改进CREAM模型的船舶引航员人因可靠性预测

Improved CREAM Model for Prediction of Human Reliability Concerning Harbor Pilots

摘要: 为有效预测船舶引航员人因可靠性,以认知可靠性与失误分析方法(Cognitive Reliability and Error Analysis Method,CREAM)为基础,将模糊数引入通用行为条件(Common Performance Condition,CPC)绩效效应评价中,将先验条件概率从确定值转变为概率值,降低专家判断的主观因素对评价结果的影响。基于贝叶斯网络技术建立改进CREAM模型,确定控制模式并计算出人因差错率(Human Error Probability,HEP),根据结果分析出实际情景环境下引航员的人因可靠性。结果表明:改进的CREAM模型能获得特定情景环境下引航员HEP的精确值,相比CREAM基本法有更好的可靠性和敏感性,可为船舶引航员人因可靠性分析提供定量评估数据。

Abstract: The cognitive reliability and error analysis method（basic CREAM（Cognitive Reliability and Error Analysis Method） method） is improved by introducing fuzzy scoring into the performance evaluation on CPC（Common Performance Condition） factors. This makes the prior conditional probability a probability value instead of a definite value so as to handle experts’ subjectivity better. With the help of Bayesian network technology, the CREAM model decides the control mode and calculates the HEP（Human Error Probability）, and analyzes the human factor reliability of the pilots in actual scenario. The improvement of the model is verified with practical examples.