【作者】 朱洪前；

【导师】 胡豁生； 桂卫华；

【作者基本信息】 中南大学 ， 控制科学与工程， 2010， 博士

【摘要】 随着陆地矿产资源的不断枯竭,深入开发各种海洋矿产资源受到世界各国的重视。在浩瀚的海底,富含一种结核状矿石,内有锰、钴、镍、铜、铅、锌、金、银、铝等金属元素,品位非常高。海底采矿机器人需要携带采矿头、动力站、电子设备、输矿软管和电缆等自主行走于稀软的海底,收集各种矿石。深海采矿机器人的作业环境很复杂,如：连接深海采矿机器人与中继矿仓的输矿软管长达数百米,在作业过程中,软管张紧度不断变化,引起软管张力动态变化,中继仓受海浪影响上下起伏,引起软管产生摆动。由于海底稀软,软管阻力将引起履带打滑率及其测量噪声的波动,产生一个复杂动态干扰,作用在深海采矿机器人组合定位系统与轨迹跟踪控制系统,影响到深海采矿效率与深海采矿系统安全。论文针对复杂作业环境下的深海采矿机器人轨迹跟踪所做的主要研究工作和贡献总结如下：1)鉴于海上实验的困难,为了降低系统开发的风险,针对采矿机器人在稀软海底行驶时打滑严重,打滑率难以准确测量等问题,基于悬链线法,通过对超越方程组数值求解,实时计算软管阻力,建立了新的深海采矿机器人动力学模型与履带打滑率计算模型,为组合定位与轨迹跟踪控制系统分析模型的建立奠定了基础。2)针对长基线声学定位受工作噪声干扰以及航位推算精度受打滑干扰等问题,采用附加打滑参数的履带机器人运动学模型,根据湖试数据对定位系统过程噪声与测量噪声进行了描述,利用新息序列实现噪声统计特性自适应,然后考虑测量数据时延带来的影响,通过卡尔曼滤波器将长基线定位信息与航位推算信息进行融合,得到采矿机器人的位置估计。研究结果证实,该自适应卡尔曼滤波器能有效地适应过程噪声与测量噪声统计特性的变化,比常规卡尔曼滤波器具有更好的深海采矿机器人定位效果。3)针对扩展卡尔曼滤波(EKF)在海底采矿机器人组合定位系统中应用时存在着计算复杂、线性化误差大等问题,将无色卡尔曼滤波(UKF)用于基于长基线(LBL)与航位推算(DR)的深海采矿机器人组合定位系统中。考虑测量数据时延的影响,组合定位系统融合LBL与DR信息,得到海底采矿机器人的位置估计。研究结果表明UKF方法能够明显减少组合定位系统的线性化误差,提高海底采矿机器人定位系统的精度与稳定性。4)针对组合定位系统存在的非线性、打滑噪声变化大、声纳检测信息延迟等导致的滤波器性能下降问题,采用了一种新的过程噪声方差自适应调整方法,并将其与UKF相结合,形成自适应无迹卡尔曼滤波(AUKF)。研究结果表明：该方法可以实时调整过程噪声方差,有效地避免由于过程噪声统计特性不准确所带来的滤波性能下降的问题,其深海采矿机器人位置估计效果明显优于常规UKF方法。5)基于视线导航方法,得到期望的深海采矿机器人速度与航向角。考虑履带打滑的影响,并加入一个与航向角误差成正比的分量,建立两侧液压马达期望角速度的计算模型,实现深海采矿机器人的航向角控制。针对液压系统的非线性和不确定性,建立了液压马达角速度控制系统分析模型,提出了参数自整定模糊PID控制方法与在线自学习并积模糊控制方法。研究结果表明：期望马达角速度计算时,履带打滑率补偿精度越高,深海采矿机器人越能够准确地跟踪预定的采矿路径。更多还原

【Abstract】 Among deep-sea mineral resources, manganese nodules (include useful metals such as manganese, copper and nickel) have been the most important targets of ocean exploration because of increasingly shortage of mineral resources and huge commercial benefits. In the deep ocean mining system, a tracked vehicle is normally used to collect manganese nodules on the seabed. The vehicle is equipped with a pickup unit, crusher, hydraulic system, electronic box, flexible pipe, etc. It is designed to move on deep seabed autonomously, with high localization precision, so as to ensure high efficiency operation.This thesis describes the development of a mining vehicle model and the associated navigation system for a skid-steered tracked vehicle that is expected to operate on the deepsea floor. The system is to navigate accurately and reliably in an unstrucutered environment, which is a difficult task and requires accurate localisation and robust trajectory control. The thesis made five main contributions towards accurate and reliable navigation and control of tracked mining vehicles as follows:1) The thesis presents a novel approach to soil-track interaction modelling by incorporating track slips, the robot slip angle and track forces. The equations developed characterise the relationship between the forces acting on the vehicle, the robot parameters and key soil properties. The soil-track interaction equations are then used to develop a comprehensive motion model of a tracked robot. Incorporating kinematic and dynamic equations of robot motion with the soil equations allows robust and reliable estimation of the robot’s position using an extended Kalman filter.2) Since the long base line (LBL) based sonar localization system of seabed mining vehicles is seriously affected by the noise in a working environment, and the accuracy of dead reckoning (DR) is seriously affected by vehicle slippage, the thesis introduces a kinematic model of a tracked mining vehicle in presence of sliding parameters, and describes its process and measurement noises based on experiment data collected from a lake. The innovation sequence is deployed to achieve the adaptive statistics features of both process and measurement noises. Taking into account the influence of measurement data delay, the Kalman filter fuses LBL and DR data to obtain the localization estimate of the seabed mining vehicle. Simulation results prove that the adaptive Kalman filter can deal with the changing statistics features of process and measurement noise very well, and has better localization estimation of the seabed mining vehicle than a normal Kalman filter.3) Since there are some defects when the extended Kalman filter (EKF) is applied in the nonlinear state estimation, unscented Kalman filter (UKF) is introduced in an integrated navigation system (LBL/DR) for the localization of a deep-sea mining vehicle. Compared with the EKF, the UKF not only improves the location accuracy, but also avoids the calculation burden of Jacobian matrices. This data fusion algorithm is easy to implement, and meets the requirements of low cost and high precision. The simulation result shows that the UKF method is more accurate and reliable than the EKF method for deep-sea mining vehicle navigation.4) This thesis introduces a kinematic model of a deep-sea mining vehicle in presence of sliding parameters. The model describes both the noise features of sliding parameters and the deep-sea condition features. To handle sliding parameters noises, a recursive algorithm is adopted to minimize difference between the filter-computed and the actual innovation covariance, which is a novel integrated navigation method based on unscented Kalman filters (UKF). Taking into account the influence of measurement data delay, UKF fuses the data from both LBL sonar localization and DR to perform the state estimation. Simulation results show that the adaptive UKF has better localization performance than a normal UKF for a deep-sea tracked vehicle (DTV).5) Based on the line of sight navigation method, the desired speed and heading angle of the mine collection vehicle are obtained. Considering the influence of track sliding, a sub-factor that is proportional to the heading error is added, a desired angular velocity model of left and right hydraulic motors set up, the heading angle control of the mine collection vehicle is achieved. Considering the non-liner and uncertain features of a hydraulic system, the analysis model of the angular velocity control system of a hydraulic motor is built and a fuzzy-PID control algorithm is proposed. The simulation results show that:the proposed system can effectively compensate track slip, resist the effect of various disturb factors, and track a desired mining path accurately.更多还原