Abstract: Mechanical defects of gas insulated switchgear(GIS) equipment is one of the critical factors leading to its failure. Therefore, we studied the evolution laws of time domain and frequency domain vibration spectra of disconnector base looseness defects under the excitation of variable frequency currents, and quantitatively analyzed six vibration characteristic values, such as waveform factor and gravity frequency. The support vector machine optimized by the Grey Wolf Optimizer algorithm was used to propose a method for identifiying the severity of GIS disconnector base looseness defects based on variable frequency current excitation mode. The results show that there is a strong correlation between the vibration characteristic values of GIS disconnector base looseness defects and the frequencies of excitation currents. Compared with the single power frequency current excitation mode, variable frequency current excitation has the advantage of more effective detection of GIS mechanical defects. Based on the verification of the 126 kV true GIS experimental platform, the average recognition accuracy of the disconnector base looseness defects identification model is over 90%. The results of this paper provide important references for revealing the nonlinear mechanical vibration characteristics of GIS equipment and improving its mechanical defect detection level.