Research on multi cross-section stage-discharge relationship based on BP neural network

To overcome the deficiencies of discharge measurement method in open channel, BP neural network (BPNN) is applied in fitting stage-discharge calculation, based on which design a multi cross-section stage-discharge method. BPNN-based single stage-discharge fitting model and BPNN-based dual stage-discharge fitting model were established as the research objects. The stage-discharge data of open channel acquired from the velocity-area method are used to train and test the model. Experiment and simulation results show that, compared with the single cross-section stage-discharge fitting model, the multi stage-discharge measurement model has higher prediction accuracy, and can exclude effect of backwater, it has been applied in flow prediction of open channels.