Fault Diagnosis Study And Application On Key Parts Of Reciprocating Piston Compressors

Machinery monitoring and fault diagnosis is a comprehensive technology, which essentially is pattern recognition of machine operating conditions. The key issue is feature extraction and classification. Research on fault diagnosis for reciprocating compressors is of particular interest due to the wide application of reciprocating compressors in industry manufactures. This dissertation is based on the project of "Research on Local Wave Method and its Engineering Application" (supported by Chinese National Nature Science foundation, Project No.50475155) and surface vibration signals from reciprocating compressors is chosen as the subject. The fault feature of the vibration signals obtained from the compressor surface is successfully extracted by combining the methods often used for non-stationary signal studies such as local wave method and fractional Fourier transform. An adaptive pattern classification method is also introduced and modified for accurate fault classification and quantification. The main work of this dissertation is as follows:1. Based on the analysis of the structure of reciprocating compressors, the response to surface vibration in relation to the major vibration source and the transmission path is summarized. With respect to the commonly occurred problems in the vibration test practice, the significance of signal reproducibility is elaborated and it is pointed out that signal reproducibility could be one of the most important characteristics in fault diagnosis and may have great impact on the reciprocating compressor fault diagnosis studies. The analysis of signals from the compressor valves shows that the spectra and energy of signals from the same part varied significantly under different test parameters or sampling positions. Therefore, it is concluded that only the signals obtained under the same test conditions can be used for the compressor fault diagnosis.2. The decomposition characteristics of Local Wave Method are discussed in detail, especially the role of sampling frequency. The components of decomposition error are analyzed and the relation between the error and sampling frequency is summarized. If possible, over sampling is recommended whenever decomposition of Local Wave Method is used. By comparing the signal decomposition ability of different methods, it is noted that none of the currently available decomposition methods are effective in minimizing decomposition errors. Therefore, a new method based on over sampling interpolation is proposed for decomposition of Local Wave Method and its effectiveness is confirmed by examples.3. Local Wave AR Mode is introduced to reciprocating compressor fault diagnosis and it is proved to be a useful method in the fault feature extraction. Since serious noise disturbance...