Abstract: In comparison between the tide bore of rivers and the tide of ocean, the occurrence  time and height of the river tide bore has the same feature as the ocean tide ； but the river tide bore is influenced more easily by astronomy, weather and other factors, so it has the instability that is different from the ocean tide ； the tide level of the tidal bore is the comprehensive result influenced by all kinds of factors. Since the BP neural network can approach any non-linear function, a short-term prediction model of the tidal height has been constructed. Some historical tidal data, as a training sample, are used to train the model, and others are used to predict the tidal data. Finally, the validity and the feature of the prediction model have been verified.

Key words: tide bore ,  , BP neural network  , tide level ,  , short-term prediction