Pressure fluctuation is one of the main parameters to judge turbine stability. However, there is limitation in judging turbine stability based on the amplitudes of pressure fluctuation only. By applying the multi-channel high precision pressure fluctuation test system, the pressure fluctuation in different parts of a high head Francis turbine under partial load were collected synchronously. Then the principal components of each pressure fluctuation data were extracted by cross-correlation analysis and complete ensemble empirical mode decomposition (CEEMD) method. Finally, the sample entropy of principal component data of each pressure fluctuation signal was calculated. And then the distribution of sample entropy of different monitor points in operating range was obtained. Base on the results, it can be concluded that vortex rope in draft tube was still the main factor affecting the stability of high head Francis turbine under part load condition. And the frequency information of the principal component of the pressure fluctuation at each part of the Francis turbine can be used to infer the type of the hydraulic instability source. According to the joint comparison of the pressure fluctuation amplitude distribution and the sample entropy distribution at key monitor points, it can be proved that the variation trend of the sample entropy of principal component of pressure fluctuation can reflect the stability of hydraulic turbine. Therefore, the trend analysis of the sample entropy of principal component of pressure fluctuation can be popularized as a technical method to study the variation of the inner flow state of turbine.