论文部分内容阅读
针对传统潮流分析方法在感潮河段实施走航数据潮流分离中的不足,提出了一种基于梯度训练法的径向基函数潮流分离方法,解决了传统潮流分析方法时序数据长、潮流分离实施复杂问题;也解决了基于贪婪拟合法的径向基函数潮流分离算法存在的节点位置无法准确确定以及过度拟合导致结果不稳定等难题;根据潮流分离结果重构流场,在徐六泾走航断面实验中取得了优于0.25m/s的外推精度。
Aiming at the shortcomings of the traditional tidal current analysis method in tidal current data separation in the tidal reach, a radial basis function tidal current separation method based on the gradient training method is proposed, which solves the problems that the traditional tidal current analysis method has long time series data and tidal current separation But also solve the problems that the location of nodes in the radial basis function tidal current separation algorithm based on greedy fitting method can not be accurately determined and the result of over-fitting results in instability. According to the tidal current separation results, the flow field is reconstructed, Extrapolation accuracy better than 0.25 m / s was obtained in the experiment of the sagittal plane.