【作者】 张强；

【导师】 孙尧；

【作者基本信息】 哈尔滨工程大学 ， 导航、制导与控制， 2011， 博士

【摘要】 本文研究了小型AUV水下组合导航系统所涉及的几个关键技术：导航器件误差参数的辨识与滤波、航姿参考系统的姿态解算以及组合导航系统的信息融合策略。小型AUV的组合导航系统由于受到艇体体积和成本的限制,往往选用体积小、成本低、功耗小的MEMS惯性器件,以及体积和功耗均较小的Doppler测速仪。这些导航器件虽然能够满足小型AUV的机械与电气特性要求,但是往往测量精度低。采用上述导航器件构成的组合导航系统不但定位精度低,甚至会影响AUV的制导与控制系统的稳定性。论文的前半部分主要就是针对小型AUV采用的导航器件上述问题展开研究。首先在对MEMS惯性器件的确定性误差进行标定后,根据经典Allan方差技术,分别根据直接采样和交叠采样技术推导了递推Allan方差辨识算法,使得MEMS惯性器件随机误差参数的在线辨识成为可能。然后运用时间序列分析技术建立了组合导航系统中相控阵Doppler测速仪的噪声模型,并借鉴S面控制算法提出了适用于小型AUV的Doppler测速仪的Kalman滤波器。最后针对小型AUV采用的航姿参考系统中电子罗盘子系统需要进行自差校正的问题,提出了一种基于UT变换的强跟踪UKF算法,随之又改进了算法中渐消因子矩阵的计算方法,并又将噪声参数在线估计技术引入到该算法中,使得该算法不但自适应性好而且鲁棒性强,解决了小型AUV在海面进行自差校正时遇到的海浪干扰问题,提高了UKF算法对自差参数的辨识能力。在随后的章节中,运用频域内连续信号的分解与重构技术,基于FFT算法提出了角速率输入下的频域姿态解算方案,并在Matlab仿真环境下实现了频域姿态解算方案,通过与四元数微分方程的四阶龙格-库塔求解方法相比,该算法能够有效减小载体做圆锥运动时姿态解算存在的圆锥误差,提高系统的姿态解算精度。组合导航系统中电子罗盘子系统虽然精度较高,但在小型AUV运动过程中,往往会受到非重力加速度的干扰,导致其输出的航姿信息产生较大的跳变误差。而基于MEMS陀螺组件解算得到的航姿信息虽然不易受非重力加速度的干扰,却存在较大的积累误差。基于上述特点,采用自适应加权算法,将电子罗盘输出的航姿信息与基于MEMS陀螺组件解算得出的航姿信息相融合,平滑了电子罗盘输出的水平姿态角和航向角,提高了整个航姿系统的动态性能。最后针对小型AUV的水下组合导航系统在海流干扰下存在模型误差的问题,提出了一种带模型误差的自适应UKF算法,该算法基于虚拟噪声的思想,利用次优MAP估值器对虚拟噪声的统计量进行实时估计,提高了小型AUV导航系统的定位精度和滤波能力。更多还原

【Abstract】 The dissertation has investigated several key technologies of underwater integrated navigation system of small AUV (Autonomous Underwater Vehicles). They were devices error parameters identification technique and their noise filtering technique, attitude determination of AHRS (Attitude and Heading Reference System), and the fusion strategy of integrated navigation system.Owing to restrictions on hull size and cost of small AUV, its integrated navigation system always adopts MEMS inertial devices, which have small volume, low cost and low power consumption, besides small, low power consumption Doppler velocity log. These navigation devices can meet the requirements of mechanical and electrical characteristics of small AUV, although they often have low accuracy. The integrated navigation system consists of these devices has poor performance, and even affects the small AUV’s guidance and control system stability. The first half of dissertation focused on the above problems of the navigation equipment used by small AUV.First of all, after calibrating deterministic error of MEMS inertial devices, Allan variance recursive identification algorithm was derived using overlapping and direct sampling technique respectively, founded on classic Allan variance identification technique, which made it possible to identify the parameters of random noise online for each MEMS navigation devices.Then the noise model of phased-array Doppler velocity log adopted by small AUV’s navigation system was built by time series analysis technique. Referencing the S plane control algorithm, the Kalman filter to Doppler velocity log was designed, suitable for the small AUV.At last, strong tracking UKF algorithm based on unscented transformation was designed, using in the small AUV’s compass calibration process. The strong tracking algorithm was introduced into UKF completely based on UT technique, and the algorithm of its fading factor matrix was improved, meanwhile estimator of noise parameters was also introduced into this algorithm, so that made the algorithm have good adaptability and robustness. The algorithm solved the wave interference problem, and improved the recognition ability of error parameters of UKF algorithm, when small AUV’s compass calibrating on the sea surface.In the following chapters, angular rate input attitude determination solution was designed based on FFT (Fast Fourier Transformation) algorithm, using continuous signal decomposition and reconstruction technique in frequency domain. The algorithm was programmed in Matlab and compared with the quaternion attitude determination solved by fourth order Runge-Kutta algorithm in time domain. The comparison showed that frequency domain algorithm can reduce the coning error effectively, and improve the accuracy of attitude determination.Although the accuracy of electronic compass in integrated navigation system is higher, it is easily interfered by non-gravitational acceleration, causing jump error in output attitude signal. Meanwhile, the heading and attitude information solved by MEMS gyro unit is not easily spoiled by non-gravitational acceleration, but its accumulation error is large. Based on the above characteristics, the attitude information respectively solved by electronic compass and MEMS gyro unit was fused by adaptive weighted algorithm, so that the attitude and heading information by electronic compass was smoothed, and the dynamic performance of the whole AHRS was improved as well.In the last part of this thesis, an adaptive UKF algorithm with model error was designed to reduce the navigation system model error caused by ocean current disturbance. Based on the idea of virtual noise, this algorithm used suboptimal MAP estimator to calculate statistics of virtual noise real time, which helped to improve the accuracy of navigation system and to enhance the ability of filtering.更多还原