基于音频信号工业机器人关节异常运行检测方法研究OA

An industrial robot operational status monitoring system is designed.The system adopts a joint monitoring device to acquire audio signals generated by the operation of robot joints.To address the difficulty in abnormal feature analysis of audio signals,an SVMD_IBWO_MCKD method is proposed.First,decomposes the audio signal into multiple intrinsic mode functions(IMFs)using the sequential variational mode decomposition(SVMD)method,and then screens out the optimal IMF through the Gaussian weighted kurtosis index.Secondly,the improve beluga whale optimization(IBWO)algorithm is utilized to adaptively select the parameters T,M and L of maximum correlation kurtosis deconvolution(MCKD),and perform MCKD processing on the selected optimal IMF.Finally,fault features in the robot joint audio signals are extracted through envelope spectrum analysis.The experimental results show that neither the BWO_MCKD nor the WOA_MCKD method can extract the effective octave,and the SSA_MCKD method can only extract the 3-octave component.In contrast,the SVMD_IBWO_MCKD method can effectively extract the fourfold frequency of the periodic fault frequency in the joint audio signal.