KNOWLEDGE-BASED FRAMEWORK TO AUTOMATE HAZOP ANALYSIS

基于知识的框架来自动化危险分析

• 批准号: 2277684
• 负责人: VENKAT VENKATASUBRAMANIAN
• 金额: $ 15.26万
• 依托单位: PURDUE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 1993
• 资助国家: 美国
• 起止时间: 1993-09-15 至 1996-09-14
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/2277684
• 关键词: artificial intelligence; chemical synthesis; industry; occupational hazard; occupational health /safety

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.

危害和可操作性（HAZOP）分析是系统的研究 识别所有可能的偏差，所有可能的异常原因 对于这种偏差以及这种偏差的危险后果 化学厂的偏差。现代化学植物极为 因此，复杂的和很难分析和评估这个Hazop 透视图，从而引发了严重的职业安全问题。 由于这种复杂性，它们也更容易受到设备的影响 由于最近发生的化学厂事故所见证的失败。专业 该建议的目标是为基于知识的系统框架开发 自动化过程工厂的HAZOP分析。这样的自动化系统是 对于改善化学的职业安全至关重要 植物。 HAZOP分析通常由一群专家进行 在流程表上持续数周或数月。因此，Hazop分析是 一个非常困难，劳动力密集和耗时的过程。在我们的 提出的基于知识的框架，我们认识并利用了两个重要的 HAZOP分析的特征：（i）即使每个Hazop分析都是 独特的过程，它是系统的和合乎逻辑的； （ii）许多方面 对于不同的过程流量表来说，分析的分析相同。因此， HAZOP分析可以通过使用基于知识的方式自动化 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 攻击这个问题。这些是：（i）将知识基础分解为 过程特定和过程通用知识，（ii）开发通用 各种过程和过程设备的因果模型，以及 （iii）面向对象的实现框架。我们已经成功 在小型典型案例研究中测试了所提出的方法。 在初步结果的鼓励下，我们建议调查这一点 通过开发更全面的通用模型库，进一步进一步接近 复杂的过程设备，过程互动，过程材料， 并通过对复杂的工业规模进行测试HAZOP案例研究。我们也是 建议开发有效的搜索技术来管理 复杂性，处理回收和反馈循环，并开发出智能的 该系统的面向对象的图形接口。