Fault Detection Of Bearings In Joints Of Robot Arms Based On Wavelet Packet And Radial Basis Neural Network

Rotating machinery is used in many fields,rolling bearing is a necessary part to ensure the normal operation of rotating machinery.It has a direct impact on the accuracy and reliability of rotating machinery.Therefore,the rolling bearing fault is one of the most common causes of rotating machinery fault.Bearings have a direct impact on the accuracy and reliability of rotating machinery,bearing in operation often appear crack defects,metal peeling,corrosion and other damage,so that the equipment can not work normally,so it is necessary to study the fault diagnosis method of rolling bearing.At present,many domestic and foreign scholars put forward a large number of research methods for fault diagnosis of high-speed full-cycle rotating bearings.However,there is little research on fault diagnosis of intra-articular bearing which can not rotate continuously for a whole cycle.In this paper,taking the intra-articular rolling bearing in a simple link manipulator as an example,the dynamic simulation of low-speed non-full-cycle rotating bearing is studied.Feature extraction of vibration signals of fault bearing with crack defects on inner ring,outer ring and rolling element,and the classification and identification of rolling bearing faults.The main contents of this paper are as follows:(1)The motion simulation of the robot arm is carried out.The shoulder inner ring,outer ring,rolling element fault bearing and joint inner ring,outer ring and rolling element fault bearing are simulated respectively.The acceleration vibration signals of different fault bearings in different parts are given.The fault vibration signal of shoulder without space motion is compared with that of bearing in joint in space motion.Preparation for bearing dynamic modeling.(2)The dynamic model of bearing with joint internal fault is established.Using the angle of the inner ring,outer ring and rolling body defects relative to the coordinate system,the deformation of each fault bearing rolling body is calculated,and the motion differential equations of different fault bearings are obtained.After solving the equations,the vibration acceleration simulation signals of three kinds of fault bearings are obtained.The time-frequency analysis of the fault signals reveals the difference from the whole rotating bearing,which lays a foundation for extracting the characteristic parameters in the frequency domain.(3)The fault diagnosis method of internal bearing of mechanical arm joint based on wavelet packet analysis and radial basis function neural network is studied.GT filter bank is used to extract the effective outer ring,inner ring and rolling body fault impact signals,and envelope analysis is carried out on the fault vibration signals with the most obvious impact signals in the filter channel.Wavelet packet analysis is used to decompose and de-noise the bearing fault signal in the joint.Demodulating the 8-order energy band of the reconstructed fault vibration signal,the frequency band energy values of vibration signals are normalized and arranged according to scale as the bearing fault feature vector,which is the input layer of radial basis function neural network.A large number of fault data are trained by RBF neural network to establish fault diagnosis model and test and compare the performance of the model.The faults of bearing can be identified accurately,and the fault diagnosis of bearing in joint of manipulator is realized.