基于CREAM和贝叶斯网络的空管人误概率预测方法

针对研究管制人因可靠性时存在的模糊性和片面性问题，采用认知可靠性与失误分析方法（CREAM）中的扩展预测法，计算10项管制通用任务的人误概率；在此基础上，以管制行为形成因子作为根节点构建贝叶斯网络，建立其与情景控制模式的不确定关系模型，对管制员在多任务中的人误概率进行预测。研究结果表明:在由相同评判者给出行为形成因子影响效应的前提下，由CREAM扩展预测法和构建贝叶斯网络的方法预测得到的多数任务的人误概率差异较大，从方法的客观性、合理性和适用性角度分析，贝叶斯网络在研究该问题时更具优势。

Prediction method of human error probability in air traffic management based on CREAM and Bayesian network

Aiming at the problems of fuzziness and one-sidedness when researching the human reliability of air traffic control(ATC),the human error probabilities in ten generic task types(GTTs) of ATC were calculated based on the extended prediction method belonging to cognitive reliability and error analysis method(CREAM).On this basis,a Bayesian network(BN) which adopted the performance shaping factors(PSFs) of ATC as the root nodes was established,then its uncertain relation model with the contextual control models(COCOMs) was established to predict the human error probabilities of controllers in multiple tasks.The results showed that under the premise of the influential effect of PSFs provided by the same estimators,the human error probabilities of most tasks predicted by these two methods had great differences,and the BN was superior to solve this problem considering the objectivity,reasonability and feasibility of the methods.