基于递归免疫网络在线辨识的AUV三维轨迹跟踪控制  

AUV 3D trajectory tracking control identified online by recurrent immune network

作　　者：王舜 江亚峰 张亮 刘维 袁明新[1,3] WANG Shun;JIANG Yafeng;ZHANG Liang;LIU Wei;YUAN Mingxin(School of Mechanical Engineering,Jiangsu University of Science and Technology,Zhenjiang 212003,China;T-SEA Marine Technology Co.,Ltd.,Zhangjiagang 215600,China;Zhangjiagang Industrial Technology Research Institute,Jiangsu University of Science and Technology,Zhangjiagang 215600,China)

机构地区：[1]江苏科技大学机械工程学院,江苏镇江212003 [2]张家港江苏科技大学产业技术研究院,江苏张家港215600 [3]中科探海(苏州)海洋科技有限责任公司,江苏张家港215600

出　　处：《舰船科学技术》2024年第13期119-125,共7页Ship Science and Technology

基　　金：工信部高技术船舶项目([2019]360);张家港市产业链创新产品攻关计划资助项目(ZKC2206);张家港市产学研预研资金资助项目(ZKYY2253)。

摘　　要：为了提高海流、海浪、水下噪声等扰动下的AUV三维轨迹跟踪精度,提出基于递归免疫网络在线辨识的PID自整定轨迹跟踪控制器(PID-RINN)。首先建立AUV的运动学模型,并将其深度距离、首向角和俯仰角作为控制变量,设计了基于神经网络在线辨识的PID控制器;然后借鉴生物免疫系统的信息处理机制构建了递归免疫网络;接着将水下机器人在水平面上距预瞄路径点的侧向距离定义为疫苗,并联合细胞隐层输出接种到递归免疫网络的突触隐层;最后基于梯度法实现了递归免疫网络辨识下的PID控制器在线自整定。测试结果表明,与PID、PID_GA、PID_RBF相比,文中轨迹跟踪控制的平均和最大位置误差分别平均减少31.91%和25.81%,平均和最大首向角误差分别平均减少32.54%和25.27%,以及平均和最大俯仰角误差分别平均减少61.93%和61.26%,从而验证了文中递归免疫网络在线辨识的AUV三维轨迹跟踪具有高控制精度,以及强扰动抑制优点。In order to improve the 3D trajectory tracking accuracy of autonomous underwater vehicle(AUV)under disturbances such as ocean current,ocean wave and underwater noise,a PID self-tuning trajectory tracking controller identified online by recurrent immune neural network(PID-RINN)is proposed.First,the kinematic model of the AUV is established,and the depth distance,heading angle and pitch angle of the AUV are taken as the control variables,and a PID controller identified online by the neural network is designed.Then a recursive immune network is constructed by referring to the information processing mechanism of the biological immune system.After that,the lateral distance between the underwater robot and the preview waypoint on the horizontal plane is defined as the vaccine,and is inoculated into the synaptic hidden layer of the recurrent immune network together with the output of the cellular hidden layer.Finally,based on the gradient method,the online self-tuning of the PID controller identified by the recursive immune network is realized.Test results show that compared with PID,GA_PID,RBF_PID,the average and maximum position errors of the proposed PID-RINN are respectively reduced by an average of 31.91%and 25.81%,the average and maximum heading angle errors are reduced by an average of 32.54%and 25.27%respectively,and the average and maximum pitch angle errors are reduced by an average of 61.93%and 61.26%respectively,which verifies that the 3D trajectory tracking identified online by the recursive immune network has the advantages of high control accuracy and strong disturbance suppression.

关 键 词：自主水下机器人 三维轨迹跟踪 递归免疫网络 在线辨识 PID控制器 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]