基于SVM的水下机器人预测S面控制  

Predictive method of S surface control using on SVM for AUV

作　　者：张国成[1,2] 万磊[1,2] 李岳明[1,2] 

机构地区：[1]哈尔滨工程大学船舶工程学院,黑龙江哈尔滨150001 [2]哈尔滨工程大学水下机器人重点实验室,黑龙江哈尔滨150001

出　　处：《华中科技大学学报（自然科学版）》2012年第3期40-44,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(51009040)

摘　　要：针对舵桨联合操控水下机器人系统的非线性特点,将预测控制的思想引入经典S面控制中,构造了一种基于支持向量机(SVM)的预测S面控制器,改善了S面控制器的控制效果,增强其自适应性.用支持向量机辨识水下机器人的非线性系统模型,充分发挥了SVM的泛化能力,能准确预测其运动状态.构造二次型性能优化函数以获取S面控制器的最优控制参数,进而获得水下机器人最优控制律.仿真结果表明:基于支持向量机的预测S面控制器具有结构简单、响应速度快、鲁棒性好等优点可行且有效.A predictive S surface controller based on support vector machine（SVM） was obtained to analyze typical nonlinearity control system of torpedo-shaped autonomous underwater vehicles（AUV）.The control algorithm was based on the S surface controller,which imported the idea of prediction control and combined with the excellence of support vector machine.Firstly the model of AUV was established to predict the motion states by the SVM.Then the control parameters of S surface controller were optimized by the quadratic performance function.At last the best control law was obtained.Simulation on AUV shows that the controller has merits of easy design,quick response,and excellent robustness,which is feasible and effective.

关 键 词：智能水下机器人 运动控制 支持向量机 预测S面控制 参数优化 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]