遗传神经网络在二维潮流特性模拟中的应用

本文将水动力学模型与遗传神经网络方法结合，对深圳湾生态敏感点潮流的实时变化特性进行了预测。利用人工神经网络得出的模拟结果与经过实测资料验证的海湾二维潮流模型的模拟结果十分吻合，从而说明了将遗传神经网络用于二维潮流运动特征模拟的可行性。

Application of genetic algorithm-based artificial neural networks in 2D tidal flow simulation

A hybrid approach combining the 2-D hydrodynamic model for tidal flow with genetic algorithm-based artificial neural networks (GA-ANN) is presented. The site-specific knowledge and numerical results from the hydrodynamic model for several typical tidal patterns can be encapsulated in an artificial neural network and taken as the basis of the training in ANNs, which can significantly enhance the simulation speed. A case study is carried out for the real time process prediction of tidal characteristics in Deep Bay, Southern China. The GA-ANN functioned as non-linear dynamic system effectively reproduces the behaviors of the tides in the Bay for any given open boundary condition at the bay mouth. The verification results of GA-ANN are acceptable as compared with the results of numerical models.