A Fault Diagnosis Method for Hexapod Robot Combining Digital Twin Technology with GCN-LSTM Framework
In order to solve the problems of difficult real-time monitoring,less fault characteristic data and low fault diagnosis accuracy when multi-legged robots working in a closed and complex environment,a fault diagnosis method of six-legged robots based on digital twin virtual data and GCN-LSTM,which hybridizes graph convolutional network (GCN) and long short term memory network (LSTM). First,the dynamics model of the hexapod robot is analyzed,and the high-fidelity and high-confidence twin model of the robot is constructed in CoppeliaSim simulation software. Second,virtual fault injection was applied to the digital twin model,ensuring the safety of the robot. Under simulated fault conditions,the digital twin model controlled the physical robot's gait,obtaining high-confidence fault data samples from each physical sensor. Finally,to fully exploit the spatial correlations and temporal dependencies in sensor data,GCN and LSTM were integrated to achieve precise fault classification. Experimental results show that GCN-LSTM has the highest fault diagnosis accuracy compared with other algorithms of the same type. The combination of high-confidence fault data from the robot's digital twin system with GCN-LSTM enables accurate diagnosis of robot faults.
digital twinfault diagnosisGCN-LSTMsix-legged robot
万方数据
维普
