Research On Fault Feature Extraction Of One-way Valve Of High Pressure Diaphragm Pump

Reciprocating high pressure diaphragm pumps are devices that provide power for transportation during pipeline transportation.The check valve is one of the core components of the reciprocating high pressure diaphragm pump.The one-way valve exerts a pressure on the slurry through direct contact with the slurry to function as a flow blocking and a flow.Because the diaphragm pump check valve is a typical reciprocating machine,it has the characteristics of multi-excitation source,response mutual coupling and other working conditions,and its complex internal structure leads to incomplete analysis of the operating mechanism.In the event of a failure of the check valve,the slurry delivery efficiency will also be seriously affected,and even the diaphragm pump as a whole will be damaged.Therefore,the fault feature extraction of the high-pressure diaphragm pump check valve is a research direction with practical application value.In this paper,the vibration signal of the high-pressure diaphragm pump check valve is taken as the research object,and the problem of screening the fault sensitive feature of the check valve under complex conditions is studied,which provides a criterion for the screening and extraction of the fault sensitive feature of the check valve.This paper mainly made the following research:(1)Aiming at the problem that the initial fault characteristics of the one-way valve is not obvious,a fault feature extraction method based on wavelet approximate entropy and K-means clustering is proposed.From the perspective of fault mechanism,the method uses wavelet theory to decompose the original signal into several components,selects the effective component,and obtains a relatively pure signal representing the abnormal high-frequency vibration of the one-way valve,and then takes the approximate entropy of the different frequency band components as the characteristic parameters.Features that are not sufficiently noticeable in the original signal are highlighted in different frequency band components.Experimental results show that this method makes the feature clustering effect significantly improved.(2)In order to avoid a certain degree of one-sidedness in single feature extraction,a fault diagnosis method for high-pressure diaphragm pump check valve based on multi-time domain feature and support vector machine is proposed.Based on the research in(1),the method further enriches the types of characteristic parameters and eliminates the decomposition process of signals.Firstly,the vibration signals of the normal and fault conditions of the check valve are analyzed,and the characteristics of the fault time domain waveform are summarized.For the impact component of the signal,three time-domain indicators of the check valve vibration signal are selected:kurtosis,approximate entropy and Margin factor;the three indicators are entered into the diagnostic model as one-way valve eigenvalues.The test results show that the selected three characteristics can better reflect the correlation characteristics of the signal impact components,and the recognition accuracy meets the engineering requirements.(3)To illustrate the evaluation criteria of feature selection,and further screen out the fault-sensitive features,this chapter proposes the K-means symbolization and rough set theory for the fault feature extraction method based on the research of(1)and(2)..The method fully considers the actual working conditions and engineering requirements of the check valve,and achieves the purpose of fuzzy fault boundary by symbolizing the processing of various characteristic parameters.The attribute reduction is reduced in the case of ensuring the correct rate of state recognition,and the feature dimension is reduced.The test results show that the method can effectively eliminate irrelevant features,improve the speed of fault diagnosis,and can be used as an evaluation index of feature validity and sensitivity.Based on the exploration of the fault feature extraction of the check valve,the fault feature extraction and sensitive feature selection method of the high pressure diaphragm pump check valve are proposed.Through the multi-band single feature,multi-features and feature fuzzy screening,the effects of “characteristic zooming”,“multi-features” and “screening sensitive fault features” are realized respectively,which have certain practical application value.