基于输入输出线性化SVR的船舶航迹控制  

Ship Track-Control Method Based on SVR with Input-Output Linearization

作　　者：孙巧梅[1] 任光[1] 

机构地区：[1]大连海事大学轮机工程学院,大连116026

出　　处：《中国造船》2012年第4期91-99,共9页Shipbuilding of China

摘　　要：支持向量回归机(Support Vector Regression,简称SVR)是系统控制的一种新方法,相对于神经网络来说它的优势在于避免过学习、陷入局部极小点,获得全局最优解。结合输入输出反馈线性化理论,提出了反馈线性化SVR方法,采用交叉验证方法对SVR的参数进行优化,通过逆动态模型和逆误差补偿项的离线辨识,将辨识的逆模型作为控制器,设计出自适应逆控制系统。控制器应用到船舶航迹控制系统当中,在Matlab平台下,针对设定的航迹方案进行数字仿真试验。仿真结果表明该控制方案具有良好的动态响应性能和控制效果。Support Vector Regression （SVR）, featured as being capable of effectively avoiding over learning and local minimum, and getting global optimal solution, is a new method for model control problems. In combination with input-output linearization, an algorithm of feedback linearization-based SVR is proposed with inverse dynamic model and compensation term for the inversion error identified off-line. Then the inverse model is used as a controller to constitute an adaptive inverse control system. Parameters of SVR are optimized by K-fold Cross Validation algorithm. Finally, numerical simulations are conducted for predefined reference on the platform of Matlab, showing that the proposed control scheme has good performance of dynamic resoonse and control effect.

关 键 词：支持向量回归机 航迹控制 输入输出线性化 

分 类 号：U664[交通运输工程—船舶及航道工程]