Research On Multi-fault Classification And Diagnosis Method Of Planetary Gear Transmission System

The service status of planetary gearbox has a vital influence on the safety of transmission equipment.In order to meet the requirements of long-term variable service conditions and complete the important requirements of power transmission and adjustment of equipment,the planetary gearbox presents the characteristics of complex transmission structure and strong load tolerance.However,due to the enhanced interaction of functional components,the planetary gearbox signals obtained by the system have obvious problems such as non-linearity,instability,mode mixing and signal coupling,etc.,which improve the difficulty of fault pattern recognition,feature extraction and prediction of planetary gearbox.Therefore,the research on multi-fault classification and diagnosis method of planetary gearbox transmission system has high application value for realizing planetary gearbox service state detection and intelligent control,achieving the purpose of fault pattern recognition,feature extraction and prediction,reducing operation and maintenance costs and avoiding the occurrence of safety accidents.The planetary gear transmission system is the research object in this thesis.Aiming at the problems of vibration signal aliasing and low signal-to-noise ratio in the process of planetary gearbox fault feature pattern recognition,a fuzzy hidden Markov model(FHMM)multi fault pattern recognition algorithm is proposed to solve the problem of low accuracy of planetary gearbox fault feature pattern recognition;In order to solve the problem of fault feature extraction of planetary gearbox under strong noise,an Intensifying variational mode decomposition(IVMD)fault feature extraction method based on instantaneous frequency and kurtosis is established.This method can overcome the influence of prior knowledge on fault feature extraction and meet the requirements of adaptive fault feature extraction,It provides important theoretical and technical support for early fault feature extraction of planetary gearbox;In order to meet the requirement of local fault trend prediction of planetary gearbox,a multi model hybrid prediction method based on enhanced self-organizing feature map(ESOFM)and support vector regression(SVR)was constructed.Based on the time-domain vibration signal,this method solves the problem that it is difficult to predict and maintain the local faults of the planetary gearbox,and provides the key technical support for the smooth operation of the planetary gearbox.The main contents of this thesis include the following aspects:(1)This paper analyzes the modeling method of planetary gearbox fault diagnosis,summarizes the traditional processing methods and intelligent optimization algorithm in the field of fault diagnosis,and lists the signal analysis indexes and application conditions in detail.At the same time,according to the three fault diagnosis requirements of fault pattern recognition,feature extraction and trend prediction,the corresponding simulated fault signals are designed respectively,and the simulation test and vibration signal collection of gear tooth fracture fault are completed by using the planetary gearbox simulation fault test equipment,which provide data support for the subsequent chapters to carry out analysis of fault pattern recognition,fault feature extraction and fault trend prediction verification.(2)In order to improve the accuracy of local fault pattern recognition of planetary gearbox,a fuzzy hidden Markov model(FHMM)multi fault pattern recognition optimization algorithm is proposed,which is based on fuzzy C-means(FCM)clustering analysis and hidden Markov model(HMM)stochastic process.In this algorithm,FCM is used to classify the fault signals,and the mode state space A is established and optimized by HMM algorithm.FHMM multi fault pattern recognition optimization algorithm improves the independence and recognition accuracy of fault feature mode,and realizes the optimal recognition of local multiple fault modes of planetary gearbox.(3)Aiming at the difficulty of feature extraction of gear vibration signal in planetary gearbox transmission system,the IVMD method is designed by introducing the instantaneous frequency index of modal component.This method realizes the adaptive and accurate selection of modal component parameter K in variational modal decomposition,and overcomes the defect of prior knowledge in setting parameters of variational modal decomposition.According to the given modal component parameter K,taking the kurtosis value of modal signal as the standard,the balance constraint parameters are optimized α,which improve the decomposition accuracy of modal components.Finally the FHMM multi fault pattern recognition optimization algorithm completed the effective identification of broken tooth fault features.(4)Aiming at the problem of local fault trend prediction of planetary gearbox,a multi model hybrid prediction method based on ESOFM-SVR is proposed.Firstly,membership function u is introduced to optimize the inertia weight ω of SOFM output key points,and ESOFM method is used to extract the key points of modal components.Then,the SVR model is used to establish a 20:1 data prediction ratio to predict the modal component and the original signal respectively.Finally,the prediction value is compared with the original data to verify that ESOFM-SVR multi model hybrid prediction method has higher prediction accuracy.