KNOWLEDGE-BASED FRAMEWORK TO AUTOMATE HAZOP ANALYSIS

危险与可操作性(HAZOP)分析是对 找出每一个可能的偏差，所有可能的异常原因 以及由此产生的不利的危险后果 化工厂中的偏差。现代化工厂非常 很复杂，因此很难从该HAZOP中进行分析和评估 因此，引起了与职业安全有关的严重关切。 由于这种复杂性，它们也更容易受到设备的攻击 最近的化工厂事故证明了这一点。少校 这项提议的目标是开发一个基于知识的系统框架，以 自动化加工厂的HAZOP分析。这样的自动化系统是 对提高化学品职业安全极为重要 植物。哈索普分析通常由一组专家进行研究 几周或几个月的工艺流程图。因此，哈索普分析是 这是一个非常困难、劳动密集型和耗时的过程。在我们的 提出了基于知识的框架，我们认识和开发了两个重要的 哈索普分析的特点：(I)即使每个哈索普分析都 独一无二--对一个过程来说，它是系统的和合乎逻辑的；和(Ii)许多方面 对于不同的工艺流程图，分析的结果是相同的。因此， 可通过使用基于知识的 系统。在这个项目中，我们提出了一种基于知识的系统方法 它在基于模型的框架中引入了几种新技术，以 解决这个问题。它们是：(1)将知识库分解为 过程特定知识和过程常识；(Ii)开发一般知识 各种工艺和工艺设备的因果模型，以及 (三)面向对象的实施框架。我们已经成功地 在一个小的原型案例研究上对所提出的方法进行了测试。 受到初步结果的鼓舞，我们建议对此进行调查 进一步开发更全面的通用模型库 复杂的工艺设备、工艺相互作用、工艺材料、 并通过复杂工业规模的哈索普案例研究进行测试。我们也 建议开发高效的搜索技术来管理 复杂性，处理回收和反馈循环，并开发智能 本系统采用面向对象的图形界面。

Hazard and operability (Hazop) analysis is the study of systematically identifying every conceivable deviation, all the possible abnormal causes for such deviation, and the adverse hazardous consequences of that deviation in a chemical plant. Modern chemical plants are extremely complex and hence are difficult to analyze and assess from this Hazop perspective, thus raising serious occupational safety related concerns. Because of this complexity, they are also more vulnerable to equipment failures, as witnessed by the recent chemical plant accidents. The major goal of this proposal is to develop a knowledge-based system framework for automating Hazop analysis of process plants. Such automated systems are extremely important for improving the occupational safety of chemical plants. Hazop analysis is often carried out by a group of experts poring over the process flowsheets for weeks or months. Thus, Hazop analysis is a very difficult, labor-intensive, and time-consuming process. In our proposed knowledge-based framework, we recognize and exploit two important features of the Hazop analysis: (i) even though each Hazop analysis is unique-to a process, it is systematic and logical; and (ii) many aspects of the analysis are the same for different process flow sheets. Thus, Hazop analysis can be automated through the use of knowledge-based systems. In this project, we propose a knowledge-based systems approach which introduces several novel techniques in a model-based framework to attack this problem. These are: (i) decomposing the knowledge-base into process specific and process general knowledge, (ii) developing generic cause-and-effect models of various processes and process equipments, and (iii) an object-oriented implementation framework. We have successfully tested the proposed approach on a small prototypical case study. Encouraged by the preliminary results, we propose to investigate this approach further by developing more comprehensive generic models library of complex process equipments, process interactions, process materials, and by testing on complex industrial scale Hazop case studies. We also propose to develop efficient search techniques for managing the complexity, handle recycle and feedback loops, and develop an intelligent object-oriented graphical interface for this system.