SIA OpenIR  > 海洋机器人卓越创新中心
水下机器人水动力参数辨识及合成射流水下操纵机理研究
Alternative TitleStudy on the hydrodynamic coefficient identification method of underwater robot and underwater steering mechanism based on synthetic jet
耿令波1,2
Department海洋机器人卓越创新中心
Thesis Advisor林扬
Keyword水动力参数辨识 运动激励优化 神经网络 合成射流 耦合水动力
Pages165页
Degree Discipline机械电子工程
Degree Name博士
2019-05-24
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract面向水下机器人对精确水动力模型及高机动性的需求,本文采用系统辨识、神经网络、CFD、势流理论等方法,从水下机器人本体及合成射流操纵部件两个层面,研究了水下机器人水动力建模方法。详细研究内容如下:(1)为满足水下机器人在不同环境、不同运动方式下对精确水动力模型的需求,研究了基于自适应扩展卡尔曼滤波的水下机器人水动力参数在线辨识方法。针对水动力模型建模误差难以预估的问题,提出了一种多步更新变权重自适应EKF算法;针对微小型水下机器人内部空间有限难以安装声学计程仪,故而线速度难以测量的问题,推导了基于线加速度+角速度的水动力参数辨识方法,通过仿真对比发现,相对于基于速度的辨识,基于加速度的辨识在噪声条件下具有更好的鲁棒性;为提高噪声环境下水动力参数辨识精度,研究了运动激励对水动力参数辨识精度的影响,并提出了一种运动激励优化算法;模拟真实传感器噪声特性,对本文所提出的辨识方法及运动激励优化方法开展了仿真验证,结果表明,采用优化的运动激励,参数辨识误差减小6倍。(2)传统水动力建模方法是基于泰勒展开得到的,仅适用于低速、弱机动。面向高速、高机动水下机器人建模需求,本文研究了基于神经网络的无参数水动力建模方法。比较了不同网络结构,包括全连接网络、径向基网络、模糊网络、小波网络以及模糊小波网络用于水动力建模的精度及泛化能力,结果表明在相同网络规模及训练次数下,非对称型模糊小波网络的建模精度及泛化能力最强;基于非对称型模糊小波网络,研究了不同运动激励对神经网络水动力模型泛化能力的影响,结果表明,采用经过优化的运动激励可以显著提升神经网络水动力模型的泛化能力。(3)面向水下机器人对高机动性的需求,研究了基于合成射流的新型水下仿生操纵技术。面向操纵控制对精确操纵力模型的需求,针对传统模型在高频率、大振幅下操纵力预报精度不足的问题,基于势流理论建立了合成射流操纵力模型,该模型将合成射流操纵力分为三个部分,分别是射流动量引起的操纵力、激励器内外流体加减速引起的惯性力以及激励器喷口处压强力;建立了合成射流时均操纵力模型,与现有模型相比,本文模型引入了动量系数修正因子,其预测结果与实验测量结果具有更高的吻合度;面向激励器效率优化对功率模型的需求,建立了合成射流激励器功率及效率模型,基于该模型对不同条件下激励器效率变化的物理机理进行了阐释;搭建了合成射流推力实验平台,基于数值仿真与实验测量相结合的方法,研究了不同驱动参数、不同腔体构型以及不同激励函数对合成射流激励器推力及效率的影响,为合成射流激励器优化设计指明了方向。(4)合成射流用于水下机器人操纵时,外流场对其推力及效率具有较大影响,这种影响及其作用机制目前没有相关研究,为此本文分两个层次研究了外流场对合成射流水动力性能的影响。研究了外流场对单个激励器的影响,通过对不同外流场速度及激励器不同驱动参数下速度分布、压强分布及涡量场的分析,阐明了外流场对合成射流推力、效率影响机制;研究了合成射流激励器与水下机器人耦合水动力,发现,航速每增加2节,合成射流操纵力衰减约23%,功率消耗增加约5%;本文还提出了一种基于双临近合成射流的矢量操纵技术,通过改变激励器的相位差,可以实现最大约12度的矢量角。本文的研究成果可以为水下机器人精确水动力建模、高机动性实现提供理论及技术支撑。
Other AbstractThe turbulence and ocean current has great impact on the motion of UUV. For some tasks such as underwater 3D restruction, the accuracy of the motion is strongly required. To fight with the disturbance, UUVshould has the ability of accurate motion control and highly mobile. In this thesis, the hydrodynamic modeling method including the vechile hydrodynamic and steering hydrodynamic is studied based on system identification, neural network, computational fluid dynamics and potential flow theory. For vehicle hydrodynamic modeling, online hydrodynamic coefficient identification based on adjustable EKF is studied. And taking into consideration of the model uncertainty, non-parameteric modeling method based on neural network is also investigated. For steering hydrodynamic, a novel steering technique based on synthetic jet is investigated. The thrust of synthetic jet is estabiished using potential flow theory. The mechanism behind thrust generation and high efficiency of synthetic jet is investigated using numerical method. The thrust characteristic of synthetic jet under different actuation and geometrical parameters is also studied. The main contents of this thesis are concluded as follows: (1) To fulfill the demand for accurate hydrodynamic model under different motion, the online hydrodynamic coefficient identification technique based on adaptive EKF is developed. An adaptive EKF algorithm with adjustable learning rate is developed for the computation of the modeling error which is essential for the EKF. Identification method based on acceleration and angular velocity is developed considering the difficulty in measuring velocity underwater. To enhance the accuracy of hydrodynamic identification using data with noise, the effect of motion input on identification accuracy is studied. And an algorithm for roboust motion input design is developed. (2)the hydrodynamic model is easily affected by the motion of the vehicle. The traditional hydrodynamic modeling method only applies to limited motion status. To make the hydrodynamic model more roboust to motion status, a non-parameteric modeling method based on neural network is studied. The performance of different types of neural networks in hydrodynamic modeling is investifated. The influence of motion input on the performance of neural network hydrodynamic model is studied. (3)to fulfill the demand for high mobility, a novel underwater steering method based on synthetic jet is studied. The mechanism behind the thrust generation and high efficiency of synthetic jet is studied using numerical method. The thrust model of synthetic jet is established using potential flow theory. The thrust characteristic of synthetic jet under different actuation and geometrical parameters is investigated and the mechanism behind these thrust characteristics is illustrated using the model established in this thesis. (4)The coupling hydrodynamic of synthetic jet and underwater vehicle is studied at last. The thrust and efficiency of synthetic jet underwater different crossflow velocities is investifated using numerical method. The numerical method is validated using experimental data. The thrust, efficiency and the total steering force under different cruising speed when synthetic jet is usind for underwater vehicle steering is also studied.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25165
Collection海洋机器人卓越创新中心
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
耿令波. 水下机器人水动力参数辨识及合成射流水下操纵机理研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
水下机器人水动力参数辨识及合成射流水下操(28590KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[耿令波]'s Articles
Baidu academic
Similar articles in Baidu academic
[耿令波]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[耿令波]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.