Remote Detection of Control Fault Signals in Robotic Arms Based on Time-frequency Clustering Characteristics
In practical industrial applications,in order to timely detect the control faults of the robotic arms,real-time online data acquisition,transmission and signal processing of the key parts of the arms are needed.However,when the detection signal of the field sensor is transmitted to the control room,the current fault signal will cause data interference,error code,loss and other problems,which cannot guarantee the accuracy of remote collection and detection of the cloud platform.Therefore,remote detection of robotic arm control fault signals is proposed based on time-frequency clustering features.Based on the the generalized S transform of the control fault signal,the improved threshold function of the index function and the regulation parameters are designed,and the time-frequency distribution of the remote control fault signal is analyzed.After the mechanical arm is faulty,it is verified that the effective fault signal component has time-frequency aggregation.According to this conclusion,the wavelet analysis and Wiener filtering method are used to detect control fault signal and realize mechanical arm control fault signal detection.The experimental results show that the instantaneous eccentricity results of each joint of the proposed method's robotic arm remain relatively constant within the range of 0～300 rad,and the harmonic components of each joint are maintained between 0～20 Hz during rotation.The remote detection accuracy of control fault signals is maintained at around 98％,indicating good detection ability.It is recommended to use this method in mechanical fault detection.
wavelet packet transformmechanical armcontrol fault signaltime frequency aggregationgeneralized S transformationWiener filtering
NETL
NSTL
万方数据
