Research On Phm System Of Robot End Position Accuracy Based On Clearance Joint Simplified Joint

As a cutting-edge product that combines artificial intelligence with traditional machinery and electricity,industrial robots are known as the jewel in the crown of the manufacturing industry.Industrial robots are widely used in high-precision processing fields such as welding,laser cutting,grinding and polishing.During longterm operation,there will be clearances in the joints,resulting in the deterioration of the accuracy of the end.At the same time,when the robot is running at high speed,the clearance may cause violent impacts and collisions between the rotating secondary elements,increasing the dynamic stress of the components,resulting in increased elastic deformation of the components,increased wear,and noise and vibration.It is necessary to establish a characterization model between joint clearance and end accuracy degradation,and to carry out research on the state monitoring and health management(PHM)methods of industrial robot end positioning accuracy.Firstly,the positive transmission relationship between the joint clearance and the end position error of the five-joint industrial robot is established.A virtual prototype model of a parametric five-joint industrial robot is established based on ADAMS.Considering the problem of joint wear,it is simplified into a clearance hinge.The clearance effect evaluation index system is established,and the influence of different joint clearances on the accuracy and stability of the end is quantitatively analyzed,which lays the foundation for the subsequent establishment of feature extraction and joint clearance recognition algorithms.Then feature extraction is one of the key steps of the PHM system.The end vibration signal of the established five-joint industrial robot virtual prototype is used as the data source for system monitoring to construct a joint clearance feature data set.According to different joint clearance parameters,the end vibration signal will have different characteristics,and the end acceleration signal is selected as the analysis object;use the ADAMS secondary development function,use the cmd command to automatically modify the joint clearance size and run the simulation script to obtain the end vibration signal.A feature extraction method of continuous wavelet transform based on complex morlet wavelet is proposed.This method can accurately locate the time when the feature frequency appears,adaptively select the center frequency of band-pass filtering,and realize the unification of signal bandpass filtering and Hilbert transform.Using this method,the vibration data is subjected to complex analytic continuous wavelet transformation to obtain the time-frequency diagram of the sample,and a data set of joint clearance characteristics is constructed.Then take the established time-frequency graph data set as the input of the convolutional neural network,and the corresponding clearance parameter as the output to realize the function of the PHM method.Pattern recognition based on smallscale convolutional neural networks solves the problem that the currently commonly used maximum pooling only takes the maximum value of neuron activity in the region,which leads to the loss of a large amount of information and average pooling.The weights of neuron activity values in the region are the same but fuzzy The problem of characteristics.Using the data set of joint clearance features to optimize the structure of the small-scale convolutional neural network,the recognition results verify the effectiveness and accuracy of the model method proposed in this paper.Finally,a PHM system software platform was built based on My SQL and Pyqt5 to realize data management and algorithm integration.The position and size of the joint clearance that causes the end accuracy to decrease can be identified based on the end state.