More and more hydropower plants are operating as regulator of renewable energy such as solar and wind energy resources. Because these renewable energy resources are highly volatile, so hydraulic turbines have to switch working condition in a wide flow range to guarantee the steady output to electric grid. It's known that there are many fluctuating frequency components in runner passage of hydraulic turbines, especially at part-load conditions, which are induced by unsteady flow phenomena like guide vane wake, flow separation or rotor-stator interaction (RSI) and so on. These fluctuation frequency components are varied with working condition due to different dominant inducing resources. Although we can estimate the frequency component resources according to rotating speed, numbers of blades and guide vanes, we can't know exactly coherent structure pattern of every frequency component. In present study, firstly, we will compute the internal flow of a Francis turbine at lower discharge, and verify the validation of computational results by experimental data. Secondly, Dynamic Mode Decomposition (DMD) method will be employed to extract the three-dimensional coherent structures in runner passage and draft tube for every frequency component based on computational snapshots. Finally, we will demonstrate the patterns and locations of coherent structures for every frequency component and predict the coherent structure or fluctuation component generation resources.