基于ANN的水下机械臂仿真控制策略研究  

Research on simulation control of strategy shipborne under water robotic arm based on ANN

作　　者：唐宇 马胜伟[1,3] 马振华 吴洽儿[1,3] 杨兴泽 王绍敏[1,3] 黄应邦 TANG Yu;MA Shengwei;MA Zhenhua;WU Qiaer;YANG Xingze;WANG Shaomin;HUANG Yingbang(Key Laboratory of Open Sea Fishery Development,Ministry of Agriculture and Rural Affairs,South China Sea Fisheries Research Institute,Guangzhou 510300,China;College of Navigation and Naval Engineering,Dalian Ocean University,Dalian 116023,China;Hainan Engineering Research Center of Deepsea Aquaculture and Processing,Sanya Tropical Fisheries Research Institute,Sanya 572024,China)

机构地区：[1]中国水产科学研究院南海水产研究所农业农村部外海渔业开发重点实验室,广东广州510300 [2]大连海洋大学航海与船舶工程学院,辽宁大连116023 [3]三亚热带水产研究院深远海养殖与加工海南省工程研究中心,海南三亚572024

出　　处：《舰船科学技术》2025年第4期76-83,共8页Ship Science and Technology

基　　金：农业农村部财政专项2024(B050101,B050102);广东省财政专项资金项目(2024-03024025);海南省科技专项资助(ZDYF2022GXJS347);海南省自然科学基金资助(323MS121);中国水产科学研究院基本科研业务费资助(2023TD58,2023TD93)。

摘　　要：针对水下机械臂在复杂水流环境下操作精度的提升问题,提出了一种基于自适应神经网络的滑膜控制策略。采用Newton-Euler法构建了适用于恒定均匀水流影响的双关节水下机械臂动力学模型,以提高模型精度并充分考虑实际应用中的环境因素。针对模型中存在的错误和不可预知的外部干扰,设计了融入自适应神经网络的滑膜控制器,有效补偿模型不准确带来的误差,抵抗多种类型的不确定外部扰动,从而显著改善了水下机械臂的轨迹跟踪速度和准确性。通过Matlab/Simulink平台进行仿真实验显示,相比于传统的PD滑模控制算法,提出的自适应神经网络滑膜控制方法在均匀水流干扰下,大大提高了水下机械臂的跟踪响应速度和精度,增强了系统在动态变化水流条件和模型不确定性面前的适应性。该策略的实施不仅提升了控制精度和响应速度,还保证了系统的稳定性,为水下机械臂在现代化海洋牧场等复杂水下作业环境中的应用提供了可行的解决方案。To address the problem of improving the operating accuracy of underwater robotic arms in complex water current environments,a slip membrane control strategy based on adaptive neural network is proposed.A kinetic model of a double-jointed underwater robotic arm applicable to the influence of constant and uniform water currents is constructed by using the Newton-Euler method in order to improve the model accuracy and fully consider the environmental factors in prac-tical applications.In response to the errors and unpredictable external disturbances in the model,a sliding film controller in-corporating an adaptive neural network is designed to effectively compensate for the errors brought about by model inac-curacies and resist multiple types of uncertain external disturbances,thus significantly improving the trajectory tracking speed and accuracy of the underwater robotic arm.Simulation experiments via the Matlab/Simulink platform show that,com-pared with the traditional PD sliding mode control algorithm,the proposed adaptive neural network sliding film control method greatly improves the tracking speed and accuracy of the underwater robotic arm under the uniform water current dis-turbances,and enhances the adaptability of the system in the face of dynamically changing water current conditions and mod-el uncertainties.The implementation of this strategy not only improves the control accuracy and response speed,but also en-sures the stability of the system,which provides a feasible solution for the application of underwater robotic arms in com-plex underwater operating environments,such as modernised sea ranches.

关 键 词：机械臂 神经网络 滑膜控制 自适应 

分 类 号：U664.6[交通运输工程—船舶及航道工程]