| Hydraulic valve is the control element of hydraulic system.No matter how simple the hydraulic system is,hydraulic control valve is indispensable and can be replaced.However,hydraulic valve often fails due to processing,assembly,working environment and working mechanism.Because the hydraulic valve is in the sealed hydraulic pipeline system,and the hydraulic system is often in bad working environment,complex operation conditions,often in high temperature,strong vibration,big noise,strong interference working conditions,the relationship between fault mode and fault symptom is not clear,and the nonlinearity of the state signal will be further aggravated with the occurrence of the fault,which brings challenges to the fault diagnosis of the hydraulic valve.In this paper,based on the acceleration vibration signal,the wear fault diagnosis of electro-hydraulic directional valve is studied.The specific work is as follows:1.The leakage fault state signal of electro-hydraulic directional valve is easy to be disturbed by strong noise,and often presents non-linear and non-stationary characteristics.The fault characteristic information of different severity is similar and difficult to distinguish,especially for the problems of unclear positioning of fault characteristic information of Electro-hydraulic Directional Valve and large redundant component.This paper presents a fault diagnosis method of hydraulic valve leakage based on IMF and weighted Densely Connected Convolutional Networks(WDense Net).In order to verify the effectiveness of the method,a hydraulic test-bed is designed for research.The research results show that the method can not only accurately and efficiently diagnose the leakage fault in the Electro-hydraulic Directional Valve(the diagnostic accuracy is not less than 98%),but also find that the state signal of before of reversing of the hydraulic valve contains more characteristic information which is conducive to diagnosis.2.Multifractal detrended fluctuation analysis(MF-DFA)can effectively reveal the multifractal characteristics hidden in nonlinear and non-stationary vibration signals,so it provides a new idea for the research of hydraulic diagnosis.However,the traditional MF-DFA method has some problems: the uncertainty of fitting order selection and the discontinuity of segmentation points lead to inaccurate feature parameter extraction.This paper presents a fault diagnosis method of hydraulic valve based on improved MF-DFA and random forest.The improved MF-DFA method can effectively solve the problem that the traditional methods extract the characteristic parameters(multifractal spectrum parameters)of fault is not accurate,and then the random forest classifier is used to diagnose the leakage fault in the electro-hydraulic directional valve.Through the experimental verification,it is found that the diagnosis accuracy of the proposed method for the leakage in the electro-hydraulic directional valve can reach 96.7%.This method has a certain reference significance for the diagnosis of non-stationary and non-linear systems.3.In view of the inconsistent performance of wear fault features of Electro-hydraulic Directional Valve in different degrees and positions,it is difficult to fully characterize the fault features only by a single fault feature,and the diagnosis stability is not high.In this paper,a hydraulic valve fault diagnosis method based on multi-dimensional feature extraction and evidence fusion theory is proposed.Combining the results of the former two methods,a new data set is formed by selecting the forward reversing data which contains the most abundant characteristic information,the multi-feature information is extracted,and the feature is used as the input of the Grey relational analysis method based on entropy weight.The experimental results show that,compared with the single feature diagnosis results,the proposed method has higher diagnostic accuracy(98.7%)in the leakage fault diagnosis of Electro-hydraulic Directional Valve,and the method is suitable for small sample data,which provides a possible basis for online fault detection of hydraulic valves. |
