Abstract: Turbine draft tube pressure fluctuation signals are very complicated and show a strong instability in fault diagnosis of hydroelectric generating units. This paper presents a new method for denoising pressure fluctuation signals of hydraulic turbines based on variational mode decomposition (VMD) and permutation entropy (PE). In this method, a pressure fluctuation signal is decomposed by VMD, and a number of its BIMF components are obtained. Then, the permutation entropy of each modal component is calculated, and the pressure fluctuation signal is screened and reconstructed by using these permutation entropies’ sensitivity to the feature of random signals. Comparison of simulation results shows that our new method can effectively denoise the signals and is better than the commonly-used EEMD filtering algorithm.