predicting human error probability in practical engineering application,which is highly uncertain.Combined with the root cause system of WANO(World Association of Nuclear Power Plant Operators),a new factor system after improving and optimizing CPC system is proposed,which includes the following eight contents:organizational management process;Communication and feedback;Availability of procedures/plans;Adequacy of training and experience;Adequacy of MMI and operational support;Working conditions;Time of day;Personal factors.(3)Taking the WANO event reports as the data source,the structural equation model(SEM)is used to verify and analyze the optimized factor system,which proves that it fits well with the actual data and is reasonable and feasible.Based on the fitting results of structural equation model,re-quantify the factor weights of the new classification,and improve the formula for calculating the probability of human error,which makes the calculation result more accurate than CREAM method.(4)Structural equation model is used to analyze the structure of the factor system,and eight factors are divided into three groups according to the strength of correlation.Three intermediate nodes are set up by expressing the correlation between factors through V-shaped structure,and then transformed into Bayesian network,which is the SEM-BN model for calculating the probability of human error.The reliability of SEM-BN model is verified and analyzed by examples.