Investigation Of Fault Feature Extraction Based On Entropy And Early Fault Diagnosis For Gearbox

Gearbox as an important part of mechanical equipment is used to transmit the motion and power,and its health will directly affect the operation safety of the whole mechanical equipment.Because the structure of the gearbox is complex and working environment is harsh,the gearbox is susceptible to fault even causing great economic losses and casualties.In this paper,the gearbox is taken as the research object.On the basis of summarizing the existing fault diagnosis methods of gearbox,dynamic modeling method is applied to supplement the fault data and find the fault characteristics,and then fault feature extraction and early fault diagnosis of gearbox are studied by applying complexity characterization based on entropy algorithm and advanced filtering technology.The main contents are as follows:In order to analyze the fault characteristics of the gearbox theoritically,at the same time,supplement the experimental data of pitting fault type and severities for gearbox,the dynamic simulation model of a gearbox with tooth pitting fault is studied.First,establish the dynamic model of a gearbox(two gears),which is a mass-spring-damper(6 degrees of freedom)dynamic model with both torsional and lateral vibrations considered.Second,gear meshing stiffness is the main internal excitation of gear dynamic system.With the increasing of pitting fault severity,the meshing stiffness will change and consequently dynamic properties of gear systems will also change.In this paper,the time-varying meshing stiffness is used to describe the changes of tooth pitting fault for gearbox.Last,the fault symptoms of different pitting fault severity are obtained.The dynamic simulation model is verified and the early fault characteristics and the vibration features under different fault severities of gearbox are analyzed.Focused on the different complexity of vibration signals under different gear fault types,a feature extraction method based on adaptive permutation entropy for fault types of gear is proposed.The adaptive permutation entropy integrates the adaptive characteristics of local mean decomposition and the high computational efficiency of permutation entropy.It can effectively extract the fault features of gear vibration signals for different fault types.Compared with the traditional multi-scale permutation entropy,the adaptive multi-scale permutation entropy algorithm apply soft threshold division,which can describe the fault information of vibration signals more accurately and improve the diatinguishability between samples effectively.The proposed method is numerically and experimentally demonstrated to be higher accuracy and be able to recognize different fault types of gear.Focused on the disadvantages of the MFE method,a feature extraction method based on generalized composite multi-scale fuzzy entropy(GCMFE)is proposed to recognize different levels of fault severity for gear.Aiming at the shortcomings of coarse-grained procedure in MFE algorithm: the concept of generalized composite multi-scale fuzzy entropy can effectively enhance the stability of fuzzy entropy in large scale and take into account the fault information of low-frequency component and high-frequency component of the vibration signal.First apply the proposed method to extract the fault feature of different levels of fault severity.After obtaining the feature vectors using GCMFE,laplacian score(LS)method is applied to select the features with most important information.Last,extreme learning machine(ELM)classifier is employed for pattern recognition.Simulation and experiment demonstrated that the extracted fault features using GCMFE presented better divisibility than MFE,which can effectively recognize the different levels of fault severity for gear.Focused on the difficulty of extracting fault signatures at early stage due to the weak fault symptoms and low signal to noise ratio,a strategy of combination of ensemble empirical mode decomposition(EEMD)and improved adaptive multi-scale morphology analysis(IAMMA)is proposed.In order to remove the noise,EEMD is firstly proposed to decompose the vibration signal at early stage into a series of intrinsic mode functions(IMFs),select the IMFs with the important information and reconstruct a signal.AMMA is then used to demodulate the reconstructed signal.Focused on the mean value method in AMMA,a characteristic frequency ratio(CFR)is used to propose IAMMA.The larger CFR index is,the larger weighted index of AMMA demodulated results are.The simulated and experimental vibration signals are employed to evaluate the effectiveness of the proposed method.The results demonstrate that the proposed method can effectively extract the weak fault characteristics and complete the early fault diagnosis of gear.