混合建模法在船舶横摇运动建模中的应用研究

Application of Mixed Modeling with CNN in Ship Roll-Model

为了反映出船舶横摇运动机理模型没能体现出的船舶横摇运动的非线性特征和机理模型所作的简化处理所遗漏的各种动态、静态信息,文章以复合神经网络作为补偿模型,建立了船舶横摇运动混合模型。仿真结果表明,复合神经网络混合模型输出在120秒后非常接近期望输出,比一般神经网络混合法更接近期望输出,从而证明复合神经网络混合建模法可以得到精确的、强泛化力的船舶横摇运动模型。复合神经网络混合建模法不仅适用于船舶运动建模,也适用于其他复杂系统、不确定系统的建模。

For the sake of consideration of nonlinear characteristic of ship,which is not shown by the ship roll-mechanism model,and the dynamic and static information,which is missed because of model simplification,this paper used the combined neural network as compensation model to deduce the mixed roll model of ship.The simulation demonstrates that the output of the mixed model approaches the expectation greatly after 120 seconds.So it＇s proved that more accurate ship roll-model with better generalization ability can be found by this approach.The mixed model can be not only applied to the simulation of ship motion,but also can be applied to the simulation of other complex or uncertain systems.