Research On Fault Diagnosis Method For Reciprocating Compressor Based On LMD And MFE

Reciprocating compressor is the core equipment in the field of petroleum and chemical industry, once the equipment is faulty, which will not only to the enterprise to great economic loss, also will bring serious damage to the personal safety of the staff. Therefore, fault diagnosis technology of reciprocating compressors become research focus. Because of the nonlinear and non-stationary characteristics of the vibration signal of the reciprocating compressor, the traditional time-frequency analysis method which represented by Fourier transform has some limitations. Local mean decomposition(LMD) is an adaptive decomposition method, which has a unique advantage in the non-stationary signal processing. In this paper, a new method of fault diagnosis for reciprocating compressors is presented by combining the LMD method with the multiscale fuzzy entropy, and carried out a series of research work of the application of the proposed method in the fault diagnosis of reciprocating compressor.First, through access to a large number of reciprocating compressor fault diagnosis related literature, on reciprocating compressor structure, working principle and common failure mechanism are summarized, and expounds the research status of the development of fault diagnosis technology of reciprocating compressors and common fault diagnosis methods.Secondly, on the basis of in-depth study of the theory of LMD method, aiming at the original LMD method sliding average number of iterations is too high and cubic spline LMD method appeared “over envelope” and “under envelope” phenomenon resulting in decomposition of distortion problem, this paper proposes an improved LMD method. In this method, the Hermite interpolation is used to construct the envelope, and the fitting accuracy of the envelope is improved by increasing the value of the extreme symmetry point. The experimental results show that the improved LMD method can accurately extract the signal characteristics, and it is an effective method for the analysis of non-stationary signals.Again, this paper introduces the fault feature recognition method based on multiscale fuzzy entropy. Multiscale fuzzy entropy can measure the complexity of the signal at different scales, and compared with the fuzzy entropy of a single scale, the extracted fault feature information can reflect the fault essence more fully. The multiscale entropy and multiscale fuzzy entropy are compared by using the simulation signals, the results show that the scale has good relative consistency of fuzzy entropy, the entropy curve is more stable, parameter selection is free. Experimental verification of the performance of the multiscale fuzzy entropy in the feature extraction of the bearing fault of reciprocating compressor.Finally, a fault diagnosis method for reciprocating compressor based on improved LMD and multiscale fuzzy entropy(MFE) is presented for the non-stationary and nonlinear characteristics of the vibration signals of reciprocating compressors. In this method, the LMD method is used to decompose the vibration signal into several PF components, and the multi scale fuzzy entropy is calculated to form the feature vector, and the classification and recognition are carried out by the support vector machine. In the process of extracting the characteristics of the state, the use of the euclidean distance to the feature vector is optimized, so that the status of the distinction is more obvious. By using the proposed method, the fault data of bearing and valve of reciprocating compressor were analyzed. The results show that the method can effectively extract fault features and realize the diagnosis of different fault types.