改进多新息卡尔曼滤波法辨识船舶响应模型  

Identification of ship response model based on improved multi-innovation extended Kalman filter

作　　者：谢朔 陈德山 初秀民 柳晨光 

机构地区：[1]国家水运安全工程技术研究中心,湖北武汉430063 [2]武汉理工大学能源与动力工程学院,湖北武汉430063

出　　处：《哈尔滨工程大学学报》2018年第2期282-289,共8页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(61273234);国家青年科学基金项目(51609193);湖北省自然科学基金项目(2015CFA111)

摘　　要：为在船舶自主航行控制中提供参数精确的船舶响应模型,本文提出一种改进的多新息扩展卡尔曼滤波参数辨识方法。引入遗忘因子以降低历史干扰数据的累积影响,并从理论上分析了改进后算法辨识的收敛性,证明了改进后的辨识算法在一定条件下有界收敛。为验证所提出辨识方法的有效性,本文在获取真实的模型船Z型试验数据基础上,将改进算法辨识方法与传统扩展卡尔曼滤波辨识方法进行对比,结果表明所提出的改进算法辨识的船舶响应模型参数更加精确,预报均方误差可达到2(°)2以下。In order to provide an accurate ship response model in ship's autonomous navigation control,an improved multi-innovation extended Kalman filtering method for parameter identification was proposed. A forgetting factor was introduced to reduce cumulative influence of historical interference. Convergence of improved algorithm identification was theoretically analyzed,and it was proved that the improved algorithm has a bounded convergence under certain conditions. Based on true Z experimental data of a model ship,the identification results of improved method were compared with those of traditional Kalman filter identification method to verify the effectiveness. The experimental results show that the improved method can identify ship response model parameters more accurately,and the prediction mean square error can reach below 2( °)2.

关 键 词：响应模型 参数辨识 扩展卡尔曼滤波 多新息 遗忘因子 收敛性 Z形试验 操纵性预报 

分 类 号：U661.3[交通运输工程—船舶及航道工程]