Fault Diagnosis Method Research For Robot Harmonic Reducers Based On Drive Motor Condition

Robot reducer is the core component of the robot,the reliability and safety of the robot system are closely related to the deterioration condition of the robot reducer.The harmonic reducer is a common reducer used in robot joint.Failure of the robot reducer can cause malfunction and even cause an accident.The closed structure of harmonic reducers has a great influence on the condition parameter extraction of the robot reducer.Therefore,It is of great significance to carry out fault diagnosis method of the robot harmonic reducer.At present,the field of fault diagnosis of robot reducer mainly lies in mechanism of robot reducer deterioration and failure.The research direction of fault diagnosis of the robot reducer still needs further development.The main research contents of the dissertation are as follows.1)This paper introduces the development status of mechanical equipment fault diagnosis and harmonic reducer,and summarizes the basic working principle of the robot harmonic reducer and the main forms of degradation failure.2)For the research direction of the axial displacement fault diagnosis of the input shaft and the overrun transmission error fault diagnosis of the harmonic reducer based on the motor condition,the experimental data acquisition platform was built,and the data acquired by the data acquisition system is used as the samples of the two these research directions.3)For the axial displacement failure of the input shaft of the robot harmonic reducer,the Simulink electromechanical system model of the harmonic reducer is established based on the exponential LuGre friction.The axial displacement failure of the input shaft is mathematically modeled in this model,the simulation results show that the axial displacement failure has influence on the condition of the drive motor.Aiming at the axial displacement failure of the input shaft of the robot harmonic reducer,this paper proposesd the fault diagnosis methods based on deep learning network.The diagnosis methods include two deep learning network diagnostic models based on the wavelet packet energy entropy value and EMD decomposition characteristics of the motor current,and these two models passes the experimental collection sample training and verification.4)For the fault diagnosis of the overrun transmission error of the robot harmonic reducer,this paper proposes a method for evaluating the transmission error level of the harmonic reducer based on the one-dimensional convolution fusion coding and Hidden Markov Chain model.This model extracts the motor condition data and applies the one-dimensional convolution network fusion coding.Finally,coded sample of the motor condition is used to train the HMM to obtain the optimal HMMs model referenced to the K-fold cross validation.After the sample verification test,the harmonic reducer transmission error level estimation model based on the drive motor state can effectively evaluate the transmission error level of the harmonic reducer.