基于BP神经网络在森林防火专家系统不确定推理中的应用研究

提出将改进的BP神经网络应用于森林防火专家系统的不确定性推理中,其良好的自学习和泛化能力,可以解决基于可信度规则的知识表示在实际应用中导致的规则激增,推理速度缓慢的问题。该方法将不确定的知识用可信度区间表示,通过知识编码,设计并训练BP网络,最后用MATLAB进行仿真。实验结果表明:BP神经网络可以自动学习专家的典型经验,并且能将之准确的推广,隐含层神经元个数的确定和典型样本的选取决定了准确精度。在实际的专家系统不确定推理应用中具有应用价值。     

Handling and updating uncertain information in bow-tie analysis

Bow-tie analysis is a fairly new concept in risk assessment that can describe the relationships among different risk control parameters, such as causes, hazards and consequences to mitigate the likelihood of occurrence of unwanted events in an industrial system. It also facilitates the performance of quantitative risk analysis for an unwanted event providing a detailed investigation starting from basic causes to final consequences. The credibility of quantitative evaluation of the bow-tie is still a major concern since uncertainty, due to limited or missing data, often restricts the performance of analysis. The utilization of expert knowledge often provides an alternative for such a situation. However, it comes at the cost of possible uncertainties related to incompleteness (partial ignorance), imprecision (subjectivity), and lack of consensus (if multiple expert judgments are used). Further, if the bow-tie analysis is not flexible enough to incorporate new knowledge or evidence, it may undermine the purpose of risk assessment.Fuzzy set and evidence theory are capable of characterizing the uncertainty associated with expert knowledge. To minimize the overall uncertainty, fusing the knowledge of multiple experts and updating prior knowledge with new evidence are equally important in addition to addressing the uncertainties in the knowledge. This paper proposes a methodology to characterize the uncertainties, aggregate knowledge and update prior knowledge or evidence, if new data become available for the bow-tie analysis. A case study comprising a bow-tie for a typical offshore process facility has also been developed to describe the utility of this methodology in an industrial environment.     

基于知识推理的发电企业设备风险管理应用研究

基于发电企业设备自身的特点及工作人员对风险知识认知的局限性,有必要分析发电企业设备在运行、检修中潜在的风险.分析了设备风险知识与风险管理的关系,并以发电企业各类大型设备中应用最广泛的电动机为例,解决了电动机风险知识的表达及推理问题,为实现基于知识的设备风险管理提供了一种思路,对指导发电企业的设备日常维护、巡检以及状态监测、避免不必要的损失有着积极的意义.     

基于案例推理的城市典型灾害应急处置专家系统构建研究

全球自然灾害频发、强度不断增强,且城市化进程飞速发展,其所带来的相关城市灾害风险与日俱增.在重视灾害预报、预警的同时,城市灾害应急处置工作的有效组织与开展,亦是至关重要.把人工智能和城市灾害学知识相结合,通过构建灾害应急处置专家系统来解决该问题是目前的一个研究方向.本文运用案例推理等人工智能领域知识进行相关问题研究,阐述了如何对典型城市灾害案例进行知识化存储处理,给出了典型城市灾害分类推理规则和应急救援案例匹配计算方法.最后,利用VC++和CLIPS相结合,构建了基于案例决策的城市灾害应急处置专家系统.通过研究,充分验证了该模型方法的可行性,对于提高城市应急处置能力具有很好的现实意义.     

Ontology-based computer aid for the automation of HAZOP studies

Hazard and Operability (HAZOP) studies are conducted to identify and assess potential hazards which originate from processes, equipment, and process plants. These studies are human-centered processes that are time and labor-intensive. Also, extensive expertise and experience in the field of process safety engineering are required. There have been several attempts by different research groups to (semi-)automate HAZOP studies in the past. Within this research, a knowledge-based framework for the automatic generation of HAZOP worksheets was developed. Compared to other approaches, the focus is on representing semantic relationships between HAZOP relevant concepts under consideration of the degree of abstraction. In the course of this, expert knowledge from the process and plant safety (PPS) domain is embedded within the ontological model. Based on that, a reasoning algorithm based on semantic reasoners is developed to identify hazards and operability issues in a HAZOP similar manner. An advantage of the proposed method is that by modeling causal relationships between HAZOP concepts, automatically generated but meaningless scenarios can be avoided. The results of the enhanced causation model are high quality extended HAZOP worksheets. The developed methodology is applied within a case study that involves a hexane storage tank. The quality and quantity of the automatically generated results agree with the original worksheets. Thus the ontology-based reasoning algorithm is well-suited to identify hazardous scenarios and operability issues. Node-based analyses involving multiple process units can also be carried out by a slight adjustment of the method. The presented method can help to support HAZOP study participants and non-experts in conducting HAZOP studies.     

Fault propagation behavior study and root cause reasoning with dynamic Bayesian network based framework

The Bhopal disaster was a gas leak incident in India, considered the world's worst industrial disaster happened around process facilities. Nowadays the process facilities in petrochemical industries have becoming increasingly large and automatic. There are many risk factors with complex relationships among them. Unfortunately, some operators have poor access to abnormal situation management experience due to the lack of knowledge. However these interdependencies are seldom accounted for in current risk and safety analyses, which also belonged to the main factor causing Bhopal tragedy. Fault propagation behavior of process system is studied in this paper, and a dynamic Bayesian network based framework for root cause reasoning is proposed to deal with abnormal situation. It will help operators to fully understand the relationships among all the risk factors, identify the causes that lead to the abnormal situations, and consider all available safety measures to cope with the situation. Examples from a case study for process facilities are included to illustrate the effectiveness of the proposed approach. It also provides a method to help us do things better in the future and to make sure that another such terrible accident never happens again.     

石油钻井井控风险诱因专家调查分析

引发井控风险的因素很多,各因素对风险的影响也大小不一,做好井控风险诱因影响程度大小的评估工作是有效降低井喷事故发生率、减少事故损失的基础工作。专家调查法是一种简便易行的调查方法,但是组织大量的专家进行井控风险诱因的调查难度很大,本文采用改进的专家调查法,通过调研专家的文章、谈话、评论的方式来代替组织专家开会研讨的方法研究石油钻井井控风险诱因,改进的专家调查法所得出的调查结果与现场实际一致,表明了该方法科学、简便易行。这种对专家调查法的改进可以推广应用在其他行业上。     

A novel Dempster-Shafer theory-based approach with weighted average for failure mode and effects analysis under uncertainty

Risk priority number (RPN) is a commonly-used prioritization method in failure mode and effects analysis (FMEA) for systemic reliability and safety study. However, conventional RPN confronts wide criticisms, due to the neglect of the uncertainties of experts' opinions. It is important to handle the conflicts among experts' multi-opinions which could be described in imprecise, incomplete or crisp forms in lack of knowledge. In this paper, a novel method is proposed to alleviate and/or eliminate counter-intuitive behaviors against conflicts among multi-opinions under Dempster-Shafer theory. Firstly, the abnormity test for experts’ opinions is undertaken to identify and discard those distinct judgments. Then, an approach to determining the subjective weights of experts is proposed, which is used in weighted average for multi-opinions before calculating RPNs. Finally, the risk priority evaluation on main engine crankcase explosion failure on-board ship is implemented to verify the feasibility of the proposed approach.     

Fire risk management system for safe operation of large atmospheric storage tanks

This paper has been prepared by its authors to show the benefits coming from the application of the fire risk assessment methodology prepared by the “LastFire^© Project” group of experts. Now a days this methodology seems to be very effective to face the fire risk connected with large atmospheric storage tanks. It has been developed during the period 1997–2005 the HSE & operations experts from experts 16 worldwide operating oil companies. The authors, by “ad hoc” created software package, show the benefit coming from the applications of the LastFire^© methodology and how, from this, using the methodology and the supporting tad, realize a coherent fire strategy in the form of a Fire Risk Management System. Same workflow could also be extended to the issues connected with other problems related with large atmospheric tanks storing hydrocarbons, such as environmental impact by soil pollution, to create a common frame walk of assessment not can be shared with authorities as well. The proposed methodology, moving from the original project, sets as an effective “Fire Hazard Management Approach” in line with the most recent trends (even enforced by the regulations) shared at international levels, that propose the integration of risk assessment with prevention and protection measures selection based on achieved risk reduction, costs, time needed, etc.     

基于贝叶斯网络的复杂工程安全管理决策支持方法研究

为改进当前复杂工程安全风险管理中普遍存在偏重事前管理、过分依赖专家经验等不足,探讨复杂工程贝叶斯网络(BN)的构建流程,提出基于BN的安全管理辅助决策分析(DSA)方法,并以武汉越江地铁盾构隧道渗漏水事故为例进行实证研究。结果表明:利用BN能够融合多领域专家关于具体工程的实践经验,以有向图形式直观表达安全事故致因变量间的不确定性关系,在事前阶段运用正向推理技术预测事故发生概率以界定警情等级及早防范,事中运用重要度分析辨识关键致险因子以明确过程控制要点,事后运用反向推理技术快速诊断查明最可能致因组合,从而为复杂工程安全事故全过程管理提供实时辅助决策支持。     

Human error analysis in furnace start-up operation using HEART under intuitionistic fuzzy environment

To address human error in system reliability, Human Reliability Analysis (HRA) is an essential issue. Human Error Assessment and Reduction Technique (HEART) as a rather straightforward technique for HRA has successfully been used in many areas to predict human error probability (HEP). However, knowledge acquisition of experts during assessed proportion of affect (APOA) calculation is subjected to vagueness and ambiguity. To overcome this challenge, in this paper Intuitionistic Fuzzy (IF) set due to their advantage to represent more fuzzy information than a classical fuzzy set adopted through APOA calculation. To demonstrate this hybrid approach short for, IF-HEART, the furnace start-up operation is handled, since analysis shows that most of explosions and losses occur during furnace start-ups operation. Further, a sensitivity analysis is carried out to approve the proposed integrated approach. In addition to its academic contribution, the results of the paper enable to improve the overall safety level of a furnace by taking into account potential human error.     

基于模糊神经网络的火灾事故处理专家系统研究

以系统科学和行为科学理论为基础,结合人工神经网络、模糊推理与计算机技术,形成基于模糊神经网络的火灾事故处理专家系统,在火灾发生时为事故处理指挥人员提供良好的决策支持.本文首先给出了基于模糊神经网络的火灾事故处理专家系统的总体框架（任务管理模块,知识库与数据库管理和维护,火灾事故模型管理系统,解释机构,推理机控制系统,信号采集与处理以及结果输出模块）,并说明了各模块的功能;然后着重介绍了该专家系统中知识表示与获取方法,即将计算智能与符号智能结合起来解决专家系统知识获取的“瓶颈”问题;最后介绍该专家系统中知识库建立的方法以及所采用的推理机制,即模糊推理与规则推理相结合的推理方法.     

Factors contributing to commercial vehicle rear-end conflicts in China: A study using on-board event data recorders

Introduction: In the last 30 years, China has undergone a dramatic increase in vehicle ownership and a resulting escalation in the number of road crashes. Although crash figures are decreasing today, they remain high; it is therefore important to investigate crash causation mechanisms to further improve road safety in China. Method: To shed more light on the topic, naturalistic driving data was collected in Shanghai as part of the evaluation of a behavior-based safety service. The data collection included instrumenting 47 vehicles belonging to a commercial fleet with data acquisition systems. From the overall sample, 91 rear-end crash or near-crash (CNC) events, triggered by 24 drivers, were used in the analysis. The CNC were annotated by three researchers, through an expert assessment methodology based on videos and kinematic variables. Results: The results show that the main factor behind the rear-end CNC was the adoption of very small safety margins. In contrast to results from previous studies in the US, the following vehicles' drivers typically had their eyes on the road and reacted quickly in response to the evolving conflict in most events. When delayed reactions occurred, they were mainly due to driving-related visual scanning mismatches (e.g., mirror checks) rather than visual distraction. Finally, the study identified four main conflict scenarios that represent the typical development of rear-end conflicts in this data. Conclusions: The findings of this study have several practical applications, such as informing the specifications of in-vehicle safety measures and automated driving and providing input into the design of coaching/training procedures to improve the driving habits of drivers.     

Epistemic uncertainty in fault tree analysis approached by the evidence theory

Process plants may be subjected to dangerous events. Different methodologies are nowadays employed to identify failure events, that can lead to severe accidents, and to assess the relative probability of occurrence. As for rare events reliability data are generally poor, leading to a partial or incomplete knowledge of the process, the classical probabilistic approach can not be successfully used. Such an uncertainty, called epistemic uncertainty, can be treated by means of different methodologies, alternative to the probabilistic one. In this work, the Evidence Theory or Dempster–Shafer theory (DST) is proposed to deal with this kind of uncertainty. In particular, the classical Fault Tree Analysis (FTA) is considered when input data are supplied by experts in an interval form. The practical problem of information acquisition from experts is discussed and two realistic scenarios are proposed. A methodology to propagate such an uncertainty through the fault tree up to the Top Event (TE) and to determine the belief measures is supplied. The analysis is illustrated by means of two simple series/parallel systems. An application to a real industrial safety system is finally performed and discussed.     

基于模糊数学的石油化工装置HAZOP风险分析技术研究

对传统的HAZOP分析中偏差原因发生可能性进行量化。对于有统计数据的,根据行业数据、公司经验及企业事故建立HAZOP风险分析统计数据库;对于没有统计数据的HAZOP分析偏差原因发生概率,通过专家主观评判,用模糊数理论将专家自然语言转换为模糊数,采用左右模糊排序法将模糊数转换为模糊失效概率值。研究了偏差后果严重程度的划分标准,并根据偏差原因概率和偏差后果严重程度确定风险等级,利用风险矩阵得出偏差风险的大小。从而把HAZOP分析方法从定性改进为半定量的分析方法。据此对石油化工装置进行了HAZOP风险分析。     

Exploiting process plant digital representation for risk analysis

Hazard analysis is a crucial task in plant design. It is expensive in terms of money and time, and involves many specialists in different disciplines who are required to analyse aspects dispersed throughout the plant documentation. Product Lifecycle Management systems allow all the project data to be handled in a complete, integrated and consistent manner; they permit tools to be developed for capturing and reusing the design information necessary to evaluate specific plant aspects concerning potential hazards, thus providing automatic verification of some safety criteria. In this paper we present a knowledge-based tool aimed at supporting the expert in performing HAZOP studies in a plant project managed by a CAD/PLM system. The tool is based on an extended plant model that includes all the data and relationships representing the knowledge on which the supported hazard identification method is based.     

A novel tool for Bayesian reliability analysis using AHP as a framework for prior elicitation

Quantitative risk assessment (QRA) is a powerful and popular technique to support risk-based decisions. Unfortunately, QRAs are often hampered by significant uncertainty in the frequency of failure estimation for physical assets. This uncertainty is largely due to lack of quality failure data in published sources. The failure data may be limited, incompatible and/or outdated. Consequently, there is a need for robust methods and tools that can incorporate all available information to facilitate reliability analysis of critical assets such as pipelines, pressure vessels, rotating equipment, etc. This paper presents a novel practical approach that can be used to help overcome data scarcity issues in reliability analysis. A Bayesian framework is implemented to cohesively integrate objective data with expert opinion with the aim toward deriving time to failure distributions for physical assets. The Analytic Hierarchy Process is utilized to aggregate time to failure estimates from multiple experts to minimize biases and address inconsistencies in their estimates. These estimates are summarized in the form of informative priors that are implemented in a Bayesian update procedure for the Weibull distribution. The flexibility of the proposed methodology allows for efficiently dealing with data limitations. Application of the proposed approach is illustrated using a case study.     

Availability organisational analysis: Is it a hazard for safety?

This main issue of this article analyses the possible way to use for availability improvement, the organisational analysis methodology initially developed for accident safety investigations. As the last decade examples in the industrial world prove that some organisational weaknesses could either impact safety or availability, we have for purpose to make some important clarifications, with the help of the organisational paradigm, and grounded on our knowledge of safety accidents or local inquiries in hazardous technical complex systems.We will first give our definition of an availability event, by comparison with a safety event and recall what is for us an organisational analysis. Then we will consider the safety organisational paradigm pathogenic factors in wondering if these factors could also be seen as pathogenic factors for availability; or if specific availability pathogenic factors can be inferred from these safety pathogenic factors.In the end we will try to assess the common points and the differences between an availability oriented organisational analysis and a safety oriented one, with a particular attention to possible negative follows-up on safety issues and to the methodology issue.