水下滑翔器惯性组合导航定位关键技术研究

发布时间：2018-01-20 10:27

本文关键词： 惯性导航系统误差校正与补偿姿态与位置解算数据融合滤波　出处：《东南大学》2015年博士论文　论文类型：学位论文

【摘要】：随着水下潜器技术的日益成熟,水下滑翔器作为一种新型且重要的水下潜器受到越来越多的关注。水下滑翔器在海洋工程应用方面发挥着重要作用,尤其是低功耗、长航时、小体积等特点更使其成为目前研究的热点。准确的位姿信息对滑翔器长时间水下作业起着必不可少的作用,因此持续高精度导航定位是水下滑翔器研究的关键技术之一借助水流和自身调节作用顺水而滑,几乎不需外界提供能源是水下滑翔器有别于其它水下潜器的重要特点,故其具有重要的应用领域和实用价值。低功耗使得滑翔器本体设计简单且体积小,所搭载的导航装置数量尽量少。作为陆地成熟且定位精度很高的全球定位系统(Global Positioning System, GPS)不能应用于水下,有自主导航解算能力的惯性导航系统(Inertial Navigation System, INS)成为替代GPS的较优选择。INS通过自身的陀螺仪、加速度计等传感器测量载体当前时刻的旋转角速度和线加速度,测得的数据通过积分等运算得到载体的姿态、速度及位置,但INS的测量误差会随着时问而积累,单独长时间工作会严重降低导航精度。应用于水下滑翔器的惯性导航系统,因成本和体积的限制只能选用微机电系统(Micro-Electro-Mechanical System, MEMS)惯性测量单元(Inertial Measurement Unit, IMU),该惯性测量单元的误差和随机漂移更加明显。有低功耗、长航时、小体积及低成本等条件的制约,加之导航传感器精度低且外界辅助导航少,要实现高精度高可靠性的水下导航与定位是目前研究的重点也是难点。在参考国内外大量低精度惯性导航元件实现高精度长航时导航文献的基础上,对水下滑翔器运动模型进行分析,建立航位推算(Dead Reckoning, DR)模型并与惯性导航组合,设计出应用于水下滑翔器的导航系统,对惯性测量单元误差模型深入分析并针对不同的导航传感器进行误差校正与补偿,提出应用于水下环境的高精度高可靠性高效的数据融合算法,使该系统及算法更适合实际工程应用。论文的主要工作和创新点如下：(1)对水下滑翔器运动模型进行分析,建立航位推算模型并与惯导系统组合,完成用于水下滑翔器的导航系统设计。针对水下实际应用环境,结合滑翔器自身特点,详细分析水下滑翔器运动模型,进行航位推算并与惯性导航系统组合,对惯性导航进行辅助及校正。以低功耗MEMS惯性元件为主体设计出适用于水下滑翔器的小体积低成本导航系统,满足水下长航时、低功耗的工作要求,并能提供高精度高可靠性的导航与定位信息。另外,在水下工作一段时间后滑翔器浮出水面,该系统能智能接收GPS或其它卫星信号对导航信息进行更新。(2)对MEMS惯性传感器及磁传感器误差进行建模,并针对不同的传感器误差模型提出相应的校正与补偿方法。MEMS惯性测量单元的零位漂移、刻度因子和安装误差等是准确建立误差模型的重要参数。设计静态八位置实验,利用多位置对称测量原理确定陀螺仪和加速度计的零位漂移,通过陀螺仪速率实验计算陀螺仪各轴刻度因子及安装误差。本论文提出加速度计动态测量方法,相比传统的静态多位置法更实时有效方便地计算出加速度计的刻度因子和安装误差。由系统误差模型推导误差补偿模型,代入相关误差参数分别对陀螺仪和加速度计误差进行补偿。处理磁传感器采集的磁场信息是将原始含有软硬磁等干扰信息的磁场去噪后拟合三维椭球磁场,得到拟合参数对磁传感器输出进行校正与补偿,提高地磁场的测量精度。(3)针对实际非线性模型,为了提高导航信息的估计精度提出将扩展卡尔曼和龙格库塔法融合的滤波算法,在此基础上为进一步提高精度及适用范围,提出平滑参数的无迹卡尔曼滤波算法。滑翔器在一定深度的水中滑翔,水环境相对稳定,在较长的滑翔时间内可将非线性模型划分成若干线性模型处理,基于这样的思想可简化建模,在不太增加算法复杂性及计算量的情况下达到提高精度的目的。本文提出将扩展卡尔曼(Extended Kalman Fiter, EKF)与龙格库塔法(Runge-Kutta,RK4)相结合的EKF/RK4滤波算法并应用于实际水下导航系统,EKF在非线性程度不太高的系统中优势较明显,再结合高精度数值计算方法(龙格库塔法),实验表明比起传统的EKF和UKF,EKF/RK4数据融合算法能有效减小误差,姿态角和位置估计精度也显著提高。为了扩大算法的适用范围并进一步提高估计精度,鉴于无迹卡尔曼滤波(Unscented Kalman Filter, UKF)在更多的非线性模型中的性能优势提出平滑参数的无迹卡尔曼滤波算法(Smooth Variable Unscented Kalman Filter, SVUKF)。基于UKF进行参数平滑处理使非线性系统的最优估计更加精确,经实验验证,算法的估计精度及稳定性得到有效提高,并且算法的鲁棒性也有所增强。(4)对较复杂的系统模型提出改进高斯混合粒子滤波算法(Improved Gaussian Mixture Particle Filter, IGMPF)并用实验对其性能进行验证：以解决实际问题为目的,权衡精度计算速度及可靠性等多项评估标准,综合比较本论文提出的所有算法,提出高效高精度回溯解耦自适应扩展卡尔曼滤波算法(Back Decoupling and Adaptive Extended Kalman Filter, BD-AEKF)。水下滑翔器在滑翔过程中不排除在个别时刻或区域出现环境突变且含有非高斯噪声的情况,这种较复杂的环境可能会使UKF算法表现欠佳,本文基于粒子滤波对系统进行混合高斯建模并提出改进高斯混合粒子滤波算法,通过和其它算法对比可得,IGMPF算法的姿态角和位置估计精度有所提高,但实时性变差计算速度下降,以此为代价来提高精度在实际应用中可能并不能广泛使用。另外,对于惯性元件,其安装轴和参考轴之间的安装误差是不可避免的,这种固有误差将导致三个姿态角(航向角、俯仰角和横滚角)之间存在交叉耦合,造成姿态角解算不准甚至错误,当俯仰角或横滚角发生变化时,这种现象变得更加明显。俯仰或横滚运动在滑翔器滑翔过程中很常见,交叉耦合导致的姿态角解算误差将不断出现并积累。针对这一实际问题提出BD-AEKF算法,用回溯解耦算法判断解算错误的节点,然后消除姿态角之间的交叉耦合；自适应扩展卡尔曼滤波实时自适应调节导航参数,对滤波输出进行平滑,最终达到消除耦合提高解算精度及稳定输出的目的。对于实际应用问题,需要权衡估算精度、计算速度、算法复杂度、鲁棒性及可靠性等多项评估标准来选择较优的算法,综合比较本论文提出的所有算法,可看出BD-AEKF估算精度比较高且解算速度并不慢,是实际应用中较可行的高效高精度算法。
[Abstract]:With the underwater vehicle technology has become more sophisticated, underwater glider is a new and important underwater vehicle has attracted more and more attention. The underwater glider plays an important role in ocean engineering applications, especially for low power consumption, long endurance, small size and other characteristics has become a hot topic at present. Accurate position information plays an essential role in the glider long time underwater, so continuous high precision navigation is one of the key technologies of underwater gliders with water and their own regulation of smooth and slippery, almost without external energy is glider has an important characteristic different from other underwater vehicles under the water, so it has important applications in the field of low power consumption and practical value. The glider body has the advantages of simple design and small size, the number of navigation devices are equipped with as little as possible. As the land is mature and positioning precision The global positioning system is very high (Global Positioning System, GPS) can not be applied to underwater inertial navigation system of autonomous navigation capability (Inertial Navigation System, INS GPS) as an alternative to the optimal choice of.INS through its own gyroscope, accelerometer and other sensors for measuring the angular velocity and the vector of the current line the acceleration, the measured data obtained by the carrier's attitude integral calculations, speed and position, but the measurement error of INS with time and the accumulation of long time working alone would seriously reduce the navigation accuracy. The inertial navigation system is applied to the underwater glider, due to limited cost and volume selection of MEMS (Micro-Electro-Mechanical System MEMS) inertial measurement unit (Inertial Measurement Unit, IMU), the error of the inertial measurement unit and the random drift is more obvious. With low power consumption, long endurance, body Restrictive conditions of product and low cost, and low precision navigation sensor and navigation outside less, to achieve high precision and high reliability of the underwater navigation and positioning are the focus of the present study is difficult. On the basis of a large number of domestic and foreign low precision inertial navigation components to achieve high precision long voyage navigation on the literature of gliding the motion model of underwater is analyzed, the establishment of the dead reckoning (Dead Reckoning, DR) model and combined with inertial navigation, design a navigation system for underwater glider, and according to the in-depth analysis of different navigation sensors for error correction and compensation of the error model of inertial measurement unit, the data of high precision and high reliability and efficient used in underwater environment fusion algorithm, the system and the algorithm is more suitable for practical engineering application. The main work and innovation are as follows: (1) the movement model of underwater glider Type analysis, the establishment of the dead reckoning model and inertial navigation system, complete the design of navigation system for underwater gliders. The underwater glider combined with the actual application environment, its own characteristics, the glider motion model is analyzed in detail under water, and for dead reckoning and inertial navigation system combination, auxiliary and correction of inertia navigation. With low power consumption MEMS inertial components as the main design of small size and low cost navigation system for underwater glider, meet the water under long endurance, low power requirements, and can provide navigation and positioning information of high precision and high reliability. In addition, work for a period of time in the underwater glider surfaced, the system can receive GPS or other intelligent satellite signal of the navigation information is updated. (2) to model the error of MEMS inertial sensors and magnetic sensor, and different sensor error model is proposed The corresponding correction of zero drift and compensation method of.MEMS inertial measurement unit, scale factor and installation error is an important parameter for accurate error model is established. The eight experimental design of static position, determine the zero drift of gyroscope and accelerometer using the principle of symmetrical position measurement, through the experimental calculation of each axis rate gyro gyroscope scale factor and installation error. This thesis puts forward the dynamic measurement method of accelerometer, compared with many traditional static method is more effective in real time convenient to calculate the scale factor and installation error of accelerometer. The system error model error compensation model, substituted the relevant parameters on the error of gyro and accelerometer error compensation. The magnetic field information acquisition and processing of magnetic sensor is the original with the soft and hard magnetic interference information of the magnetic field after denoising fitting ellipsoid fitting parameters on the magnetic field. The output of the magnetic sensor calibration and compensation, improve the measurement accuracy of the geomagnetic field. (3) according to the actual nonlinear model, in order to improve the estimation accuracy of the navigation information of the extended Kalman filtering fusion algorithm Calman and Runge Kutta method, on this basis to further improve the accuracy and applicability of the proposed filtering algorithm, unscented Calman smoothing parameters. The glider gliding in certain depth of water, the water environment is relatively stable, in a long period of time can be gliding nonlinear model is divided into several linear models, this idea can be simplified based on the modeling, to improve the precision of algorithm complexity and increase in less computation time. In this paper, the extended Calman (Extended Kalman Fiter, EKF) and Runge Kutta method (Runge-Kutta, RK4) combined with EKF/RK4 filtering algorithm and applied to the underwater navigation system, EKF in nonlinear The system is not too high in the obvious advantages, combined with high precision numerical method (Runge Kutta method), experiments show that compared with the traditional EKF and UKF, EKF/RK4 data fusion algorithm can effectively reduce error, attitude angle and position estimation accuracy is significantly improved. In order to expand the scope and improve the estimation algorithm in view of the unscented filtering precision, Calman (Unscented Kalman Filter, UKF Calman) unscented filtering algorithm performance advantages in nonlinear model more in the smoothing parameter (Smooth Variable Unscented Kalman Filter, SVUKF UKF). The optimal smoothing parameter estimation for nonlinear system based on more accurate and verified by experiment, and the estimation accuracy the stability of the algorithm effectively improve the robustness of the algorithm, and has also been enhanced. (4) proposed Gauss hybrid particle filter algorithm for the system model is complex (Improved Gaussian Mixture Particle Filter, IGMPF) and its performance is demonstrated by experiments: in order to solve practical problems for the purpose of balancing precision calculation speed and reliability evaluation standard, all the algorithms proposed in this paper, a comprehensive comparison, put forward a high precision backtracking adaptive decoupling extended Calman filter algorithm (Back Decoupling and Adaptive Extended Kalman Filter, BD-AEKF). The underwater glider does not exclude the environmental catastrophe and contains non Gauss noise in the individual time or area in the gliding process, the complicated environment may cause the poor performance of the UKF algorithm, the particle filter for mixed Gauss model of the system and put forward the improved Gauss hybrid particle filter algorithm based on the available by comparing with other algorithms, IGMPF algorithm's position and attitude estimation accuracy is improved, but the real-time variation calculation This rate of decline, in order to increase the accuracy in practical application and can be widely used. In addition, the inertia element, its installation installation error between the shaft and the reference axis is inevitable, this error will lead to three attitude angle (heading angle, pitch angle and roll angle) cross coupling between. Not even cause the attitude calculation error, when the pitch or roll angle changes, this phenomenon becomes more and more obvious. The pitch or roll movement is very common in the glider gliding process, cross coupling leads to the attitude angle calculation error will appear and accumulation. BD-AEKF algorithm is proposed to solve this problem. Backtracking algorithm to determine the error decoupling solution node, and then eliminate the cross coupling between attitude angle; adaptive extended Calman filter adaptive real-time navigation parameters, smoothing filter output, the most The final solution to eliminate the coupling to improve the accuracy and stability of output. For practical problems, need to weigh the estimation accuracy, computing speed, complexity, robustness and reliability evaluation criteria to select the optimal algorithm, the algorithm proposed in this paper is a comprehensive comparison, shows relatively high estimation accuracy and BD-AEKF the calculating speed is not slow, is a high precision algorithm is feasible in practical application.

【学位授予单位】：东南大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U666.1

【相似文献】