Research On Gearbox Fault Diagnosis Method Based On Dictionary Learning

Gearbox is an indispensable power transmission component in mechanical transmission,and is widely used in various types of transmission equipment.The operating state of the gearbox often reflects the operating state of the mechanical transmission equipment,and once the gearbox fails,the mechanical transmission equipment is often paralyzed.The vibration signal is the carrier of the gearbox fault characteristic information.Fault diagnosis analysis is of great significance.This paper takes gearbox vibration signals as the research object,and proposes a dictionary learning method to extract gearbox fault features,and uses support vector machine(SVM)to classify and identify gearbox faults.By analyzing the failure type and local fault vibration characteristics of the gearbox,a mathematical model is given.Gears and bearings are the most vulnerable parts in gearboxes.When the gears are faulty,the vibration signals that characterize the operating state will often produce amplitude modulation and frequency modulation.When the rolling bearings are faulty,the vibration signals that characterize the operating state will often appear transient.Impact components,which generally decay rapidly and have periodicity.Aiming at the characteristics of non-linearity and non-stationarity of fault signals,a dictionary learning algorithm is used to extract the characteristics of fault vibration signals.Firstly,the basic concepts of dictionary learning algorithm are introduced.Aiming at the problems that the traditional K-SVD algorithm is susceptible to noise interference and the dictionary's atomic coherence is insufficient to represent the internal structure of the signal during the construction of the dictionary,a set of empirical mode decomposition is proposed(EEMD)combined with low-coherence K-SVD for fault feature extraction.Taking the extracted fault features as the input of the fault diagnosis model,the SVM gearbox fault diagnosis model is established,and the selection of the SVM classifier model parameters has a great impact on the classification accuracy of the model to a certain extent,and there is currently no uniform selection standard.In this paper,a gearbox fault diagnosis model of genetic algorithm and SVM is proposed.Finally,the gearbox vibration signal is taken as the research object.By comparing and analyzing with the fixed base dictionary and the traditional K-SVD algorithm,the experimental results show that the combined empirical mode decomposition and lowcoherence K-SVD are used to extract fault features.A dictionary matching the characteristic components of the signal can be accurately constructed,and the signal reconstruction performance is improved,so the fault characteristics of the fault vibration signal can be accurately extracted.Optimizing SVM parameters through genetic algorithms can avoid human interference and adaptively obtain optimal parameter values,thereby improving the accuracy of fault diagnosis.