基于BP神经网络的厦门沿海风暴潮预报应用

Application of storm surge forecasting by BP artificial neural network off coast of Xiamen

以BP人工神经网络可有效描述非线性问题的特性应用于短期风暴潮增水预报,利用风暴潮增水与各项影响因素的关系,建立厦门沿海的风暴潮增水预报的人工神经网络模型。该模型将以台风中心最低气压、最大风速,七级大风半径、台风中心距测站位置的距离和测站当地气压、当地风速、天文潮位及增水值、作为主要的输入因子,预测未来1 h、2 h、3 h及6 h风暴潮增水值。分别探讨厦门沿海的风暴潮増水在3种代表性热带气旋路径的影响下的模型应用情况,由预报结果的分析显示:该BP神经网络模型所预报的风暴潮增水较好的拟合了实际变化趋势,表明本模型对于厦门沿海风暴潮増水的预报具有相当不错的成效。

The relations between storm surge and the influence factors are used to establish the storm surge forecasting model based on Back-propagation artificial neural network off coast of Xiamen. The inputs of model are the major parameters of the typhoon, including the current data about the central pressure of typhoon, the maximum wind speed of typhoon, the seven-grade typhoon radius, the local station pressure, the local station wind velocity, the local station astronomical tide, the storm surge height, the distance from the center of typhoon to the station. The model can be used to forecast the storm surge height one hour to six hour ahead. The study discussed the model analysis situation of the effects of three different typhoon routes. Then the original data of three typhoons is used to verify the present model. The results indicate that the storm surge predicted by the BP neural network model is in good agreement with measured data. It＇s indicated that the model built on the back-propagation artificial neural network can be effective in storm surge forecasting off coast of Xiamen.