面向地铁突发事件的行车调度系统人误预测研究

发布时间：2018-05-06 00:25

本文选题：人误预测技术 + 人误致因　；参考：《北京交通大学》2013年博士论文

【摘要】：摘要：近年来,随着地铁列车运行控制系统技术的发展,行车调度系统自动化程度越来越高,降低了由行车调度员失误引发事故的可能性,但与此同时,当地铁突发事件发生后,由于行车调度员处理不当而导致事故发生或结果恶化的问题日益凸显。行车调度系统是地铁安全运营的重要保障,行车调度员则是地铁突发事件下系统的最终保护和恢复机制,如何有效地避免和减少地铁突发事件中行车调度系统人误是各国地铁亟需解决的安全问题。为此,本文以国家高技术研究发展计划(863计划)“列车运行综合优化控制技术”为背景,从分析地铁行车调度系统人误致因出发,提出研究地铁突发事件下行车调度系统人误预测的两个关键方面——人误行为风险预测和人误诱发因素风险预测,提出基于场景的突发事件人误行为识别、预测指标定级和人误行为风险评价技术,同时提出考虑因果作用的人误诱发因素风险预测技术,并形成与提出的人误预测技术相匹配的规范化的行车调度系统人误数据采集和管理方法,为地铁行车调度系统人误预防与减少提供理论方法和技术支撑,具有重要的理论与实际应用价值。首先,对行车调度员突发事件处理任务和行车调度系统人机交互特点进行了详细分析,在此基础上构建了地铁行车调度员人误机理模型,从人误行为和人误诱发因素两个方面对人误机理进行描述,先后基于层次任务分析法和m-SHEL模型建立了人误行为分析模型和人误诱发因素识别模型,形成了以多资源理论为基础的行车调度员人误模式分类框架和以Jae W. Kim研究为基础的行车调度系统人误诱发因素备选库,并以国内某地铁公司1997-2011年间的98份行车调度员突发事件处理的人误分析报告为对象,通过灰色关联分析、数据挖掘技术和评分者信度评估验证了人误模式分类框架、人误诱发因素分类、人误模式分析模型和人误诱发因素识别模型的合理性和正确性。针对行车调度系统人误数据缺乏而引起的无法直接或间接计算、估计和预测人误率的问题,参考硬件故障模式、影响及危害性分析技术提出了以人误行为可能性、可恢复性和后果严重性为指标的人误行为风险预测模型,并结合行车调度系统人误模式分类框架确定了人误行为可能性等级划分标准；结合系统人误屏障的特点确定了人误行为可恢复性等级划分标准；结合地铁事故管理规定确定了人误行为后果严重性等级划分标准,形成了对人误行为风险进行评价和定级的度量图和风险度量矩阵。考虑到人误行为风险预测模型定量化的需要,参考硬件故障模式发生概率等级的评分准则和系统人误屏障的失误率减少作用确定了人误行为可能性和可恢复性指标的定量化定级标准,研究了基于贝塔分布的定量化人误数据的采集和估计方法,解决了人误率数据采集难的问题,设计并开发了基于典型任务的ATS模拟实验系统,通过视线追踪技术对定量化人误数据进行采集,完成了对采集方法的验证,为行车调度系统定量化人误数据收集提供思路。在建立的人误行为风险预测技术上研究突发事件场景人误行为风险的识别方法。根据行车调度员突发事件处理任务可模块化的特点,建立了基本任务模块库及相应的突发事件人误场景生成技术,以人误场景构建规则为对象,分析和确定了突发事件场景的失误后果严重性等级、可恢复性等级、可能性等级和风险等级的计算规则,以及关键人误行为识别技术,并通过对接触轨断电突发事件人误风险预测实例验证了方法的实用性。在人误诱发情景风险预测方面,充分考虑人误诱发因素间因果关系,以及因果关系而导致的其对系统人误的综合和整体效应的变化情况,构建了以模糊认知图来进行描述的人误诱发情景图模型,引入模糊认知图在因素关系推理、权重计算方面的理论成果,形成人误诱发情景影响效应的评价技术,完成了对某地铁公司人误诱发情景风险预测的实例分析。为了保证构建的图模型的正确性,采用证据理论和不一致性判断方法对专家判断进行综合,同时引入小样本数据筛选技术提高了图模型构建的效率。最后,根据建立的行车调度系统人误预测技术特点,对地铁行车调度系统人误数据需求和不确定性进行了详细分析,建立了规范化的人误数据采集方法。
[Abstract]:Summary :

In recent years , with the development of the technology of metro train operation control system , the automation degree of traffic dispatching system is becoming higher and higher , and the problem of accident occurrence or deterioration caused by improper handling of traffic dispatcher is becoming more and more obvious .

Firstly , a detailed analysis is carried out on the human - machine interaction characteristics of the traffic dispatcher ' s emergency handling task and the traffic dispatching system . Based on this , the human error mechanism model is constructed , and the human error - induced factor identification model based on multi - resource theory and the human error - induced factor identification model are established , and the human error - induced factor classification framework and the human error - induced factor identification model based on the multi - resource theory are established , and the rationality and the correctness of the human error - induced factor classification framework , the human error - induced factor classification , the human error mode analysis model and the human error - induced factor identification model are verified by the grey correlation analysis , data mining technology and the scoring person reliability evaluation .

Aiming at the problem of the lack of direct or indirect calculation , estimation and prediction of the human error rate caused by the lack of human error data in the traffic dispatching system , the human error behavior risk prediction model based on the possibility of human error behavior , the recoverability and the severity of the consequences is put forward with reference to the hardware failure mode , the influence and the hazard analysis technology , and the classification standard of the probability level of human error behavior is determined in combination with the human error pattern classification framework of the traffic dispatching system ;
Based on the characteristics of human error barrier in the system , the author determines the criterion of the recoverable grade of human error behavior ;
Combined with the regulations of metro accident management , the criterion of severity grade of human error behavior is determined , and the risk evaluation and risk measurement matrix of human error behavior risk are formed .

Considering the need of quantitative analysis of human error behavior risk prediction model , the quantitative order criterion of human error rate and recoverable index is determined with reference to the scoring criterion of probability level in hardware failure mode and the error rate reduction function of systematic human error barrier .

In this paper , the method of identifying the risk of human error behavior in the incident scene is studied based on the established human error behavior risk prediction technology . According to the characteristics of the modularization of the emergency handling task of the traffic dispatcher , the basic task module library and the corresponding incident human error scene generation technology are established , the human error scene construction rule is used as the object , the calculation rule of the severity level , the recoverability grade , the probability level and the risk level of the incident scene are analyzed and determined , and the key human error behavior identification technology is analyzed and determined , and the practicability of the method is verified by an example of the error risk prediction of the contact rail power - off emergency .

In the aspect of human error - induced scenario risk prediction , the human error - induced causal relationship between human error - induced factors and the change of the overall effect of human error - induced scenario are fully taken into account , and an example analysis of human error - induced scenario risk prediction based on fuzzy cognitive map is constructed . In order to ensure the correctness of the constructed graph model , the expert judgement is integrated by using the evidence theory and the inconsistency judgment method , and the efficiency of the construction of the graph model is improved by introducing small sample data screening technology .

Finally , according to the characteristics of the human error prediction technology of the established traffic dispatching system , a detailed analysis of the human error data requirement and uncertainty of the subway traffic dispatching system is carried out , and a standardized human error data acquisition method is established .

【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：U298;U231

【引证文献】