Design And Implementation Of Gearbox Fault Diagnosis System Based On Vibration Signal

With the continuous improvement of scientific and technological productivity,in the long-term production process,there will be wear,broken teeth,pitting and other faults in the gearbox of production equipment.If it is not detected and repaired in time,it will have a serious impact on production efficiency and production safety.However,as a key load-bearing component,gearbox is in a complex and harsh working environment,there will be problems such as high failure rate,complex fault causes and difficult signal acquisition,which will bring great difficulties to equipment maintenance.Traditional detection methods relying on hardware circuit can no longer accurately diagnose the fault node.Aiming at these problems,a gearbox fault diagnosis system based on vibration signal is designed.The main contents are as follows:The correlation between gearbox fault and vibration signal is studied.Based on this,the gearbox vibration signal detection model is established.The vibration signal characteristics of gearbox under various states such as no fault,broken tooth fault,pitting fault and wear fault are comprehensively analyzed.The overall scheme design of the system is completed,which lays a theoretical foundation for the realization of the system.Aiming at the problem of real-time acquisition of gearbox vibration signal,a multi-channel and high sampling rate vibration signal acquisition and fault diagnosis software is developed,and the modular design of software function is realized,which mainly includes user management,system setting,signal acquisition and storage,real-time waveform display,characteristic parameter display,fault analysis,judgment and synthesis.The coupling between each functional module is small,which improves the portability of the system.Aiming at the problem that the gearbox vibration signal is not stable and the hidden feature information is not fully utilized,the multi-scale feature signal decomposition method is used to analyze the time-frequency of the vibration signal.Two signal processing methods of ensemble empirical mode decomposition(EMD)and empirical wavelet transform(EWT)are analyzed,and the synthetic signal is decomposed and simulated respectively.The results show that EWT has better effect on decomposition speed and feature reflection,and can better reflect the amplitude frequency characteristics of the original vibration signal of the gearbox.Aiming at the problem of gearbox fault classification,a network model training method based on support vector machine is used to train the vibration signal features.Using logistic regression,support vector machine and k-nearest neighbor algorithm to train the eigenvectors of vibration signal data under different fault conditions,and input the test signals into the model for fault diagnosis.Support vector machine algorithm can quickly identify the presence and type of gearbox fault according to the relevant characteristics of vibration signal,with high diagnostic accuracy,It is regarded as the diagnosis algorithm of the system.Moreover,based on the adaptive fault diagnosis algorithm,the adaptive diagnosis of new fault types of gearbox is realized,and the diagnosis ability and adaptability of the system are strengthened.Finally,taking the gearbox as the research object,a set of vibration signal acquisition and analysis experimental platform is built to verify the method.The software is used to collect and store the vibration signal in real time.The system analyzes the characteristic parameters of the signal,and then performs data display and fault alarm,and tests the fault rate analysis.The feature extraction and support vector machine fault classification of gearbox vibration signal based on empirical wavelet transform are verified by experiments,The experimental results show that the gearbox fault diagnosis method based on vibration signal is correct and efficient.