基于支持向量回归机的船舶操纵运动在线辨识建模

Online Identification Modeling of Ship Manoeuvring Motion Using Support Vector Regression

采用滚动时间窗的方法实现支持向量机的在线辨识。以船舶操纵运动响应模型为研究对象,并由10°/10°和15°/15°仿真Z形试验数据构造支持向量机参数辨识所需的训练样本对,应用支持向量机进行船舶操纵运动在线辨识建模,回归操纵运动响应模型中的操纵性指数,并利用建立的响应模型进行Z形试验的数值模拟。将Z形试验数值模拟结果同仿真Z形试验数据进行比较,结果表明,在线式支持向量回归机是一种进行船舶操纵运动在线辨识建模的有效方法。

Support vector machine(SVM)is applied to conduct the online identification modeling by using the sliding time window.The response models of ship manoeuvring motion is taken as the research object,the training sample pairs required for the parameter identification of the SVM are constructed by using the simulated 10°/10°and 15°/15°zig-zag test data,and the SVM is used to establish the online identification model of ship manoeurving motion.The manoeurving index in the manoeurving motion response models is regressed,and the numerical simulation of the zig-zag test is performed using the established response models.The numerical simulation results of the zig-zag test are compared with the simulated zig-zag test data.The results show that online support vector regression machine is an effective method for online identification modeling of ship manoeurving motion.