基于动态贝叶斯网络的石化装置火灾事故动态风险分析

为揭示石油炼化装置事故风险动态特性和事故情景演变路径,在对石化装置进行风险因素分析的基础上构建石化装置火灾事故故障树,基于贝叶斯网络非常规突发事故的演变过程,构建情景演变下的动态贝叶斯网络模型,在综合考虑应急措施的基础上,利用MATLAB软件和联合概率公式计算出各种事故场景的状态概率.以丙烯精馏装置火灾事故为例,结果表...     

Fatal workplace accidents in outsourced operations in the manufacturing industry

Several studies have indicated that outsourcing increases the risk of accidents and presented some explanations for this phenomenon. For example, higher accident proneness of external employees has been presented, but the common causes of accidents have not been reviewed in depth. This paper provides information about typical accidents, the contributing factors, and preventive measures of fatal occupational accidents that occurred in outsourced manufacturing tasks. This paper also compares whether these factors differ from accidents that occurred when tasks were performed in-house in the manufacturing industry. The focus is on operations executed in the factory area for or by an organization operating in the manufacturing business. The accident analysis is based on information gathered from accident reports for fatal workplace accidents that occurred in Finland during 1999–2008. At outsourced operations in manufacturing, accidents occur most commonly when installations or work preparations are being performed. According to the reports, dangerous work practices and insufficient hazard identification most frequently contributed to accidents. In order to prevent typical accidents, e.g., occupational instructional and guidance, hazard identification, work practices, supervision, and task planning should be improved. Statistical differences between outsourced and in-house operations were also found, mainly within the contributing factors. Therefore, the safety of outsourced manufacturing operations should be considered in detail in order to prevent accidents and ensure occupational safety also when operated with other performers.     

基于数据挖掘的化工生产事故致因主题抽取

为充分挖掘化工生产事故数据中的有效信息和潜在规律,提高对化工事故认知水平,针对某化工集团2010—2016年共1 578起事故数据,利用社会网络分析等方法揭示事故要素间的关联关系;运用潜在狄利克雷分配（LDA）模型进行事故聚类,并抽取到5个事故致因主题。研究结果表明:LDA主题模型等数据挖掘技术能有效挖掘大量事故数据中的潜在信息;5个事故致因主题中,4个涉及到人因或组织层面的缺陷;员工注意力不集中和现场风险管理不足这2个致因主题间具有较强相关性;员工注意力不集中、现场风险管理不足以及设备问题是导致事故发生的主要原因。     

基于贝叶斯网络的一种事故分析模型

贝叶斯网络被认为是人工智能研究中不确定性知识表示和推理的重要工具。当前在系统安全领域中已开始运用贝叶斯网络技术进行故障诊断分析，然而故障只是诱发事故的因素之一，无法系统的评价事故背后的隐患，对事故后果的预测也甚少涉及。笔者将贝叶斯网络作为一种事故分析手段，在事故致因理论的基础上提出了一种基于危险因素-事故-事故危害的三层贝叶斯网络拓扑模型；阐述了网络模型层次间的因果关联关系、各层次的构成、节点的描述方法以及网络模型的构建方法；最后通过一个天然气球罐的分析案例验证了该模型分析方法的可行性和有效性。     

Accidents at work during temporary agency work in Finland – Comparisons between certain major industries and other industries

Accidents at work during temporary agency work (TAW) are analysed and compared with those occurring in other industries in Finland on the basis of national statistics databases. The general trend is analysed between 1998 and 2007. The years 2006–2007 are analysed in more detail. Frequency distributions and accident frequencies are calculated for data analysis. Statistically significant differences between TAW and other industries are also tested. The results suggest that the accident risk is increasing in TAW and is higher than in other industries. This might be due to increased use of TAW in traditionally accident prone industries. However, workplace accidents were not so often severe in TAW compared to other industries. The results also indicate that certain work assignments, namely manual work in production and construction, etc. are more common in TAW compared to other industries, which affects the typical kind of accidents. Differences in workplace accidents between TAW and other industries should be taken into account in accident prevention. However, further research is needed to identify the underlying accident mechanisms and their significance in order to guide accident prevention effectively. Information retrieval concerning accidents at work in TAW should continue to be made available in the future.     

基于贝叶斯网络的煤矿事故人因推理

煤矿事故的不可重现性决定了事故原因的调查具有很强的不确定性,如何通过事故发生后的相关信息提高事故深层次原因调查的准确性是非常重要的。将HFACS与贝叶斯网络相结合,以煤矿事故HFACS分析结果为样本,通过卡方检验和让步比分析建立人因的贝叶斯网络因果图,进一步利用最大似然估计算法确定了煤矿事故人因的贝叶斯网络参数。最后,以双柳煤业顶板事故的调查信息为证据推理导致煤矿事故发生的深层次原因,提高事故原因调查的准确性,从而验证模型的有效性。     

Accident analysis model based on Bayesian Network and Evidential Reasoning approach

In this paper, an accident analysis model is proposed to develop the cost-efficient safety measures for preventing accidents. The model comprises two parts. In the first part, a quantitative accident analysis model is built by integrating Human Factors Analysis and Classification System (HFACS) with Bayesian Network (BN), which can be utilized to present the corresponding prevention measures. In the second part, the proposed prevention measures are ranked in a cost-effectiveness manner through Best-Fit method and Evidential Reasoning (ER) approach. A case study of vessel collision is analyzed as an illustration. The case study shows that the proposed model can be used to seek out accident causes and rank the derived safety measures from a cost-effectiveness perspective. The proposed model can provide accident investigators with a tool to generate cost-efficient safety intervention strategies.     

Workplace and organisational factors in accident analysis within the Food Industry

This paper focuses on the causes of accidents at work in the Food Industry Sector and it covers both immediate and systemic causation factors, using data collected in Portugal. In the first part, accident data is analysed by looking at harmonised variables within the Eurostat recording system, allowing a portrait of the accidents occurred and their immediate causes. To complement the study, a second part was designed to deepen the insight into underlying factors, as well as the relevant organisational conditions; this complementary part comprised an in-depth analysis of 30 accidents, carried out in the field by visiting several enterprises of the sector and conducting interviews with the injured people and their managers. For eliciting and analysing this information, the WAIT method was applied together with its classification schemes. The results are presented and discussed, showing the usefulness of certain new Eurostat variables, such as the deviation and the contact. However, they also demonstrate that the current variables are not yet sufficient to clarify accident mechanisms on which to build up knowledge and develop better prevention strategies. The authors argue for the need of more detailed information and propose an additional variable, associated with the deviation, aimed at promoting the inclusion of specific underlying factors within the local workplace environment.     

Prediction of occupational accident statistics and work time loss distributions using Bayesian analysis

This paper uses Bayesian analysis tools for the stochastic evaluation of work time losses due to occupational accidents in a workplace. Models are developed for accident frequencies, duration of recovery from an accident, and the worker unavailability. The unavailability statistics are hereby derived considering a two state stochastic model, to provide estimates for the expected work time losses over a base period of workplace operation. The above models are applied on real multiyear accident data collected from the Greek Petrochemical Industry.     

事故的组织管理原因文献综述

组织管理原因是一类重要的事故致因,对其全面了解才能有效预防事故。就组织管理原因对事故的作用机理、定义与内容、事故分析对其研究的现状进行文献综述。通过文献梳理发现,在一维链式事故模型和系统论为基础的事故模型中,组织管理原因对事故作用机理不同,但对事故发生都具有重要作用。现阶段研究中,组织管理原因没有统一的定义和内容,但具有的区别于其他事故致因的特性能得到共识。现阶段常见的事故分析模型对组织管理原因进行定量和定性分析,但仍需继续完善。最后,对组织管理原因的研究方向提出展望。     

基于FDA模型的地铁列车洪淹事故致因分析*

为了从安全信息视角探究地铁列车洪淹事故原因,基于FDA（Forecast-Decision-Action）事故致因模型和真实事故案例,构建由个体（地铁行车人员）、自组织（地铁运营企业）和他组织（应急主管部门）3个事故致因子模型组成的地铁列车洪淹事故FDA致因模型,并针对安全管理重要环节提出事故预防建议。研究结果表明:阻断事故致因主链或子链、保证必要安全信息充分且完备、防范安全欺骗行为,可有效预防地铁列车洪淹事故,通过实证案例分析验证FDA模型可显著提升地铁防洪应急管理水平。     

Analysis of equipment failures as contributors to chemical process accidents

A database study of chemical process accident cases was carried out. The objective of the study is was to identify the reasons for equipment based accidents. The most frequent accident causing equipment were piping (25%), reactors and storage tanks (both 14%) and process vessels (10% of equipment accidents). The six most accident-prone equipment is process related involve nearly 80% of accidents.78% of equipment accident contributors are technically oriented including design and human/technical interface faults. Purely human and organizational reasons are the most common accident contributors for storage tanks (33%), piping (18%) and heat transfer equipment (16% of causes). For other equipment the technical accident causes are most common.The accident contributors were divided to main and sub-contributors. On average process equipment failures have 2.2 contributors. The contributors, which frequent and act often as main contributors, should be focused. These risky contributors were identified for several equipment types. Also a deeper analysis of the accident causes and their interconnections was made. Based on the analysis a checklist of main risk factors was created for hazard identification on different types of equipment.     

“三高”气田钻井过程硫化氢泄漏动态风险分析

传统的H2S泄漏风险分析方法不能很好地对事故发展过程进行动态分析,导致分析结果偏离实际。基于贝叶斯方法,构建了高温、高压、高含硫（“三高”）气田钻井过程中H2S泄漏的蝴蝶结模型并提出将其转化为贝叶斯网络,在事故已发生的情况下更新基本事件发生的概率。然后,假定事故后果在确定的时间段内发生的累积次数已知的条件下,更新安全屏障及事故后果发生的概率,从而完成对H2S泄漏的动态风险分析。结果表明,该方法克服了传统静态定量分析方法中的不足,可动态评估导致H2S泄漏的基本事件发生的概率和对顶事件发生的影响程度,并动态反映安全屏障和事故后果的风险变化,能为钻井过程中H2S泄漏的风险分析及防控措施提供参考。     

Research on the prediction model of hazardous chemical road transportation accidents

In recent years, hazardous chemicals road transport accidents have occurred frequently, causing huge casualties and property losses, and accident risk assessment has become the focus of researchers' research. To predict the risk probability value of hazardous chemical road transport accidents, first, we compiled data on road transportation accidents of hazardous chemicals in China in the past five years. And the nine nodes in the Bayesian network (BN) structure were defined in combination with relevant classification standards. The optimal Bayesian network structure for hazardous chemical road transport accidents was determined based on the K2 algorithm and the causalities between the nodes. Second, the node conditional probabilities were derived by parameter learning of the model using Netica, and the validity of the model was verified using the 5-fold cross-validation method. Last, the Bayesian network model of hazardous chemical road transport accidents is used to analyze accident examples, and the accident chain of “rear-end-leakage” is predicted, and the accident is most likely to be disposed of within 3–9 h. The study shows that the derived accident prediction model for hazardous chemical road transportation can reason reasonably about the evolution of accident scenarios and determine the probability values of accident risks under different parameter conditions.     

我国化工企业火灾爆炸事故统计分析及事故表征物探讨

从事故统计分析的角度,对2001—2006年我国化工企业发生事故的设备、事故介质进行分析;对设备发生事故的规律以及事故介质的分布特征进行总结;对在火灾爆炸事故调查中具有普遍代表性的事故表征物:爆炸抛射物、火灾爆炸容器、喷溅物和烟尘及烟熏痕迹的特征进行分析;从定量、半定量化的角度,对事故原因进行探讨,事故分析结果,对于从理论的角度分析事故原因、有效预防事故都具有重要的作用。     

基于HFACS-BN模型的危险化学品事故人因路径分析*

为提高危险化学品安全管理水平,收集我国134起以人为主因的危险化学品事故,构建危险化学品事故的HFACS-BN模型,并基于贝叶斯网络对危险化学品事故中的人因路径及各因素灵敏度进行分析。研究结果表明:行为违规是导致事故的最主要不安全行为;在复杂路径中,环境因素和操作者状态具有较高的灵敏度,与其相关的风险认知与处理不当是阻碍事故预防的关键因素;在非复杂路径中,操作者状态与运行计划不当具有较高的灵敏度,其所涉及的监督违规、组织过程和组织氛围,是导致危险化学品事故的根本起点,并可依此分析事故直接原因和间接原因;操作者状态在2种路径中均表现出高灵敏度,因此化工企业在人员管理上要加强关注职工状态,减少不安全行为的出现。     

基于最优尺度Logit模型的交通事故形态致因分析

为剖析典型交通事故形态致因,在利用最优尺度分析法诊断交通事故形态致因共线性的基础上,筛选关键影响因素,构建无序多分类Logit交通事故致因模型并对模型进行参数标定,并选择100个交通事故样本对模型进行精度验证,该模型相对误差仅为4.0%,能够较为准确地分析典型交通事故形态致因。研究结果表明:路侧及中央隔离设施、照明是事故的保护性因子,可有效降低事故率;路侧行道树设置不合理或未及时修剪,使车辆不能有效识别道路信息,是正面碰撞的重要影响因素;机非混合道是事故的危险因子,增大了侧面碰撞事故发生概率。研究结果可为交通隐患精准识别、交通事故主动预防以及交通设施科学设计提供理论依据和技术支撑。     

基于多米诺效应的定量风险评价研究

多米诺效应是引发化工重大事故的主要原因之一.本文综合国内外的研究成果,对火灾热辐射、冲击波超压等造成的多米诺效应进行了深入分析,建立基于多米诺效应的定量风险评价模型,其中包括了评价流程、传播概率、阈值距离计算、多米诺效应对事故频率的影响及后果分析的内容.最后利用Matlab7.1计算平台,以汽油储罐进行实例分析,结果表明该方法是一种适用于多米诺效应定量评价的良好方法,能够比较科学、有效的对危险单元进行风险评价,使重大事故风险评价更切合实际,为政府监管部门和化工企业进行事故的控制和预防提供决策技术.     

基于行为安全“2-4”模型的受限空间事故分析及预防

为了研究受限空间作业过程的事故特性和致因机理,指导人员有效辨识受限空间作业风险,采取正确的应急救援措施。对2012年1月至2018年12月发生的典型受限空间作业过程事故进行了统计,并根据行为安全“2-4”模型对事故致因进行了分析,改变了传统事故原因的分类方法。将事故致因从一次性行为、习惯性行为、运行行为和指导行为4个方面进行划分,将安全管理体系划分为事故的根本原因。结果表明,受限空间作业事故伤害主要来源于气体中毒或窒息,盲目施救是导致受限空间人员伤亡的主要原因。基于行为安全“2-4”模型,提出从个人行为控制和组织行为控制两个层面的受限空间作业事故防控措施,对有效预防受限空间作业和救援事故,降低事故危害程度有指导意义。     

Database for accidents and incidents in the fuel ethanol industry

The purpose of the present research is to collect information about accidents and incidents that have occurred at fuel ethanol facilities from 1998 to October 2014, and to keep complete unified records of them in a database. The developed database contains general information about the accident or incident, its sequence, mitigation measures, its causes and consequences for humans, environment and for the plant. Until now, this information is not available. The work consisted in gathering information from different documental sources and subsequent organization in a database. It complements the previous work made for biodiesel industry and fills the existing gap in the field of ethanol. Knowledge about this information enables us to manage plant risks, since the accidents that are more likely to occur and the main sources of risk can be easily identified. Also, it makes it possible to exchange information with interested third parties. Statistical analysis shows that accident frequency has an oscillatory behavior, rising in the last year. Fire is the most common type of accident, while equipment mechanical failure is the main cause of accident. Partial material loss has been identified as the most common consequence. Finally, some conclusions are obtained concerning to the importance of having an updated and complete accident and incident database.