Abstract:

Traditional HAZOP analysis is conducted through experts’ brainstorming discussions, which is time-consuming and labor-intensive and cannot guarantee the completeness of the analysis results. To resolve the problem, the HAZOP causal knowledge description model based on ontology is proposed. The study first extracted common knowledge from traditional HAZOP analysis texts, and obtained the HAZOP causal knowledge description model structure expressed in natural language. With the help of ontology, the structural modification of the model is realized through the definition of classes, attributes and instances in HAZOP causal knowledge. Finally, the study used ontology software named Protégé to express the model through OWL language, and established an ontology-based HAZOP causal knowledge description model. The model was used for finding the causes and consequences of liquid level changing to verify its reasoning process. The results shows that the proposed model can assist artificial HAZOP reasoning analysis process, and at the same time explore the shallow and deep knowledge of the HAZOP analysis process to ensure the depth and completeness of analysis results.

Key words: hazard and operability analysis (HAZOP), safety, knowledge ontology, model, knowledge inference

摘要：

传统HAZOP分析通过专家头脑风暴的讨论形式进行，耗时耗力且无法保证分析结果完备性。针对此问题，本文在语义模型改进HAZOP分析方法的研究领域，提出一种基于本体的HAZOP因果知识描述模型。本文首先对传统HAZOP分析文本抽取共性知识，获得自然语言表述的HAZOP因果知识描述模型结构；再借助本体论的方法，通过对HAZOP因果知识中类、属性、实例的定义实现模型结构化修正；最后，使用本体开发软件Protégé将HAZOP因果知识描述模型通过OWL语言进行表示，构建了基于本体的HAZOP因果知识描述模型，可辅助人工HAZOP分析，提高人工HAZOP分析的效率，同时保障分析结果的完备性。本文以寻找储罐液位变化的致因、后果为例，对该HAZOP因果知识本体的推理过程进行验证，结果表明，该HAZOP因果知识本体可辅助人工推理分析过程，同时发掘HAZOP分析过程的浅层、深层知识，保障分析结果的深度和完备性。

关键词: 危险与可操作性分析, 安全, 知识本体, 模型, 知识推理