Research On Kinematic Accuracy Reliability Of Industrial Robots Based On Information Fusion

Industrial robot is an important equipment in modern manufacturing industry.Its kinematic accuracy is an important index to describe the performance of industrial robot.The uncertain internal and external excitation has an important influence on the running accuracy,which makes evaluating the motion accuracy of industrial robot with high precision and high efficiency a key scientific problem and difficulty.Combined with the information fusion technology,this paper fuses the system’s multi-model,multi-level,multi-type and multi-moment information to analyze the kinematic accuracy and reliability of industrial robots efficiently and accurately.The main work of this paper includes three parts: the accuracy and reliability research of the joint reducer of industrial robot,the establishment of the multi-fidelity agent model based on deep transfer learning and the research of the reliability analysis method of the kinematic accuracy of industrial robot based on information fusion.The specific work is as follows:(1)Evaluation of accuracy reliability of industrial robot’s joint reducer.Joint reducer is one of the core parts of industrial robot.Its accuracy and reliability have an important effect on the kinematic accuracy of industrial robot.This paper takes RV(Rotate Vector)reducer as the object to analyze the influence of manufacturing error,assembly error,backlash and elastic deformation on its transmission performance.By quantifying these uncertain qualitative factors comprehensively,the reliability model of transmission precision of RV reducer is established,and the reliability of its transmission precision is evaluated.(2)Establishment of kinematic accuracy analysis model for industrial robot.Industrial robot is a complex system with rigid and flexible coupling,whose kinematic and dynamic model is difficult to solve theoretically.In order to improve the analysis accuracy and efficiency,this paper studies the kinematic accuracy analysis method of industrial robot based on the multi-fidelity model technology.Based on the deep transfer learning technology,kinematics model established based on D-H method and the rigidflexible coupling model established based on virtual prototyping technology are fused to establish the motion precision analysis model of industrial robot.The proposed modeling method can solve the problem of input dimension mismatch of various fidelity models in the process of uncertainty analysis,and the established model can analyze the kinematic accuracy of industrial robots with high efficiency and precision.(3)Evaluation of the kinematic accuracy reliability of industrial robots.Industrial robot is a complex series system whose kinematic accuracy reliability is determined by the performance and interaction of each subsystem.In this paper,an efficient reliability evaluation method for the rigid-flexible coupling system of industrial robots is studied,and an information fusion framework based on multi-fidelity Monte Carlo estimation is proposed.Based on the framework,the performance information of components,the information of multi-fidelity model and the information of similar equipment are fused to evaluate the reliability of the motion accuracy of industrial robots.(4)The establishment of a reliability evaluation model for kinematic accuracy of industrial robots considering joint wear.Firstly,the wear mechanism of the key transmission parts of the joint reducer is analyzed,and the wear of the key transmission parts is modeled and the uncertainty of the wear is quantified.Secondly,the influence of wear on the accuracy of the joint reducer is studied,and the accuracy reliability the joint reducer considering wear is evaluated.Then,the reliability information of multi-time system is fused to evaluate the reliability of industrial robot kinematic accuracy based on Bayesian information updating method.