Research Of Keca In Fault Diagnosis Of Acoustic Emission In Planetary Gearbox

As a complex compound gear transmission system,the planetary gearbox has the advantages of small size,large transmission ratio,and high transmission efficiency.It is widely used in many fields such as wind turbines,aviation generators,automobile gearboxes and so on.Due to the complex structure of the planetary gearbox itself,during its operation,factors such as clearance,external force,collision,friction and its dynamic behavior interact with each other,causing it to deviate from the ideal operating state,which can easily lead to failures and even catastrophic accidents.Therefore,the status monitoring and fault diagnosis of the planetary gearbox are of great research significance and application value to ensure the safe and reliable operation of the system,reduce maintenance costs,and avoid major accidents.Acoustic emission(Acoustic emission,AE)technology as a high-sensitivity,wide frequency response dynamic nondestructive detection technology,has been more and more widely used in the field of fault detection.Compared with traditional fault detection technologies(such as vibration analysis and oil sample analysis),AE has obvious advantages in the ultra-low speed operating state of the planetary gearbox and early fault diagnosis.This article takes the planetary gearbox as the research object,combined with the AE detection technology,and carried out related research on the planetary gearbox fault status recognition and health monitoring and other issues.The essence of fault diagnosis is the problem of pattern recognition,and how to extract the characteristic parameters that are sensitive to the running state of the equipment,have good separability and strong regularity from the complex fault signal is a key step.Based on this,this paper introduces the state recognition model of planetary gearbox based on Kernel Entropy Component Analysis(KECA)algorithm,and on this basis,further proposes an improved algorithm(Improved Kernel Entropy Component Analysis,IKECA).Main tasks as follows:1)It is difficult to avoid the noise mixed in the AE signal of the planetary gear box,which will cause interference to its fault diagnosis and condition monitoring.Firstly,the source and composition of noise in the collected AE signal are analyzed.According to the typical nonlinearity,non-stationary and non-Gaussian nature of the AE signal,and the characteristics of noise distribution,the wavelet packet threshold noise reduction algorithm is used to filter out the noise mixed in the AE signal.To improve the signal-to-noise ratio.And according to the characteristics of the actual measured AE signal,the wavelet basis function,decomposition layers,threshold,threshold function and other suitable selection methods were explored,and good results were obtained;2)Aiming at the correlation and redundancy existing in the mixed-domain high-dimensional feature data set extracted from the planetary gearbox AE signal,which will affect the subsequent state recognition performance,the KECA algorithm is introduced to extract the characteristics that can characterize the equipment state Important information and reduce dimensions.On this basis,an improved algorithm is further proposed.The improved algorithm directly looks for the direction that maximizes the secondary Renyi entropy value of the data as the projection direction,and fully mines the low-dimensional sensitive feature parameters embedded in the high-dimensional space,thereby improving the efficiency of fault diagnosis.And state recognition accuracy.Compared with different feature extraction algorithms,the effectiveness and superiority of the improved algorithm are verified through experiments.3)Considering that the data processed by the IKECA algorithm still needs to be input into the classifier to complete the intelligent identification of the final state,this paper studies the nonlinearity,high dimensionality and small samples of the planetary gearbox fault identification.Machine fault recognition algorithm.On the one hand,the classification accuracy is used as an indicator to further verify the effectiveness and superiority of the IKECA algorithm;on the other hand,on the basis of synthesizing the above algorithms,the WPD-IKECA-SVM fault diagnosis model is used for the analysis and processing of the measured AE signal.The results show that the diagnostic framework has higher fault recognition accuracy and diagnostic efficiency.