基于支座振动响应与知识网格的旋转机械故障诊断方法

Fault diagnosis of rotary machinery become difficult because of the uncertainty of failure type, time and position of failure occurring, and the inconvenience of sensor installation. Based on the vibration response in the base of the rotary machine and knowledge grid, a novel fault diagnosis methodology is proposed in this proposal. The following researches are covered: analyze failure mechanism of typical failures, such as unbalance, crack, loose, rub-impact, bearing fault; establish the dynamical model of the rotor system including the base of rotary machine; validate and modify the dynamical model by the results of finite element analysis and experiment; investigate the dynamical characteristic of the vibration response of the base under typical failures; establish sensitivity quantities of typical failures; build grid knowledge library about multi-mode fault features; advance intelligent knowledge-grid recognition method for the multi-mode fault features. The proposed method is compatible for failure types, time and position of failure occurring, and is significant to develop general standard for sensor fixture design and manufacture in the design stage of rotary machine, as well as the standard for online monitoring and fault warning and diagnosing of such machine. The studies in this proposal are ease to explore the fault diagnosis system, and to apply in practice. As a result, a new fault diagnosis ideal and theory can be proposed and developed for theoretical research and engineering applications of rotary machinery.

项目针对旋转机械故障诊断技术现场应用中，故障发生的类型、时间、位置未知，传感器安装不便等原因，导致现场有效的故障预警诊断困难的问题，提出基于支座振动响应与知识网格的旋转机械故障诊断方法。研究内容包括：转子各类典型故障(不平衡、裂纹、松动、碰摩、轴承故障等)机理分析；建立含支座振动响应的各类典型故障转子系统动力学模型；有限元仿真与实验对模型进行验证与修正；研究各类典型故障的支座振动响应特性，构建可表征各类相应故障的敏感特征量；构建多模式故障特征网格知识库，研究网格环境下多模式故障特征的知识网格智能识别方法。项目研究对故障发生类型、时间、位置具有适用性，有利于形成在设备设计制造时就将传感器安装结构设计和制造好的通用标准；有利于形成设备运行监测和故障预警诊断系统的标准、固化构建程式，易于系统开发和现场推广应用，为机械设备故障诊断领域理论研究与工程应用提出了一种新的旋转机械故障预警诊断理念和思路。

汽轮机、电机、风机、直升机，航空发动机等旋转机械处于生产中的核心地位，若发生故障会造成较严重的经济损失。在旋转机械故障诊断技术现场应用中，故障发生的类型、时间、位置未知，传感器安装不便等原因，导致现场有效的故障预警诊断困难的问题，项目以典型转子系统为研究对象，主要工作和结论如下：. 1、构建与轴系结构尺寸和运行参数相符的动力学有限元实体模型，掌握了多支承转子轴系动力学建模原则与基本方法；以四跨五支承串联轴系为实验对象，研究了一种多转子串轴试重虚拟动平衡方法；分析了多转子串联轴系的扭转振动特性；基于实际工况开展了立式转子动力学建模与振动响应特性分析；开展低速重载滚筒动力学特性与不平衡响应分析，揭示出滚筒转子运行状态与支承振动的内在联系；针对多跨转子轴系现场动平衡需要多次启停机配重且存在较大失败风险的问题，提出基于动力学有限元模型分析的无试重整机动平衡法；开展际工况下转子典型故障振动特性与油温加热装置设计；分析了基于气体动理学模型的谐振管内声场特性，为结构优化提供了重要依据。. 2、开展典型故障转子系统振动特征提取研究，提出了基于小波包能量熵的特征提取方法，结果表明该方法能够较直观地揭示不同状态振动信号的时频特性；开展基于典型核主元分析特征提取方法研究，针对KPCA性能受核函数及其参数影响很大，而最优参数难以选取的问题，展开旋转机械故障诊断中核参数优化选取研究，旋转机械故障特征实例验证了该方法的有效性。. 3、以异步电机与圆柱滚子轴承为对象，提出基于小波包近似熵与LMS加权特征融合的故障诊断法；以航空发动机为研究对象，提出一种基于多种聚类诊断算法融合诊断的方法；开展了某型特种车辆变速箱圆柱滚子轴承故障诊断；提出了基于基座的多传感核主元分析的故障诊断方法。