【作者】 王妙云；

【导师】 肖人彬；

【作者基本信息】 华中科技大学 ， 工业工程， 2009， 博士

【摘要】 在设备寿命周期中，初期故障期和耗损故障期的故障率较高，对设备运行可靠性的影响最大。因此要降低设备寿命周期费用、提高设备的运行可靠性，一是要改进设计和制造，消除影响设备安全高效运行的因素，提升设备的先天性能。二是要加强对设备运行状态的监测和故障诊断，发现设备早期故障隐患，及时进行操作调整，避免故障进一步发展，保障设备安全运行。本文针对这两个影响化工设备可靠性的主要因素进行研究，取得了很好的效果。首先，将质量功能配置(OFD)和设计结构矩阵(DSM)的思想和方法成功地引入到化工设备的设计和改造技术工作中。为了在尽量压缩设备设计和改造时间的同时，使设备设计和改造质量得到保证，本文巧妙地将质量功能配置和设计结构矩阵方法进行集成，通过引入“敏感度”和“进展度”两个指标将OFD与DSM有机的结合起来，充分发挥了两者的优势，在保证新产品开发质量的同时使开发时间压缩到最小。将该方法成功应用于某大型炼化企业烟气轮机改造设计工作中，较好地解决了机组原来运行中存在的问题，证明了该方法的可行性和有效性。其次，设备状态预报技术是生产、维修决策的重要技术基础，对保障设备运行的可靠性具有重要作用。针对目前设备状态预报功能缺乏而现场迫切需要的现状，本文采用人工神经网络预报方法对设备状态预报技术进行研究，首次提出了设备状态预报性能的稳定性问题，分析了影响预报性能的主要因素，提出了一种新的预报优化方法，解决了预报的稳定性问题，大大提高了预报精度。经过试验论证和现场实用，预报误差在10％以内，完全达到了工程实用的要求。第三，针对设备运行后期故障率上升的问题，本文对远程诊断系统的构架、监测分析方法库的建立、诊断流程的设计等方面进行了研究和实践，开发了一套实用的远程监测诊断系统平台，并在生产实际中得到了成功的应用。(1)针对目前设备故障分析方法不多，一些最新的行之有效的诊断分析方法不能很快得到推广应用的情况，开发了诊断分析方法库。除具有波形分析、频谱分析、轴心轨迹分析、多种综合趋势分析等常规的分析方法外，还重点开发了局部投影降噪、神经网络滤波等信号提纯技术，以及分形理论、小波分解等非线性信号处理技术等最新的分析诊断方法，并基于分形理论提出了机组运行稳定性的定量评估方法，将它们集成应用于设备运行状态分析，取得了很好的应用效果。(2)提出了基于CORBA组件模型(CCM)的分布式诊断系统构架，旨在隐藏系统底层实现细节，提高系统的可重用性和可移植性。通过对CCM软件组件思想的深入研究，给出了分布式诊断系统的CCM构架。针对系统的具体业务流程，将系统分为监测子系统和诊断子系统两部分，将监测对象、诊断方法和诊断策略抽象为CCM组件，构建了分布式诊断系统的整体框架。(3)开发了一套实用化的设备远程监测与故障诊断系统。根据企业实际，提出了远程诊断的现场级、企业级、远程级三层框架体系结构，对远程诊断中用户、专家、管理员等角色的职能和工作流程进行了周密的分析和设计，集成应用视频、音频、电子白板、WAP、手机短信、EMAIL、BBS等多种信息交互方式，以诊断请求为核心，实现了基于用户角色驱动的远程交互式诊断及专家会诊平台。(4)最后，简要介绍了本文开发的远程监测诊断系统的构成、功能及界面，并将该远程监测诊断系统应用于化工设备的监测诊断，大大提高了监测诊断的快捷性和准确性。现场应用情况表明，该系统监测数据准确、快捷，分析方法可靠，诊断功能强大，有利于充分发挥诊断专家的知识和经验，加快新技术的推广应用，及时发现设备运行中的早期故障隐患，提高设备运行的可靠性。更多还原

【Abstract】 There are two important aspects to improve operation reliability and reduce life-cycle costs of equipments. One is to improve design and manufacture so as to eliminate factors affecting safe and efficient operation of equipments and advance nature of them. The other is to enhance the monitoring and fault diagnosis of equipment running states, find out early fault hidden danger, operate adjustment in time, avoid further development of fault, and ensure safe operation of equipments. In this paper, the two main factors affecting chemical equipment relaiblity are discussed and a fine result has been achieved.At first, Quality Function Deployment (QFD) and Design Structure Matrix (DSM) are introduced to deisgn and improvement of chemical equipments. In order to compress equipment design and transformation time and to assure its design quality, the approach to integrating QFD and DSM is presented. Two factors named "sensibility" and "evolution" are introdcut to combine these two approaches so as to exert the superiorities of them and ensure product equlity and compress development time. Subsequently, a case study of gas turbine design of a certain large refinery and petrochemical enterprise is give and the result shows that the approach has preferably resovled problem existing in original aircrew, which is effective and feasible.Then, this thesis focuses on the researchment of structure of remote diagnosis system, the setup of signal analysis method library, and the plan of diagnosis program. A remote monitoring and diagnosis system is developed.(1) Firstly, aimming at the situation that new analysis technique can’t be quickly popularized in practice, a diagnosis methods library is developed. It includes not only traditional methods, for examples, vibration wave analysis, spectrum analysis, orbit analysis and trend analysis, but also some new technique up to date, such as noise depressing technique, non-linear signal processing technique, condition forecasting technique based on artificial neural network, and so on.(2) Secondly, applied forcatsting technique is demanded imminently in practice. This thesis introduces artificial neural network technique to predict machine-set’s oprating condition. Main factors that affect forecasting accuracy are discusseded minutely,. The result demonstrates that the ANN’s predictive accuracy is remarkably depended on the amplitude of noise in trained samples. Based on the conclusion above, a new optimizing approach is put forward. Experiments and applications prove that this method is applicable to all sorts of typical situation, and can meet the demand of production.(3) Thirdly, architecture of DRDS founded on CORBA component model(CCM) is proposed to hide the detail of system implementation and enhance the ability of reuse of system. On studying on CCM,approach for DRDS based on CCM is presented. Considering businesses of DRDS,the system is divided into two parts,monitoring subsystem and diagnosis subsystemwhere monitored objects,diagnosis methods and diagnosis rules are abstracted as CCM component,which constitutes the whole architecture of DRDS.(4) Fourthly, the remote monitoring and diagnosis system aims to commendably bring expert’s diagnosis experience into play to resolve problems in practice. So the diagnosis process is required to be concision, the diagnosis function to be abundant. After advantages of some distributed system are taken into account, a distributed system framework integrating manifold communication protocol is put forward. This framework is a three-level structure that includes local level, enterprise level, and remote level. The remote level consists of monitoring center, diagnosis center, case library, and self-help diagnosing module.This thesis carefully designed the work procedure of the user, the diagnosis expert, and the system administrator. Based on these works, an alternant diagnosis system is achieved. The system regards the user’s the diagnosis ask as the core. The user, the diagnosis expert, and the system administrator can conveniently communicate each other by means of video, audio, WAP, electrical mail, BBS, mobile message, etc.(5) Finally, the system’s composing, function and software interface are introduced in brief.This system’s monitoring data is veracious and in real time, it’s anlysis method is applied and advanced,and it’s diagnosis fuction is powerful. It is shown in pratice that the system can greatly improve the veracity and speediness of fault diagnisis.更多还原