Fault diagnosis of robot joint based on BP neural network

Abstract

Aiming at the problem of low accuracy of robot joint fault diagnosis, a fault diagnosis method of robot joint based on BP neural network is designed. In this paper, the UR10 robot is taken as the research object, and the end pose data of the robot are collected in real time. By injecting different joint errors and changing the sampling frequency, the joint fault database is collected and established, and the BP neural network is used for training to obtain the robot neural network fault diagnosis model. The fault diagnosis model can output the joint fault of the input end pose data. And we analyzed the influence of different joint angle errors and different training sets on the accuracy of joint fault diagnosis of the robot. The results show that when the sampling frequency is 250 Hz, the simulation result of joint fault diagnosis accuracy with the fault degree of 0.5° is 99.17%, and the experimental result is 97.87%. Compared with traditional data-driven methods, it has higher accuracy and diagnostic efficiency, and compared with existing machine learning methods, it also achieves a high accuracy while reducing the network complexity. The effectiveness of the BP neural network robot joint fault diagnosis method is verified by experiments.

Structural topic modeling-based systematic review for service robots’ fault diagnosis: content, popularity and regional differences

“…Service robots are better adapted to complex environments and meet the essential work needs of human-machine collaborative interaction. Meanwhile, regional differences have brought new requirements to service robots' fault diagnosis, creating more unique service robots [35,76,80,81,87,92]. The main contributions of this review can be concluded as follows: (1).…”

“…In terms of the application of network algorithms, compared to the other two regions, the Asia-Pacific region has employed a more diverse range of network algorithms in robot fault detection. These include Convolutional Neural Networks [76][77][78][79], Back Propagation Neural Networks [80], Deep Residual Networks [35] and Generative Adversarial Networks [81]. Besides, abnormal diagnosis of signals (topic 7) is also a preferred research content in the Asia-Pacific region.…”

Neural Network-Enhanced Fault Diagnosis of Robot Joints

“…There will be no factors such as mechanical wear and deformation to interfere with the results. To compare with similar research [2], the advantages of this research is that the database generated by multiple features of the forward kinematics, which could improve the diagnostic accuracy and classification and, in this way, the features and data fusion methods can be updated according to practical needs.…”

“…With the wide application of informatisation, digitalisation and intelligentisation technologies, the automation level of manufacturing enterprises has been significantly improved, and flexible production lines aiming at rapid automated machining of multispecies precision parts have been proposed, which is of great significance and value for the transformation and upgrading of the manufacturing enterprises' production methods, equipment manufacturing capabilities and product performance [1]. In the processing of flexible production line, it is particularly important to maintain the product quality and quantity, and improve production efficiency, which directly affect the production efficiency and economic benefits of enterprises [2]. In the past decades, a large amount of research has been devoted to production systems, focusing on modelling, performance analysis, bottleneck identification, lean design, product quality inspection and production control of production systems [3].…”

“…This approach is common to this study in that both identify errors by comparing theoretical and practical results, but the disadvantage is that fault identification based on non-linear models requires more calculations and has a certain delay in identifying faults in real time. There is a similar study that collects the trajectories of different faulty robotic arm movements to build a database and train it with the network to finally achieve fault classification [2]. The disadvantage of this research is that the results obtained by fault classification with a single feature have low credibility and it does not give the possibilities for industrial applications, i.e., it does not separate faulty situations from non-faulty situations, which becomes a disadvantage of the application of this research.…”