SIA OpenIR  > 水下机器人研究室
深海漫游者机器人关键技术研究
Alternative TitleStudy on the Key Technologies of Deep-Sea Rover Underwater Vehicle
张运修
Department水下机器人研究室
Thesis Advisor张艾群 ; 张奇峰
Keyword深海漫游者机器人 深海履带式机器人 动力学建模 自主回坞 组合导航
Pages169页
Degree Discipline机械电子工程
Degree Name博士
2021-05-19
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract在二十世纪70年代以来,有缆遥控水下机器人(Remotely Operated Vehicles, ROV)以其海底强机动性、强作业能力的特点,被广泛应用于海洋石油天然气开发和海洋科考领域。然而,近年来随着人们对深海领域的不断探索与开发,使得ROV的深海应用逐渐受到其传统有缆作业模式的制约,包括海况条件制约、成本和技术制约以及无法完成长时间的坐底作业任务等。驻留式ROV(Resident ROV)概念的提出在一定程度上突破了传统ROV受海况条件、作业时长的限制,然而,相对传统的ROV系统,驻留式ROV的遥控回坞方式、依靠推进器悬停定位或坐底作业的模式并没有改变,不具备在复杂地形条件下爬行和定点长时作业的能力。本文借鉴太空探测漫游者机器人与太空着陆器联合作业的理念,提出一种基于深海着陆器作业的新型有缆水下漫游者机器人概念,即深海漫游者机器人(Deep-sea Rover ROV,简称R-ROV)。R-ROV具备浮游和履带爬行两种模式,综合了传统ROV机动性强和履带式爬行机器人地面适应性强的优点,克服了传统ROV和驻留式ROV无法应对长时复杂底质环境下精细探测作业的缺点,并仅基于引导声纳作为定位传感器使R-ROV具备了100 m范围内的自主回坞能力。作业时,R-ROV以一种新型的深海探测装备——深海多位点着陆器“鹿岭”号作为海底基站,规避了传统ROV需要布放回收系统实时支持的限制,降低了母船的要求,并拓展了各自的探测和作业能力,构成了一种新型的深海装备和“断线区域”探测作业模式。本文结合国家重点研发计划“深海关键技术与装备”重点专项课题“漫游者潜水器技术研究(编号:2017YFC0306402)”的需求,针对R-ROV基于深海着陆器作业时在深海底质环境下的运动适应性和导航回坞若干关键问题进行深入研究,主要包括以下内容:1、R-ROV爬行模式运动机理与参数优化研究。采用椭圆原理设计了基于可伸缩摆臂的自张紧履带构型,分析了R-ROV爬行运动机理,并从目标约束和地面行走适应性角度推导了多目标优化数学模型。根据遗传算法、组合赋权求解后的设计参数,开展了R-ROV多体动力学仿真,证明了R-ROV爬行模式对沟壑和台阶典型障碍具有较强的通过性,此外还对双摆臂姿态对R-ROV形走过程中履带沉陷量和行驶阻力的影响进行了仿真验证。2、R-ROV爬行模式转向动力学建模与分析。基于履带剪切应力-剪切位移关系建立了包含履带滑移参数、水阻力以及浮力影响的R-ROV海底松软地面下稳态转向动力学模型,从理论上获得了R-ROV在海底松软地面爬行时的转向阻力矩、履带牵引力以及履带的打滑率预报;影响R-ROV转向性能的参数主要包括地面土壤条件、转向半径和前进速度等,给出了R-ROV转向运动的限定条件。通过动力学仿真实验得到的数据与理论计算结果具有较高的一致性。3、 基于声学引导的自主回坞技术研究。面向R-ROV基站式近距离自主回坞的需求,区别于传统AUV采用远距离辅以声学、近距离光学的双定位引导源自主回坞方案,本文采用回坞引导声纳作为单一引导源和低成本惯性传感器,构建了一套具有自适应容错能力的高性价比R-ROV组合导航系统。为解决回坞引导声纳数据包含野值、短暂偏置以及非连续等问题,本文提出了一种基于新息修正的改进Sage-Husa 自适应平方根容积卡尔曼滤波算法,仿真验证了该算法的优越性。基于变深度视距制导回坞控制策略,采用研制的浮游模式R-ROV作为试验平台成功开展了多组自主回坞功能试验,验证了本文基于单一声学引导自主回坞技术的有效性。4、基于双神经网络辅助的无缝导航方法。在前文提出的自适应容错滤波器的基础上,运用长短期记忆网络(Long Short-Term Memory,LSTM)构建了一种引入自学习能力的双神经网络辅助优化卡尔曼滤波器,解决声学野值、短暂偏置等故障的同时,可以有效解决R-ROV航行时遇到的长时间(60 s)声学定位数据中断导致导航定位发散的问题。通过引入深度神经网络来学习惯性传感器、滤波器内部增益和R-ROV位置增量之间的模型,以补偿定位信息失锁时惯性系统的误差;同时,增加了一个深度学习网络进行最后的估计误差补偿。本文所提算法体现了对卡尔曼滤波器内部和外部的双重优化,最后基于R-ROV的外场试验数据进行了多组对比分析,证明了所提方法的优越性。5、平台搭建与试验研究。根据R-ROV的运动实现机理,在本文搭建的模块化浮游模式R-ROV平台基础上,搭建了爬行模式R-ROV试验平台。在海上试验中,针对R-ROV系统建模、仿真分析的结果进行了验证。最后介绍了在中国南海“海马”冷泉区域和“深海勇士”号载人潜水器联合科考作业的情况。相关数据和应用成果表明,本文提出并研制的深海漫游者机器人系统具有广阔的应用前景。
Other AbstractRemotely Operated Vehicle (ROV) plays an important role in the exploration of the ocean, and it has been widely used in oil & gas (O&G) and marine scientific research since the 1970s due to their strong subsea maneuverability and operation capabilities. However, with the exploring and developing of the deep-sea in recent years, the deep-sea application of ROV is gradually restricted by its traditional cable operation mode, including the harsh deep-sea conditions, cost and technology laminations, as well as the inability to complete the long-time bottom operation task. Recently, a new kind of ROV (resident ROV) breaks through the limitation of traditional ROV restricted by the deep-sea conditions and operation duration to a certain extent. However, compared with the traditional ROV system, the mode of remote-control docking, hovering positioning by thruster and bottom operation of the resident ROV are not changed. It means that the resident ROV does not have the ability of crawling and fixed-point long-time operation under a complex terrain condition. In this thesis, a new concept of underwater ROV system based on a deep-sea lander operation is proposed, namely Deep-Sea Rover ROV (R-ROV). It draws on the concept of joint operation of the space exploration Rover robot and the space lander. R-ROV has two motion modes: the floating mode and the crawler crawling mode, which combines the advantages of the strong mobility of the traditional ROV and the strong adaptability of crawler crawling robot on the ground. The R-ROV overcomes the shortcomings that traditional ROV and resident ROV cannot detect in complex sediment environments for a long time. Meanwhile, the R-ROV has the autonomous docking capability within 100 m only based on the guidance sonar as a positioning sensor. During the operation, the R-ROV uses novel deep-sea exploration equipment, deep-sea multi-site lander which is named “Luling”, as the seabed base station. It simplifies the operating conditions of ROV and Launch and Recovery System (LARS). Additionally, it reduces the requirements of the mother ship and expands the exploration capabilities in deep-sea environment. The R-ROV system and the deep-sea multi-site lander constitute a novel deep-sea equipment and “broken line zone” detection operation mode. With the requirements of the project “Research on deep-sea Rover vehicle Technology (No.: 2017YFC0306402)”, this thesis conducted an in-depth study on several key issues concerning the motion adaptability for the deep-sea bottom environment and autonomous docking to the deep-sea multi-site lander. This work was supported by the key special project “Deep-Sea Key Technology and Equipment” of the National Key Research and Development Project. The main contributions of this thesis are: 1. Study on motion mechanism and parameter optimization of the R-ROV. A tensioned track configuration based on the retractable swing arm is designed by using ellipse principle. The crawling motion mechanism of R-ROV is analyzed, and a multi-objective optimization mathematical model is derived from the perspective of target constraint and ground walking adaptability. The multibody dynamics simulation of the R-ROV is carried out based on the design parameters results which are solved by the genetic algorithm and combined empowerment. The results prove that the R-ROV in crawling mode has strong trafficability for typical obstacles of gullies and steps. In addition, the influences of the angle of the double swing arm on the track settlement and driving resistance during the R-ROV walking were simulated and verified. 2. Steering dynamics modeling and analysis of the crawling mode R-ROV. The steady-state steering dynamic model of the R-ROV under the deep-sea soft soil bottom is established by using the track shear stress-shear displacement relationship. The dynamic model includes the influence of track skid parameters, water resistance and buoyancy. Based on the theoretical analysis, the steering resistance moment, traction force and skid rate prediction of the R-ROV crawling on the soft ground of the seabed are obtained. The parameters that affect the steering performance of the R-ROV mainly include ground soil conditions, steering radius and forward speed, and the limiting conditions of the steering motion of the R-ROV are given in this thesis. The data obtained from the dynamic simulation experiments is in good agreement with the theoretical calculation results. 3. Acoustics-based autonomous docking for the R-ROV. To demand the R-ROV short-range (< 100m) autonomous docking, this thesis uses a docking guidance sonar and low-cost inertial sensor to build an integrated navigation system that with adaptive error tolerance and high-cost performance. It is different from the traditional AUV’s docking scheme, which adopts the acoustic and short-range optical to consist a dual positioning guidance method. In order to solve the high field value and discontinuous characteristic problems of the guided sonar, a modified Residual-based Sage-Husa Adaptive Square Root Cubature KF algorithm, called RSHA-SRCKF, is proposed to fuse the heterogeneous information from a small docking guidance sonar and low-cost inertial sensors. The simulation results verify the superiority of the combined navigation algorithm. Field experiments were conducted successfully based on the proposed variable depth and line of sight guidance docking control strategy, which verify the effectiveness of the acoustics-based autonomous docking technology presented in this thesis. 4. Seamless navigation method based on double neural network. Based on the adaptive fault-tolerant filter proposed in the previous chapter, a double neural network aided Kalman filter with self-learning ability is constructed by using Long Short - Term Memory (LSTM). While solving the acoustic outliers, short offset and other fault data, it can effectively solve the problem of navigation divergence caused by long-time (60 s) acoustic positioning data interruption during the R-ROV navigation. The deep neural network is introduced to learn the relationship between the inertial sensor data, the internal gain of the filter and the R-ROV position increment, so as to compensate the errors of the inertial system when the acoustic positioning information is lost. At the same time, an additional deep neural network is added to compensate the final estimation error. The algorithm proposed in this paper embodies the internal and external optimization of the Kalman filter. Finally, several groups of comparative analysis prove the superiority of the proposed method based on the field test data of the R-ROV. 5. Study on platform construction and field experiments. On the basis of the modular floating mode R-ROV platform built above and according to the analysis of the motion mechanism of the R-ROV, a crawling mode R-ROV platform is built in this chapter. The results of R-ROV system modeling and simulation analysis are verified in the field tests. Finally, the paper introduces the joint scientific research work with the “Shen Hai Yong Shi” HOV (Human Operated Vehicle) in the cold spring area of “HaiMa” in the South China Sea. Relevant data and application results show that the deep-sea Rover ROV system developed in this thesis has broad application prospects.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/29005
Collection水下机器人研究室
Recommended Citation
GB/T 7714
张运修. 深海漫游者机器人关键技术研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2021.
Files in This Item:
File Name/Size DocType Version Access License
深海漫游者机器人关键技术研究.pdf(10516KB)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
[张运修]'s Articles
Baidu academic
Similar articles in Baidu academic
[张运修]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[张运修]'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.