Due to the large-scale construction of ultra high voltage direct current (UHVDC) transmission projects, the demand for reactive power regulation at the receiving end of UHVDC is increasing day by day, so the synchronous condensers is applied in the power system. In order to solve the vibration problem caused by unbalanced load distribution of the synchronous condensers, a fault diagnosis method based on variational mode decomposition and multiple-kernel support vector machine (VMD-MSVM) for load distribution of the synchronous condensers is proposed. Firstly, several measuring points are arranged outside the bearing seat of the synchronous condenser to collect vibration signals. Secondly, according to the characteristics of the signal nonlinear and complicated composition, the vibration signal is decomposed into multiple intrinsic mode function by using VMD modal function. Thirdly, intrinsic mode function is selectd to reflect the characteristics of failure mode, and calculate the energy entropy to constitute feature vector. Finally, the appropriate kernel function is selectd to build a multi-kernel support vector machine, and the feature vectors are input to multi-kernel support vector machine for fault diagnosis. The experimental results show that this method is suitable for the fault diagnosis of load distribution and has better fault identification effect.