【作者】 孙超；

【导师】 李平； 王海清；

【作者基本信息】 浙江大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 故障诊断技术的研究对于保障设备稳定运行，提高产品质量具有重要的意义。本文以一个实际橡胶混炼故障诊断专家系统(RMES)的构造过程作为案例，探讨了专家系统应用于故障诊断的优势和不足。本文主要做了如下工作： (1)探讨了使用通用程序设计软件和数据库管理软件进行复杂专家系统开发的流程。弥补了传统专家系统设计工具的设计风格单一，用户界面不友好以及无法和最新的理论发展结合的缺点。 (2)探讨了应用专家系统进行故障诊断时系统的结构设计。并对各功能模块进行了详细的分析。 (3)探讨了模糊理论在知识表示和知识推导方面如何和专家系统进行有效地结合，并在RMES中应用了研究成果进行知识库和推理机的设计。 (4)在知识库的设计上，采用知识分级的方式，使知识的表达更加具有层次性和合理性。通过各个知识表中的相互约束关系保证知识的一致性。 (5)在模糊推理中用故障阀值取代了一般专家系统规则阀值，使推理输出更具有合理性。此外在推理过程中设计了良好的人机互动界面使用户在推理的过程中可以学到很多专家知识。 (6)将人工神经网络应用到知识获取领域来解决专家系统知识获取的“瓶颈问题”，并提出了全新的知识获取方式。使规则的获取对领域专家的依赖性大大减少。 (7)提出了新的自动知识调整方法，使系统在运行过程中依然可以进行微调，使知识库内容更加合理。 (8)使用本文理论设计了橡胶质量预测专家系统(RMES)并在某大型橡胶炼制企业中取得了良好的应用效果。更多还原

【Abstract】 The research of fault diagnose technology has important significance on ensuring equipment operation stabilization and improving the product’s quality. This paper uses a design process of RMES as an example and discusses the advantage and disadvantage of expert system’s application on fault diagnose. The following results have been researched:(1) Discusses the development of complex Expert System by common program develop software and database management system; The new method provides a better design style, more friendly user’s interface and closer combination with the up-to-date theory than the Traditional Expert System develop tool’s.(2) Design the structure of fault diagnose expert system and analysis each function module’ structure particularly.(3) Discusses the application of fuzzy theory on expert system in the field of knowledge’s express and knowledge’s inference. And uses the fruit to construct the RMES’s knowledge base and inference engine.(4) Divide the knowledge to three level. The express of knowledge is more effective than the common expert system. Ensures the consistency of knowledge by design the restriction between the different knowledge table.(5) Use the fault’s valve value replace the rule’s valve value in common fuzzy expert system. The inference machine’s output is more reasonable, the RMES’s user interface has a good information communication between user and RMES and users can learn expert’s knowledge by use the system.(6) Discusses the application of neural network on expert system in the field of knowledge acquisition. Resolve the difficulty of knowledge acquisition. And declare a new method in knowledge acquisition. The dependence of expert is reduced.(7) Declare a new method of knowledge’s auto-adjust, the knowledge can be improved when the Expert System is running.(8) Develop the Rubber Mixing Quality Forecasting Expert System(RMES) by using this paper’s theory. And has a effective application in a large Rubber factory.更多还原