基于输出误差模型优化的甲板运动预报算法研究

本文提出了一种适用于多种复杂海况的大型舰船甲板运动预报方法，目的在于提高算法对不同海域复杂海况的适用性，以及对甲板运动模型的辨识精度与预报精度。该方法通过将量测数据的时间滞后处理引入输出误差模型来描述甲板运动的动力学模型，引入定阶准则确定了模型最优阶数数对。在此基础上应用了辅助模型递推最小二乘算法进行系统参数辨识并估计输出误差模型中的状态变量。实验结果表明，本文所提出的预报方法在系统参数辨识阶段可以将递推最小二乘算法的辨识精度提高5.13%，并且在预报阶段可以有效地将甲板运动的幅值与相位预测精度提高3.17%。该方法在复杂海况下具备良好的预测性能，适用于大型舰船甲板运动预报。

Research on deck motion prediction algorithm based on output error model optimization

This paper presents a method for large ship deck motion prediction for many complex sea conditions, with the purpose of improving the applicability to complex sea conditions in different seas, as well as the recognition and prediction accuracy of the deck motion model. The method describes the dynamics of the deck motion by introducing a time lag processing of the measurement data into the output error model. And the order estimation criterion is introduced to determine the optimal order pair of the model. The auxiliary model recursive least squares (AM-RLS) algorithm is applied to identify the system parameters and estimate the state variables in the output error model. The experimental result shows that the proposed prediction method can improve accuracy of the recursive least squares algorithm by 5.13% in the system parameter identification period and can effectively improve the prediction accuracy of the deck motion’s amplitude and phase by 3.17% in the prediction period. The method has good prediction performance under complex sea conditions and is suitable for the large ship deck motion prediction.