Research On Sensitive Feature Extraction Method Of Rotor Fault Signal

Fault diagnosis is an important means to ensure the safe operation of mechanical equipment.With the improvement of technology level,the complexity of equipment is also getting higher and higher.To perform effective and accurate troubleshooting,you need to extract a wealth of equipment failure information.Dimensional reduction is a key step in feature extraction.It can extract the essential information of faults and reduce the pressure for subsequent pattern recognition.The vibration information collected during the operation of the mechanical equipment is both linear and non-linear.Therefore,how to effectively mine the data that can truly reflect the fault information,eliminate the irrelevant components in the vibration information,and better fault diagnosis and identification are very important.The focus of this work is on the feature extraction of the rotor vibration signal.The main contents are as follows:(1)Aiming at the problem that the accuracy of rotating machinery fault identification is low,ensemble empirical model decomposition(EEMD)is combined with energy moment and neighborhood rough sets(NRS)to propose a fault data set classification method for rotor system.The method adaptively decomposes the collected vibration signals by the EEMD method,and quantizes the decomposed Intrinsic Mode Function(IMF)with energy moments.The energy moment is used as the condition attribute describing the fault state to construct the fault identification decision table.Then the attribute reduction is performed on the decision table by using the neighborhood rough set to eliminate the redundant attribute.Finally,the reduced sensitivity subset is input.Pattern recognition is performed in the decision tree C4.5(DT4.5)algorithm.The effectiveness of the proposed method is verified by the fault feature set of a typical rotor test bench.(2)Aiming at the problem that the rotor vibration signal is non-stationary and the weak fault features are difficult to extract,a fault feature extraction method based on the combination of ensemble empirical model decomposition(EEMD)singular value entropy and manifold learning algorithm is proposed.Firstly,the original vibration signal is decomposed by EEMD,and some Intrinsic Mode Function(IMF)components are obtained.According to the kurtosis-European distance evaluation index,the sensitive components with rich fault information are selected to form the initial eigenvector and the singular value entropy is obtained.Using the improved Laplacian Eigenmaps algorithm-Nearby Probability Distance Laplacian Eigenmap(NPDLE),the feature matrix composed of singular value entropy is reduced,and the fault information can be extracted.A low-dimensional feature subset that reduces the difficulty of classification.Finally,the obtained low-dimensional feature subsets are input into the KNN for pattern recognition,and the effectiveness of the improved method is verified on the double-span rotor test bench.The results show that the combination of IMF singular value entropy and NPDLE can effectively extract rotor fault features,and the accuracy of fault identification is also improved.(3)In order to stably extract rotor fault characteristics,a fault diagnosis method based on complementary set empirical mode decomposition,multi-scale permutation entropy and GK fuzzy clustering is proposed.Firstly,the fault signal is processed by the Complementary Ensemble Empirical Mode Decomposition(CEEMD).According to the correlation coefficient principle,the modal component with the largest correlation coefficient is selected as the analysis object.Then the multi-scale array entropy is used to quantify the fault feature of the modal component as the feature vector.Finally,The low-dimensional feature set after PCA dimensionality reduction is input into the GK fuzzy clustering algorithm for fault identification.The proposed method is applied to the fault feature set of a typical rotor test bench.The classification effect is tested by classification coefficient and partition entropy,and compared with the empirical mode decomposition multi-scale entropy combined with GK fuzzy clustering.The results show that the proposed method can effectively extract fault characteristics.