变电站操作不安全行为核心影响因素分析*

为分析变电站中操作不安全行为的核心影响因素，基于国家电网某省变电站员工安全心理自评数据，运用心理学模型确定9种主要心理维度,融合Bayesian Search和Peter Clark(PC)算法,并使用专家知识修正以训练贝叶斯网络模型结构，应用Expectation Maximization(EM)算法训练模型参数,并运用Mutual Information(MI)和预测推理2种方法分析核心影响因素。研究结果表明:训练模型的预测错误率为14.8%，AUC为0.945 8，变电站操作不安全行为的核心影响因素为情绪稳定性和心理承受力。

Analysis on core influencing factors of unsafe operation in substations

In order to analyze the core influencing factors of unsafe operation behavior in the substations,based on the psychological self-assessment data of substation employees in a certain province of the State Grid,firstly,nine main psychological dimensions were determined by using the psychological models.Then the integration of Bayesian Search algorithm,Peter Clark (PC) algorithm and expert knowledge correction was used to train the model structure of Bayesian network.The Expectation Maximization (EM) algorithm was used to train the model parameters,and two methods named Mutual Information (MI) and predictive inference were used to analyze the core influencing factors.The results showed that the prediction error rate of the training model was 14.8%,and the AUC was 0.945 8.The core influencing factors of unsafe operation behavior in the substation were the emotional stability and psychological endurance.