Fault Diagnosis And Application For Reciprocating Compressor Based On Time-Frequency Distribution Entropy

As for the complexity of reciprocating compressor and many vibration sources on it, fault diagnosis for reciprocating compressor is very difficult. The vibration signal on cylinder is very complex, which is full of nonlinear and nonstationary characters. In this paper, a new method by using time-frequency distribution entropy is deeply investigated according to the former achievement, which is directly to real project.Local-wave method is imported especially for extracting feature in the vibration signal. Because of the complex of the real signal, it is not so clear about the reciprocating compressor feature on the time-frequency distribution based on local-wave method. Hence, image recognition is imported for fault diagnosis based on grey chart. The method of SVD about image recognition is used to extract secondary feature. Singular value spectrum and eigenvector about the time-frequency distribution is obtained. It achieves the aim of dimension reduction. And it made state component more clearly and more convenient for fault diagnosis.In order to quantify the vibration signal, the theory of entropy is introduced. A new method of fault diagnosis for reciprocating compressor is put forward based on time-frequency distribution entropy. It provided the foundation for the intelligent fault diagnosis and the development of state monitor.According to the contract with Beijing Yanshan Petrochemical Co., Ltd. state monitor is put on the reciprocating compressor(C-1150). The theory of time-frequency distribution entropy is put into practice. The fault of packing leakage of C-1150 is predicted successfully. It proves the feasibility and veracity of theory, and brings factory notable benefit.State monitor system is development based on LabVIEW in order to monitor the compressor more accuracy. Many of modules such as the LabVIEW drive of the data acquisition card, DAQ, data transmission, the module of data analysis, subsystem of database are developed. Now the system has been completed, it provides technical support for fault diagnosis and state monitor about reciprocating compressor.