【作者】 赵涛；

【导师】 严新平； 周新聪；

【作者基本信息】 武汉理工大学 ， 载运工具运用工程， 2002， 硕士

【摘要】 由于摩擦学故障的征兆随着时间和系统的变化而变化，因此，润滑剂样本含有的信息特征要能够充分反映这种变化，才能有效地诊断摩擦学故障。事实上，单一油液监测方法所能捕捉的信息特征是有限的，不能准确诊断出零件磨损的种类、磨损程度和磨损过程等。因此，必须融合来自不同油液监测方法的数据以获取更多的原始特征信息或不断发展信息提取方法以便从润滑剂样本中获取更丰富的二次特征信息。随着数据集成和信息融合技术的迅速发展，其在故障诊断中的应用也越来越广泛。 油液监测专家系统是一个计算机程序，具有油液监测领域专家的经验、知识推理策略和解决问题的能力，一般以求解那些需要人类专家才能求解的高难度问题或不良结果问题为特征。专家系统的产生和发展，为油液监测技术的研究者解决和克服油液监测技术遇到的困难提供了新的途径。 随着分布式技术更加成熟，应用更加广泛；分布式知识库、分布式人工智能的理论与技术的发展，多专家、多知识表示、多种推理决策机制、多知识库协同联合求解以及数据库、知识库、方法库的多库协同支持，使计算机应用已由传统的在封闭系统下用户各自独立解决问题、通过问题求解、执行一定的计算，朝着帮助用户进行交互支持(同步、异步)协同工作、共同解决各种复杂问题并为用户提供一个最大限度共享计算机资源(硬件、软件、数据库、知识库、一般程序或智能程序)的环境的方向发展。 因此，利用信息融合技术进行数据级、特征级和决策级层次上的监测和诊断，把专家系统用作信息融合的辅助工具，建立基于油液监测的多种技术集成的分布式故障诊断专家系统，则是实现机器油液监测故障诊断的智能信息处理的一条可行的重要途径。 本文把信息融合技术引入油液监测系统，利用CLIPS专家系统外壳构造了油液监测的知识库，实现了以柴油机为对象的油液监测的专家系统，作为整个系统的智能组件。信息融合技术和专家系统体现了本系统的智能化的特点。利用SQL Server数据库服务器建立了油液监测中央数据库，对油液监测实验数据、监测设备基本情况以及油品分析标准等进行有效地管理；针对不 武汉理工大学硕士学位论文同类型的监测实验设备特点和数据采集接口，编写了相应的数据采集或转换程序，为监测诊断提供数据基础；在此基础上，建立了以分布式组件对象模型为框架结构、以数据库服务器为数据维护工具的监测计算机软硬件网络。更多还原

【Abstract】 The features of tribology fault are changed along with the time and system. It is efficient to diagnose tribological fault in respect that information characteristics contained in lube samples tell the truth enough. As a matter of fact, information characteristics are limited by using a single monitoring technique in order to diagnose the type of wearing parts, degree of wear and wear process. So, the data fusion from different monitoring techniques is necessary to obtain more original characteristic information, or to continuously develop information extraction technique to acquire more secondhand characteristic information from lube samples. With the rapid development of the data integration and information fusion, it is widely applied into fault diagnosis.Expert system of oil monitoring is a computer program, which is provided with experience of domain experts, strategy of knowledge reasoning and ability of resolve problems. It is the characteristic thing for expert system to answer the difficulty of or badness results of purpose which human experts are requested to. With the produce and development of expert system, a new approach is presented for researcher in oil monitoring to put to the issue.With the maturity distributed technique and the multiple base synergism of database, knowledge base and method base, computer applications go forward from resolving problems respectively in a close system to cooperating synchronously or asynchronously with each other for customers to find a solution to all kinds of complex problems. Therefore, computer resources, which include hardware, software, database, knowledge base, normal programs and intelligent programs, will be shared to large extent.It is a of consequence approach for intelligently processing information on oil monitoring and fault diagnosis to establish the distributed fault diagnosis expert system based on the integration of multiple techniques, which utilizing information fusion techniques to go along with monitoring and diagnosis, and using expert system for auxiliary means.In this paper, information fusion is imported into oil monitoring system, and knowledge base of oil monitoring are constructed by making use of CLIPS expert system shell. Then, expert system of oil monitoring for making diesel engine is established, serving as intelligent components of the whole system. The intelligentization is materialized by information fusion and expert system. Oil monitoring databases are constructed by utilizing SQL Server, which efficiently manage the experimental data, fundamental state of monitoring unit and standard of oil analysis. According to different communication ports of experimental equipments, several data collection programs or data communication programs are wrote to provide database for monitoring and diagnosis. Based on all of these, a computer network is established, which uses distributed component object model for a framework, and employs database server as data maintenance tools.更多还原