BP Model Study to the Current Velocity in Dongdagang Tidal Channel

Abstract

Based on the measured data on current velocity and current direction of six water layers of tidal channel at 4# station position in the radial sand ridge sea area for two consecutive tidal cycles, simulation and prediction concerning the current velocity and current direction of the rest water layers at the 4# station position during the tidal process has been carried out using the back propagation (BP) neural network model. Two best BP neural network models with three factors of surface current velocity, current direction and water depth as input vectors as well as current velocity and current direction of the rest layers as output vectors have been established successfully, with relatively high precision of simulation and prediction. It indicates that the vertical distribution of tidal current velocity is regular and the BP neural network models are an effective means for researching into tidal vertical structure. This research is capable of providing scientific basis for measuring and predicting current velocity and direction of tidal channel, calibrated artificial intelligence and automation.