全矢经验模态分析理论及其在转子耦合故障诊断的研究

In order to overcome the limitations due to incomplete and imprecise information, two mutually orthogonal sensors are usually installed on the same section of a rotor in the rotating machine field. Although the vibration information from two mutually orthogonal sensors belongs to homologous, two signals are not identical. The drawback of using individual source information alone to analyze signal is incomplete and imperfect, which cause some fuzzy results because of ignoring information of the relationship between the two sensors. To improve the accuracy from non-stable vibration signal for the coupling faults diagnosis in the rotor system, the full vector spectrum technology is introduced into the empirical mode decomposition analysis method. By merging vibration signal from two mutually orthogonal sensors, an advanced signal processing technique is presented. The new method can merge non-linear vibration signal from two mutually orthogonal sensors in rotating machinery vibration signal feature extraction. The features extracted by using the full vector empirical mode decomposition analysis method and then the features are constructed in the artificial intelligence. It makes the new method more easily to apply to condition monitoring and fault diagnosis of mechanical equipment. Fault diagnosis methods are changed from a single vibration direction to the section of a rotor, by using the full vector empirical mode decomposition analysis method. It has the advantage of accuracy, efficiency and comprehensiveness in detecting coupling faults vibration signal of rotor system. This may pave a new way for coupling faults diagnosis of rotor system.

在旋转机械转子系统的运行状态监测中，为了全面地监测其运行状态，通常在转子的同一截面安装一套相互垂直的传感器以拾取振动信号。但在信号分析时，常规的振动信号分析方法仅以单一传感器的信息为研究对象，其分析结果存在信号信噪比低和诊断的可信度低等问题。为了从转子系统耦合故障的非稳态信号中更全面、准确地提取出故障征兆，本项目将基于同源信息融合的全矢谱理论引入到经验模态分析方法中，提出基于同源融合信息的全矢经验模态分析新方法，实现转子系统多通道非线性振动信号的数据融合与特征提取等问题。以全矢经验模态分析方法为故障特征提取工具，利用神经网络建立一种基于全矢经验模态分析的智能诊断模型。采用全矢经验模态分析方法进行转子耦合故障的诊断，是从转子单个方向的诊断提升到整个振动平面的诊断，可以更加准确、有效地提取和分离耦合故障的各个特征频率，为转子耦合故障的诊断提供一种新思路。

转子系统作为旋转机械核心部件，一旦出现故障，将影响机械设备的使用寿命及其安全生产。如何及时、有效地发现并预测转子系统故障，是设备状态预测与故障诊断的主要研究课题。随着设备系统的集中化、高速化，转子系统常常出现具有典型非线性特性的耦合故障，转子耦合故障的振动特性较单一故障更加复杂，而且相互影响，给故障诊断带来一定困难。. 本项目基于多传感器数据层融合技术的全矢谱理论和经验模态分析方法，开展了转子系统耦合故障诊断方法研究。通过数值仿真和实验验证，分析了转子系统局部碰摩振动特性与传感器观测点的相关性，得出不同观测点对同一碰摩部位的诊断结果存在较大差异性的结论，提出了基于全矢谱理论的转子碰摩故障诊断方法；研究了不同工况下转子碰摩耦合故障中振动信号响应特点与故障原因之间的相互映射关系，提出了全矢转子碰摩耦合故障诊断方法，构建了基于全矢谱理论的主振矢、副振矢、相位、夹角四个不同参数的全矢谱碰摩耦合故障诊断体系；提出了全矢小波包能量熵的转子碰摩耦合故障诊断方法，以不同频段的能量比为特征参数，反映了碰摩耦合故障在各分频段上能量的集中和分散程度特性，实现了基于神经网络的耦合故障模式识别；构建了基于同源信息融合的全矢经验模态分析方法的数学模型及其计算方法，提出了转子耦合故障的全矢经验模态分析方法，验证了全矢经验模态分析方法在转子耦合故障诊断的可行性。. 本项目的研究揭示了转子系统发生耦合故障时同一截面两个相互垂直方向的振动响应规律，解决了转子同一截面不同方向振动分析差异性问题，为信息融合技术在转子系统耦合故障诊断中的研究提供科学依据。