Affected by reservoir regulation, the complicated flow conditions and the traditional discharge monitoring means between Three Gorges and Gezhouba Dam are far cry from meeting demands of flood control, flood reporting and hydrological data processing. On the basis of H-ADCP on-line velocity monitoring data given by the CWRC, this paper extracts index-velocity by using wavelet analyzing techniques, handles the relationship of nonlinear mapping between index-velocity and section average velocity by making use of artificial neural network, builds an H-ADCP online discharge calculation model which can calculate the real-time sectional discharge in the Three Gorges-Gezhouba area under complicated flow conditions. The model also ensured timely and accurate hydrological data for flood control, flood reporting and hydrological data processing.