【作者】 刘和平；

【导师】 龚振邦；

【作者基本信息】 上海大学 ， 机械电子工程， 2009， 博士

【摘要】 我国江河湖海水域众多,海岸线漫长,滨海、城市河流、港口、水库等水域星罗棋布。其中的资源和重要设施面临探测、开采、监控等诸多新的问题。城市河流重要地段、港口、水电站、滨海核电站、水库等重要目标面临恐怖主义、分裂主义和极端主义等三股势力的威胁,通过水下船底走私、贩毒、偷渡等违法活动时有发生。在举办重大节日庆典及重要政治、经济、体育活动时,相关邻近水域要进行严密监控。目前浅水水下机器人已广泛应用于水库堤坝检查、核电站检查、海上钻井平台水下部分的监测与修复,沉船考古、海底光缆检测、海带收割、绿藻探查以及水雷布放、猎雷与扫雷、水下侦察等等众多民用与军用领域。本项研究在于研制适合浅水域繁忙运输水道浑浊水体的无人有缆监控水下机器人及其处置机械手。针对无人有缆遥控的水下机器人(ROV)的发展、国内外现状、控制技术的进展以及ROV及其作业系统在应用中出现的问题,本文阐述了研究的背景、目标和内容,制定了ROV总体功能和指标,据此确定传感器和探测设备的性能指标,对设备进行了选型和配置。本文研究了水下机器人的结构、密封、控制系统的电路硬件、位姿采集系统及软件操控界面和光电脐带缆的设计、制造及其测试工作。对主体框架结构进行了有限元分析。研究了浮体的外形结构、材料及其性能、制造和安装工艺。给出了总体重心和浮心计算方法和结果,估算了水下机器人的运动阻力、功率和最大速度。计算了截面积形心作为确定推进器安装位置的依据。本文提出了已申请发明专利的具有创新意义的实用性网兜式机械手和抱持式机械手,从功能和结构设计进行了深入研究,分析了网兜式机械手的脱网结构,对两种机械手的手臂进行了有限元挠曲分析和应力分析,完成了运动仿真结果验证,最后对抱持式机械手进行了抓取试验。本文对ROV建立了固定坐标系和运动坐标系并对二者之间的转换进行了研究,对ROV受到的重力、浮力、推进力和水动力进行了分析,推导了水下机器人空间运动的一般方程,并根据结构对称性和使用特点推导了水平面、垂直面和横截面的简化动力学运动方程,提出了一种ROV的加速度水动力系数以及速度水动力系数的计算方法。在ROV的定向运动上,对滑模控制的优、缺点及其在ROV控制中的应用进行了总结,讨论了抖振的成因以及削弱方法的进展。重点研究了径向基(RBF)网络的理论方法和思路,对其非线性映射以及函数逼近能力进行了分析。根据误差反向传播机理,采用梯度下降法对基函数的参数和权值同时进行学习训练和调整。由于在学习过程中收敛的速度较慢,还有可能存在局部极小值问题,所以引入附加动量修正法对这种方法加以修正。在定向控制器设计上,采用滑模控制,将整个系统分成名义模型和由水动力和干扰构成非确定部分,对于确定部分采用状态反馈增益方法对名义模型进行控制,同时采用RBF神经网络作为滑模控制中动态可调的补偿控制器对非确定部分的上界进行学习和逼近。经过仿真和试验,控制效果得到了证实。对ROV的定深和姿态控制,首先对全部推进器进行敞水试验,分别得到正反向的控制电压-推力测试数据,绘出了曲线。然后用RBF神经网络进行学习,实现了非线性推力曲线的神经网络建模逼近。然后研究了模糊控制的理论和方法,并对模糊推理的逼近精度进行了分析。采用模糊控制加神经网络组成串联控制器,根据上下推进器的敞水控制电压-推力实验数据,以推力为输入,控制电压为输出,用RBF神经网络进行学习和训练,由模糊控制器根据深度误差和误差变化率进行模糊推理,得到深度调节用的推力,输入神经网络得到控制电压,从而精确控制推进器产生需要的推力。试验结果证明了控制方法的有效性。网兜式机械手作业时会引起的水下机器人姿态纵倾变化,本文提出并设计了一个姿态平衡调整装置,采用滑模控制方法调节姿态纵倾角,由于滑模控制方法在控制切换时不可避免的具有较大的抖振,为了减少抖振,采用干扰观测器对干扰项进行估计,仿真表明实现了抖振的削弱。针对水下机器人在浪涌作用下的横摇运动进行了分析、建模和数值计算,提出了一种采用ROV侧向推进器对ROV的浪涌横摇响应过程进行干扰的自适应模糊滑模控制方法,以减少横摇的幅值。数值计算和仿真结果显示了一定的效果,对浪涌下ROV的姿态调整进行了有益的探索。最后,本文对研究过程进行了总结,展望了下一步的研究方向和内容。更多还原

【Abstract】 There is much water area in China including reservoirs, rivers, lakes, and ports and with a long coastline. The water area and impotant resource and installations within it are faced with some new problems such as exploration, exploitation and monitoring. The parts of rivers through cites, ports, water power stations, nuclear power plants and reservoirs are threatened by hostile forces including the three forces of terrorism, splittism and extremism. There are also many illegal activities such as contraband, drug trafficking and steal a crossing through underwater. When holding important festival celebrations and activities in political, economical and physical fields, the related water areas must be in monitoring and control. The underwater robots have now been widely used in civil and military fields such as hidden defects detection of dam and dyke, danger inspection of nuclear power stations, underwater structure examining and repairing of offshore drilling platforms, wreck archaeologizing, repairing of submarine optical cable, kelp reaping, green algae exploration, laying and sweeping of mines, underwater reconnaissance and so on.The aim of the project is to study and develop a monitoring underwater robot along with manipulators applicable to turbid water of busy water way in shallow region. The background, goal and study contents were expatiated in this article in allusion to the development actuality of Remotely Operated Vehicle(ROV), its manipulators and control technology. The problems produced in operating practise were taken into account. The function and index in collectivity were established and performance index of sensor and dectecting equipments were determined with that. The choosing and configuration are made for some apparatus.The study in this paper ranges from the structure, waterproof technology, circuits hardware of control system, collection system of position and gesture of ROV to design, manufacture and tests of photoelectricity tether. A finite element analysis was made for main body frame structure. The figure, material, manufacture and installation process were investigated. The calculating methods and results of barycenter and floating center were provided. The movement resistance, power and maxmium speed were estimated. The shape centers of cross-section along each direction were computed as the basis to determine the installation places of the thrusters.The creative Tuck Net Manipulator and Clasp Manipulator were presented, with which patents of invention have been applied. A deep study of function and structure of manipulators were carried out and finite element analyses on deformation and stress were made for the arms. The structure of net doffing of the Tuck Manipulator was analyzed. Both motion simulations verified the designs. The grasp tests of Clasp Manipulator were accomplished.The fixed and motion coordinate systems are set up for the ROV and the mutually conversion between them were studied. The gravity, buoyancy, thrust and hydrodynamical factors were analyzed. The general equations of space motion of ROV were derived. And according to the structure symmetry and use characteristics of ROV, the simplified motion equations on horizontal, vertical and cross section were also derived. A calculating method for the hydrodynamical coefficients of accelaration and velocity of ROV was presented.For the yaw control of ROV, the sliding mode control(SMC) method was introduced first on the control theory of ROV. The advantages and disadvantages of SMC and its application on ROV control were summarized. The cause of chattering formation in SMC and development of weakening methods were discussed. Then, the Radial Basic Function (RBF) neural network was paid more attention. And the emphasis of the analysis is on the nonlinear mapping and ability of approximating a function of RBF. According to the error back-propagation mechanism, the gradient descent method was adopted to train and adjust the weights and parameters of the basic function simultaneously. Since the speed of convergence is slow and the local minimum is possible in presence, additional momentum method was introduced to make a modification during the learning process. The sliding mode control method was used. The system is divided into a nominal model and an uncertainty composed by hydrodynamical force and disturbance. The state feedback gain method was used to control the certainty-- nominal model. As the compensating controller that is dynamic adjustable, the RBF network was adopted to learn and approach the upper bound of the uncertainty. The effects were verified by simulation and tests.For the control of depth and attitude of ROV, the open water tests were done first for each thruster. The nonlinear testing data of control voltage-thrust in forward and reverse was obtained respectively. The data were used to train the RBF network and the neural network approaching models for nonlinear thrust curves were set up. Then, the theory and contents of the fuzzy reasoning were studied and the approaching precision was also evaluated. A series controller with fuzzy control and RBF network was used to control the keep-depth motion of ROV. Let thrust be the input and control voltage be the output, the RBF network was trained bsed on the open tests data of control voltage-thrust. The fuzzy controller makes the reasoning and gets the desired thrust according to the error of the depth and its rate. As the input, the thrust is then introdued into the neural network and a desired control voltage is acquired to produce a precision thrust.When working, the Tuck Net Manipulator would cause the ROV pitch. A balance device was presented and designed by using a sliding mode controller to adjust the pitching. As it is inevitable for the chattering to occur in the control switch of SMC, A disturbance observer was used to estimate the disturbances to weaken the chattering. The simulation had shown the effect.In allusion to the rolling motion of ROV in wave, the analysis, modelling and numberical calculation were carried out. An adaptive fuzzy sliding mode control method was presented to disturb the rolling response process of the ROV by using the lateral thrusters to reduce the rolling amplitude. The simulation indicated some effects. It’s a probe for the attitude adjusting of ROV in wave.Finally, the paper summarized the whole study and brought forward future direction and contents of the study.更多还原