In order to effectively suppress the periodic narrowband interference, a noise reduction method based on improved variational mode decomposition and permutation entropy is proposed. Firstly, aiming at the problem that variational mode decomposition algorithm (VMD) may under decompose or over decompose in the decomposition process, a method to determine the optimal mode decomposition number based on the mean orthogonal value is proposed. Secondly, the permutation entropy is applied to the selection and reconstruction of effective components to filter out the periodic narrowband interference according to the characteristics that the periodic narrowband interference is more regular than the partial discharge signal in a certain frequency band; This method, variational mode decomposition based on conventional method and ensemble empirical mode decomposition are used to reduce the noise of simulated a signals respectively. The results show that the method has high noise rejection ratio, small waveform distortion and can effectively remove periodic narrowband interference.