基于BP神经网络的海啸爬坡高度预测

Prediction of tsunami climbing height based on BP neural network

海啸是一种灾难性的海浪,了解它的运动状态及其近岸的爬坡高度,对近岸基础设施建设,工程的防灾减灾具有重要意义。海啸的爬坡运动是一个复杂的非线性系统,通常将海啸简化为孤立波计算,孤立波爬高通常和坡度、波高、静水深度等因素有关,为解决这一非线性问题,提出了一种基于BP(back propagation)神经网络的预测近岸孤立波爬坡最大高度的方法。通过OpenFOAM建立二维数值波浪运动模型,计算得到数值模型爬高数据,通过与理论值的对比分析,该模型能够较好地模拟孤立波的爬坡运动。仿真数据与文献现有数据共计70组,将它们随机打乱后,其中50组用于模型训练,20组用于模型验证。通过对数据的训练和验证,最终结果表明,基于BP神经网络算法的模型在预测孤立波最大爬坡高度上具有较好的精度。

Tsunami is a kind of catastrophic wave.It is of great significance to understand its motion state and its climbing height near shore for the construction of offshore infrastructure and disaster prevention and mitigation of engineering.The climbing motion of tsunami is a complex nonlinear system,which is usually simplified to the calculation of isolated wave.The climbing height of isolated wave is usually related to slope,wave height,static water depth and other factors.To solve this nonlinear problem,a method to predict the maximum climbing height of isolated wave near shore based on BP(Back Propagation)neural network has been proposed.The two-dimensional numerical wave motion model is established by OpenFOAM,and the climbing data of the numerical model is calculated.By comparing with the theoretical value,the model can better simulate the climbing motion of isolated waves.A total of 70 groups of simulation data and existing data in the literature were randomly scrambled,of which 50 groups were used for model training and 20 for model validation.Through the training and verification of the data,the final results show that the model based on BP neural network algorithm has a good accuracy in predicting the maximum climbing height of isolated waves.