SIA OpenIR  > 水下机器人研究室
全海深水下机器人建模与控制研究
其他题名Modeling and Control of Full-Ocean-Depth Underwater Vehicles
刘鑫宇1,2
导师封锡盛 ; 李一平
分类号TP242
关键词水下机器人 全海深 动力学建模 自适应控制
索取号TP242/L75/2018
页数116页
学位专业机械电子工程
学位名称博士
2018-05-21
学位授予单位中国科学院沈阳自动化研究所
学位授予地点沈阳
作者部门水下机器人研究室
摘要本文的研究工作依托中国科学院战略先导科技专项相关项目的相关课题开展研究,对全海深水下机器人建模和控制中遇到的重点和难点问题进行分析并给出解决方案,以“海斗”号自主遥控水下机器人(“海斗”ARV)为实验平台。主要研究内容如下:(1) 全海深水下机器人运动仿真及实验平台。介绍了实验平台的硬件条件。通过对全附体模型进行计算流体动力学(CFD)仿真,解决了全海深水下机器人缺乏设计母型和特殊外形下仿真精度低的问题。仿真并讨论了超高压(>100MPa)下海水密度变化对下潜阻力产生的影响。由于海沟环境及机器人自身外形的影响,需要具备强机动能力并经常切换运动方向,通过使用四元数方式进行运动学描述解决了常规方法中存在的万向节死锁问题。(2) 基于零进速推力特性的全海深水下机器人实时浮力测量方法。本研究提出了一种适用于全海深水下机器人的实时浮力测量方法,主要解决大水深环境下密度变化、材料力学参数不确定等因素造成浮力计算困难的问题。该方法基于推进器和加速度计进行浮力测量。为了减少相关动力学参数不确定性对测量结果的影响。本研究通过执行一个精心设计的动作,使用“海斗”ARV在近10900m深的海底中进行了实验验证。实验结果表明,该方法测量得到的剩余浮力结果与参考值的平均误差为0.3N。(3) 基于柔性结构的全海深水下机器人外场动力学参数辨识方法。本研究还通过设计特殊的柔性结构实现了在超过一万米的海底测量水下机器人受到的实时合外力,通过对多个传感器信息的融合,可以获得垂直方向上往复运动的高精度动态受力信息。基于该受力数据实现了垂向动力学参数的辨识,并通过实验验证了本方法获得的参数的准确性。(4) 基于岭回归的全海深水下机器人动力学灰箱建模方法。全海深水下机器人大多采用特殊的外形,有些会在运动时自发地周期摇摆。常规方法得到的模型往往是渐进稳定的,无法对水下机器人自发的摇摆状态进行仿真。本研究通过改进岭回归优化算法,可以在粘性力灰箱模型中优化出最优的粘性力系数组合方式,并针对庞大参数矩阵造成的过拟合和收敛性差的问题进行了优化。使用该方法得到了“海斗”ARV无动力下潜6自由度动力学模型,该模型的仿真结果与海上试验中无动力下潜的数据吻合度较高,并在海上试验中较好的预测了水下机器人在摇摆状态下的下潜速度。(5) 全海深水下机器人矢量欠驱动控制及L1自适应运动控制。本研究针对“海斗”ARV特殊的推进器布置和矢量控制的特点设计了基于矢量控制的欠驱动运动控制策略,使其上层控制器可以在常规运动控制通道(航行、升沉、偏航)上进行控制,并对定深控制效果进行了优化。在此基础上设计了基于L1自适应控制器的摇摆抑制控制器,可以大幅减轻下潜过程中的摇摆幅度并提升下潜速度。使用该方法下潜到11000m深度预计可以比常规无动力下潜方式节省能源约38%。该仿真结果也同时验证了本文其他研究内容中方法和结果的有效性。
其他摘要This research was financially supported by the Strategic Priority Research Program of the Chinese Academy of Sciences. This research mainly research on the key and difficult problems encountered in the modeling and control of full-ocean-depth underwater vehicle, and the HADAL Autonomous and Remote operated Vehicle (HADAL ARV) play a role as the experimental platform. The main research contents include the following aspects: (1) Full-ocean-depth underwater vehicle motion simulation and experimental platform. The experimental platform hardware conditions are introduced. Computational fluid dynamics (CFD) simulation of full appendage model solves the problem of low accuracy of simulation withot reference model or with special shape in full-ocean-depth underwater vehicles. The influence of seawater density change under high pressure (> 100MPa) on the dive resistance was simulated and compensated. Due to the influence of the trench environment and the robot's own shape, it needs to have strong maneuvering ability and switch the direction of motion frequently. The problem of universal joint deadlock in the conventional method is solved by using quaternion kinematics description. (2) Real-time buoyancy measurement method for full-ocean-depth underwater vehicle based on zero advance speed thrust characteristic. This study presents a real-time buoyancy measurement method suitable for deep-sea underwater robots, which mainly solves the problem of buoyancy calculation caused by density variations in deep ocean environment, uncertainty of material mechanical parameters, etc. This method measures buoyancy based on propeller and accelerometer. In order to reduce the uncertainty of the relevant kinetic parameters on the measurement results, we used a well-designed motion to verify this method in the depth of nearly 10 900 m using the HADAL ARV. The experimental results show that the average error between the residual buoyancy measured by this method and the reference value is 0.3N. (3) A dynamic mothod for identification of dynamic parameters of full-ocean-depth underwater vehicle based on flexible structure. In this study, we designed a special flexible structure to measure the real-time external force received by underwater vehicle at the seafloor of more than 10 000 meters. Through the fusion of multiple sensors, the force information with high-precision dynamic response of vertical reciprocating motion can be obtained. Based on this, the vertical dynamic parameters are identified, and the accuracy of the parameters obtained by the method is verified by experiments. (4) Ridge regression based full-ocean-depth underwater vehicle gray-box dynamic modeling method. Most deep-sea underwater robots use special shape, some spontaneous swinging during dive process. The model obtained by the conventional method is usually asymptotically stable and cannot simulate the spontaneous swinging state of underwater vehicles. In this study, by optimizing the ridge regression optimization algorithm, the best combination of viscous force coefficient can be optimized in viscous force gray-box model, and the over-fitting and convergence problems caused by the large parameter matrix are optimized. Using this method, the 6-DOF dynamic model of HADAL ARV in free-fall dive process was obtained. The simulation results of this model are in good agreement with the data of no- free-fall dive process in sea trials, and can be a good prediction of the swing state in the sea trials. (5) Full-ocean-depth underwater vehicle vector propeller underactuation control and L1 adaptive motion control. This study based on the unique thruster layout and vector control features of HADAL ARV, an underactuation motion control strategy based on vector control is designed so that its upper controller can be used in conventional motion control channels (sailing, heave, yaw) on the control. The depth control effect is optimized in this controller. On this basis, the design of a swing suppression controller based on the L1 adaptive controller can greatly reduce the swing speed during dive process and improve the dive speed. Desending to a depth of 11 000 m using this method is expected to save about 38% of energy than conventional free-fall dive process. The simulation results also verify the validity of the methods and results in the other studies present above.
语种中文
产权排序1
文献类型学位论文
条目标识符http://ir.sia.cn/handle/173321/21805
专题水下机器人研究室
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
推荐引用方式
GB/T 7714
刘鑫宇. 全海深水下机器人建模与控制研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2018.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
全海深水下机器人建模与控制研究.pdf(12398KB)学位论文 开放获取CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[刘鑫宇]的文章
百度学术
百度学术中相似的文章
[刘鑫宇]的文章
必应学术
必应学术中相似的文章
[刘鑫宇]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。