Fault Diagnosis And Fault Tolerant Control Of Multi-Joint Cooperative Robot

Multi-joint collaborative robots are a type of robots that have multiple joints and require collaboration with humans to complete tasks.The safety of these robots has always been highly concerned by the industry,and fault detection,diagnosis,and fault-tolerant control are crucial for their safe and stable operation.The operating environment of multi-joint collaborative robots is complex and constantly changing,and various sensors in the system are easily affected by joint movements,resulting in low fault detection accuracy,which can lead to misjudgments and endanger the safe operation of the system.Once a fault occurs,fault diagnosis and fault-tolerant control can ensure the safe and stable operation of the multi-joint collaborative robot system to a certain extent.Considering the serious coupling nonlinearity between the joints of multi-joint collaborative robots and the significant influence of external disturbances and internal parameter perturbations on their control effect,the control accuracy is reduced.This paper conducts in-depth research on the above problems in the aspects of fault detection,diagnosis,and fault-tolerant control of multi-joint collaborative robots.The main research content and innovative points are as follows.(1)In response to the problem of low fault detection accuracy caused by the complex and constantly changing operating environment of multi-joint collaborative robots and the susceptibility of various sensors in the system to joint movements,a fault detection framework based on multi-sensor data fusion considering multi-frame visual image fusion and complex extended Kalman filtering is proposed.Firstly,in order to effectively solve the problem of low precision in estimating the end position using visual sensor data due to image jitter of multi-joint collaborative robot’s visual sensors,a multi-frame visual image fusion algorithm is adopted.Secondly,a complex extended Kalman data fusion algorithm for different sensor data is designed to solve the problem of time asynchrony between low-bandwidth visual sensors and high-bandwidth position sensors,thereby improving the real-time and accuracy of end position estimation.Then,based on multi-sensor data fusion,a fault detection algorithm combining residual chi-square and improved sequential probability ratio test is proposed to improve the detection lag problem of the conventional sequential probability ratio test.Finally,the effectiveness of the proposed method is further verified through simulations.(2)A fault classification method for multi-joint collaborative robots considering spatio-temporal multi-scale information is proposed to address the problems of low accuracy of traditional fault classification due to multi-joint associative coupling of multi-joint collaborative robots,non-stationary strong noise and obvious spatio-temporal characteristics of data.First,the sensor and actuator data of the multi-joint collaborative robot are reconstructed to remove the noise in the data from the principle;second,a neural network fault classification model is designed considering the spatio-temporal multi-scale feature data of the multi-joint collaborative robot,and the attention mechanism is introduced in the model,and the deep convolutional neural network and recurrent neural network are combined to solve the problem of feature extraction with spatio-temporal feature data and Finally,the effectiveness of the proposed fault classification method is verified by simulating the multi-joint collaborative robot data.(3)A fault-tolerant control strategy for multi-joint collaborative robots based on fractional-order sliding mode is proposed to address the problem that the coupled nonlinear characteristics of multi-joint collaborative robots between joints,external disturbances and internal parameter uptake aggravate the degradation of multi-joint collaborative robot control performance under fault states and reduce the trajectory tracking accuracy.First,a state observer is designed to realize the estimation of the velocity and acceleration of the joint operation,which solves the problem that the differential operation is susceptible to noise interference;second,a nominal system model is introduced to the inner-loop torque loop,and a disturbance observer is designed to ensure that the torque performance tracks the nominal system output;then,the fractional-order sliding mode fault-tolerant controller is designed by using the output of the designed state estimator and disturbance observer and combining the fractional-order differential operator theory;finally,the fault-tolerant control performance of the algorithm on collaborative robot trajectory tracking is verified by simulation and experiment.(4)Integrating the key technologies of fault diagnosis and fault-tolerant control mentioned above,a fault detection and fault-tolerant control process for multi-joint collaborative robots was designed.Repeated experiments were conducted on the sensor and actuator fault detection of multi-joint collaborative robots using the constructed test bench.At the same time,when the actuator of a multi-joint collaborative robot malfunctions,a fractional-order sliding mode faulttolerant control algorithm software was developed using an experimental platform to verify the trajectory tracking accuracy of the multi-joint collaborative robot under actuator failures.This experimental plan was tested on the multi-joint collaborative robot platform of AUBO Company for a long time and at high frequencies,achieving the expected results.It passed the inspection of the company and on-site applications,and achieved practical proof.