基于改进加权最小二乘支持向量机的X舵潜水器操纵运动模型辨识  

Parameter Identification for Maneuvering Model of Underwater Vehicle with X-rudder Based on Improved WLSSVM

作　　者：裴天琪 于曹阳 钟一鸣 连琏[1] PEI Tianqi;YU Caoyang;ZHONG Yiming;LIAN Lian(School of Oceanography,Shanghai Jiao Tong University,Shanghai 200030,China)

机构地区：[1]上海交通大学海洋学院,上海200030

出　　处：《中国造船》2023年第2期1-12,共12页Shipbuilding of China

基　　金：国家自然科学基金(51909161);上海市自然科学基金(22ZR1434600)。

摘　　要：针对稀疏的试验数据会降低潜水器操纵运动模型辨识精度的问题,提出了一种融合三次样条插值加密和改进加权最小二乘支持向量机(WLSSVM)的潜水器系统辨识方法。基于仿真试验和原理样机湖上试验获得X舵潜水器的操纵数据,应用三次样条插值加密策略对稀释的艏向角数据进行预处理,以降低数据稀疏对辨识结果的影响。设计了双曲正切函数改进型WLSSVM,对潜水器操纵运动响应模型进行高精度参数辨识和操纵运动预报。通过原理样机的湖上试验对该方法的有效性进行验证,与未改进的方法相比,预报值均方根误差最高降低了0.12 rad。A parameter identification method by combining cubic spline interpolation and optional weighted least square support vector machine(WLSSVM)was proposed to improve the identification accuracy of maneuvering motion model of an underwater vehicle with X-rudder in the presence of sparse test data.Heading angle data from the simulation with a zigzag of 30 degrees and the experiment were pre-processed using the method of cubic spline interpolation to reduce the influence of sparsity on the identification result.An improved WLSSVM algorithm with hyperbolic tangent function was applied to the maneuvering motion response model for parameter identification and maneuvering motion prediction.Comparing the proposed strategy with the original method,the root mean square error was reduced by up to 0.12 rad.

关 键 词：X舵 潜水器操纵性 参数辨识 三次样条插值 加权最小二乘支持向量机 

分 类 号：U661.33[交通运输工程—船舶及航道工程]