Research On Recognition Of Friction State Of Rotary Pair Based On Motor Current Characteristic Analysis

Rotary joints are widely present in various types of mechanical equipment.Once they fail,they will affect the normal operation of the equipment,which may cause hidden safety hazards and even cause safety accidents.For the identification of the friction state of the rotary pair,you can evaluate the current lubrication status of the rotary pair and estimate the time from the current to the dry friction,so as to take the necessary measures in time to avoid the adverse consequences caused by the dry friction.In view of the problems of many equipments,high cost,difficulty of practical application on the spot or low accuracy,the commonly used friction state monitoring methods,a method for identifying the friction state of the rotary pair based on the motor current characteristic analysis method is proposed,this method can realize long-distance monitoring and has strong anti-interference ability.The main work includes:(1)Based on the analysis of the relationship between the friction torque of the rotating pair and the stator current of the motor,an analysis model of the stator current of the motor containing the friction characteristics of the rotating pair was established,and the influence of different factors on the model was studied.According to the frequency range of the frictional characteristics of the rotary pair in the motor stator current signal,the variational modal decomposition method is used to extract the frictional characteristic signal,and a fast Fourier transform extreme point search is proposed to determine the variational modal decomposition parameters The self-adaptive parameter selection method is used to extract friction features in the time-frequency domain.(2)Calculate the time-domain and frequency-domain feature quantities of the original signal as the auxiliary recognition feature value of the classifier.In order to reduce the feature redundancy caused by the similarity between time-domain,frequency-domain and time-frequency domain features,principal component analysis is used to reduce the dimension of the extracted multi-dimensional features to reduce the feature dimension and improve the recognition accuracy of the classifier.According to the change trend of the first principal component feature after dimensionality reduction,the dimensionality reduction sample interval category is divided and used as the training sample of the classifier.(3)By analyzing the characteristics of the friction state of the rotating pair and the change in the number of signal samples at different friction stages,a support vector machine suitable for small sample learning is selected for friction state recognition,and use the integrated learning method to train the support vector machine,construct a joint classifier,and the penalty factors and kernel parameters of the support vector machine are optimized by the quantum particle swarm optimization algorithm.A joint classifier is used to identify the friction state of the sample data and predict the time from the current friction state of the rotary pair to dry friction.(4)By collecting the working current signal of the rotary pair drive motor in the friction process,analyzing the frequency domain and time-frequency domain characteristics of the experimental signal,and establishing the friction characteristic library of the working current,used to train the multi-class support vector machine classifier and realize the friction state recognition on the classifier.The experimental research on the friction state recognition of the rotary pair based on the analysis of the working current is completed.The research results show that the friction generated during the rotation of the rotary pair causes the amplitude and phase modulation of the motor current signal,and the amplitude change trend is the same as that of the first principal component after dimensionality reduction.By extracting the friction characteristics of the rotating pair,the integrated quantum particle swarm optimization support vector machine joint recognition can accurately assess the friction state of the collected signal,and can predict the current friction state of the rotating pair to the dry friction stage.