机械故障诊断内积变换原理与验证

摘要

Signal processing methods commonly used in engineering practice, such as Fourier transform, short time Fourier transform, wavelet transform, second generation wavelet transform, and multiwavelet transform, lay the foundation for operating monitoring and fault diagnosis of key equipments. To profoundly understand their common problems, an intensive and systematic research on the physical nature and key techniques of these signal processing methods is conducted. It is found in the research that their transform principle is inner product transform. The purpose of the principle is to search the components contained in the signal, which are the most relevant or similar to the "basis function". The key point lies in the construction and selection of basis functions matching fault feature waveforms, so as to achieve condition monitoring and fault diagnosis. The principle is validated by simulations and engineering cases, which sufficiently proves its accuracy and reliability. Furthermore, some conclusions from the experimental results supplement and improve the inner product principle.%工程实际常用的快速傅里叶变换、短时傅里叶变换、小波变换、第二代小波变换和多小波变换等先进信号处理方法,为关键设备运行监测与故障诊断奠定了基础.为深刻认识其共性问题,对机械故障诊断信号处理方法物理本质与关键技术等基础问题进行研究.指出它们的变换原理都是内积变换,探求信号中包含与“基函数”最相似或最相关的分量,关键在于构造和选择动态信号中与故障特征波形相匹配的基函数,实现科学、正确的状态监测与故障诊断.通过仿真试验和工程案例,对机械故障诊断的内积变换原理进行验证,证明该原理的正确性和可靠性.同时,根据试验现象得出若干基函数性质对内积变换的影响规律,补充和完善了机械故障诊断内积变换原理.