一种基于BP神经网络的江河潮位短期预测

A Kind of Short-term Height-Prediction of Tidal Level of Rivers Based on BP Neural Network Model

江河涌潮与海洋潮汐相比较,其潮位和潮时有着与海洋潮汐规律相同的一面,但它与海洋潮汐所不同的是,其更容易受到天文、气象等因素的影响,有着异于海洋潮汐的不稳定性。涌潮的潮位值变化就是各种因素综合结果的表现。通过借助BP神经网络可以逼近任意非线性函数的能力和特点,构建一个用于短期预测潮水潮位的模型,利用一小部分潮位的历史数据作为训练样本对构建出的BP神经网络模型进行样本训练,用其它的潮位历史数据对模型进行了预测验证。模型验证结果表明,这个BP神经网络预测模型能够对潮水潮位进行有效的短期预测,其性能指标符合要求。

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.