Abstract: Aiming at the problem of low efficiency and high cost of manual positioning of abnormal noises in nuclear power plants, this paper proposes a method based on improved variational mode decomposition (VMD) to identify and locate abnormal noises in motor bearings. Firstly, this method first decomposes the motor sound signal according to the improved VMD algorithm, calculates the effective value and kurtosis index of the best mode component, and initially realizes the recognition of the abnormal sound of the motor. Secondly, through the sound propagation characteristics in the air, using the size of the effective value of the test signal at different locations was used to achieve the location of the different sound source. The simulation experiment of the location of abnormal sound of the motor verifies the feasibility and effectiveness of the proposed method.