复杂海洋环境中水下机器人控制若干问题研究
文献类型：学位论文
| 作者 | 吴宝举
|
| 学位类别 | 博士 |
| 答辩日期 | 2010-05-26 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 王晓辉 ; 李一平 |
| 关键词 | 水下机器人 系统辨识 鲁棒控制 自适应控制 干扰观测器 |
| 其他题名 | Study on Control Problems for Unmanned Underwater Vehicles in Complicated Oceanic Environment |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 复杂海洋环境是指近海面或近海底，流、浪等作用强烈的环境。这种条件下水下机器人受到自身和外部环境的各种不确定性因素的干扰，主要包括：模型自身扰动，如模型本身的非线性、水动力参数的时变以及机械手等作业工具引起的重心变化；环境扰动，如海流、海浪干扰。不确定性可归纳为结构不确定性和非结构不确定性。结构不确定性主要是线性参数时变，可以利用自适应控制理论解决，而非结构不确定性对于自适应控制是无能为力的，甚至有可能导致自适应控制失去稳定性。由于水声传输速度慢，机器人与操作者之间无法建立实时的联系，这就要求在复杂海洋环境中工作的水下机器人具有一定的自主决策和判断能力。水下机器人在接近海底航行时还可能受到海底障碍物的威胁，所以水下机器人应具有对复杂海底地形的识别能力。所以，实现复杂海洋环境中水下机器人控制需要解决三方面问题，一是从运动控制的角度，要求水下机器人在各种不确定的干扰作用下具有鲁棒性；二是水下机器人需要具有一定的自主决策能力；三是水下机器人对复杂地形的避碰能力。本文针对以上三方面问题开展研究，在运动控制方法上采用以自适应控制和滑模控制为基础控制方法，利用干扰观测器来估计非结构不确定干扰，形成一种鲁棒的自适应控制方法。针对水下机器人在复杂环境中的使命规划和自主决策问题研究了基于事件驱动的控制方法。对于复杂海底地形的避碰问题本文研究了采用前视声纳获取障碍物信息，把反应式避碰和慎思式避碰相结合的避碰策略。具体研究工作如下： (1) 首先对水下机器人进行受力分析，建立水下机器人的数学模型。分析水下机器人受的干扰力和力矩。对数学模型进行简化，建立离散数学模型。 (2) 推导出系统辨识的数学模型和简化的水下机器人物理变量之间的关系。介绍了常用的系统辨识方法，并指出这些方法在实际应用中存在的问题。分析了参数离线辨识和在线辨识的问题。分析了系统辨识参数的有效性判断依据。对系统辨识的鲁棒性问题进行研究。对“北极ARV”的航向回路和升沉回路进行辨识，并验证了模型的有效性。 (3) 传统的水下机器人运动控制系统大多采用分段PID方法，这种方法对参数变化的适应性不强，受到干扰时准确性和鲁棒性都不高。本文通过系统辨识的方法实时辨识系统的模型参数，按照预定的性能指标调整PID控制器的参数。由于系统存在外部干扰和传感器噪声的影响，在系统辨识部分采用死区修正的方法来解决辨识参数收敛问题，对控制率加入鲁棒补偿项。由于“北极ARV”具有的延迟问题，本文把预测控制应用到控制器设计当中，取得了比较好的效果。对于水下机器人受到的不确定干扰问题，本文利用干扰观测器来估计干扰，把观测值前馈到控制输入端，引入等量的补偿，实现对干扰完全抑制。同时利用滤波器来消除观测噪声对系统的影响。 (4) 滑模变结构控制对干扰有较强的抑制作用，能够消除未建模动态和外界干扰对系统的影响，但是滑模控制作为鲁棒控制的一种需要对干扰的幅值进行估计，若估计的不足会导致鲁棒性下降，估计过大控制器的输出会饱和，通常在设计过程中以估计的最大幅值作为设计依据。本文采用干扰观测器来实时估计干扰的幅值，通过估计值来调整滑模控制器的开关增益，这种控制方法的设计很简单并且对干扰具有适应性，形成一种自适应的滑模控制方法。 (5) 研究了水下机器人圆弧航路规划和跟踪算方，通过直线和圆弧航路使机器人在水平面可以走出非常复杂的轨迹。研究了基于事件驱动的使命规划方法，利用事件驱动，水下机器人可以在复杂的海洋环境中执行复杂的任务和对环境做出智能的反应。 (6) 研究基于前视声纳信息的水下机器人避碰问题。对于避碰方法主要分为两类：反应式避碰和慎思式避碰。反应式避碰方法是一种低层次的快速避碰策略；慎思式避碰是高层次的仿人行为的避碰方法。本文把反应式避碰方法和基于专家经验的慎思式避碰方法相结合，形成一种简单实用的避碰策略。 |
| 索取号 | TP242/W81/2010 |
| 英文摘要 | Complicated oceanic environment means that the vehicle is operated near the surface or near the bottom. Precise and stable control of underwater vehicles in complicated oceanic environment becomes one of the most challenging problems. Major facts that make it difficult to precisely and stably control underwater vehicles are as the followings: firstly, the disturbances come from the vehicles dynamics itself, including the intrinsically nonlinearity of the dynamics, time-vary and unknown hydrodynamics, the variability of the centers of gravity and buoyancy due to the motion of manipulator or payload change during the operation; secondly, the disturbances come from environment, such as the uncertainty of sea current, lopsided force due to the variety of temperature and salinity of water. It is difficult to achieve high performance by using a conventional control algorithm. Three problems have to be considered when designing the controller. Firstly, the controller should have the ability to learn and adapt to the changes of the dynamics of the vehicle and its environment. Secondly, the vehicle should have the ability to deal with complicated situations. Thirdly the vehicle should have obstacle avoidance system. The research of this paper is based on the problems mentioned above. The main research problems are as the followings: (1) An input/output-based simple discrete-time underwater vehicle model is deduced firstly. (2) An advanced algorithm is proposed according to the scheme of adaptive control. The parameter identification algorithm is used to deal with any changes of vehicle and environment disturbance taking advantage of its online learning ability. The online identifier is based on a recursive prediction error algorithm. The introduction of dead-zone term during training makes the algorithm more robust than the standard recursive least-squares method. Its performance is verified through numerical simulations. 3) Unstructured disturbance may reduce the performance even destroy the stability. An observer scheme is proposed in this thesis to solve the unstructured problem. In this control scheme, the observer utilizes the information from the sensors and the data calculated from the model for reconstructing the disturbance signal and a compensator is used to reduce the influence of the disturbance. (4) A sliding-mode controller is designed to compensate disturbance. The stable conditions and attraction region of the sliding-mode control scheme can be regulated by parameters obtained by identification algorithm. The range of the disturbance is estimated by an observer .Then the range is used to tune the coefficients of the sliding-mode controller. Effectiveness of this control scheme is demonstrated by numerical simulation and pool experiment. (5) In order to solve more complicated task an FSM(finite state machine)is used for modeling the task，and based on the FSM method，an event-based coordinating control strategy is provided for improving flexibility and robustness of the system．The experiment results show the feasibility and effectiveness of the design． (6) The forward looking sonar is used to identify obstacle information in complicated oceanic environment．The reactive avoidance method and artificial potential method is integrated to the avoidance planning system. Reactive avoidance method is very efficiency when the environment is simple. When the situation becomes complicated UUV can cope with it in virtue of artificial intelligence. The hybrid method make use of the merits of the two methods. When facing complicated obstacles under uncertain environment，UUV can complete obstacle avoidance effectively．The validity and feasibility were verified through simulation test. |
| 语种 | 中文 |
| 公开日期 | 2012-07-27 |
| 产权排序 | 1 |
| 分类号 | TP242 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/9265]  |
| 专题 | 沈阳自动化研究所_水下机器人研究室 |
| 推荐引用方式 GB/T 7714 | 吴宝举. 复杂海洋环境中水下机器人控制若干问题研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2010. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
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。