【作者】 徐鹏；

【导师】 裴福俊； 刘红云；

【作者基本信息】 北京工业大学 ， 控制科学与工程， 2014， 硕士

【摘要】 微小型自主水下航行器是20世纪以来世界各国探测开发海洋环境的重要工具，因此自主导航系统成为了自主水下航行器完成勘探任务的关键技术。而由于受到探测器体积和功耗的限制和工作环境的限制，无法采用例如GPS导航等传统的导航方式。因此微机械惯性测量单元就成为了自主水下航行器首选的导航设备，但是由于惯性导航系统具有测量误差随时间累积的特点，所以需要采用外辅助器辅助测量来对导航误差进行矫正。由于受到体积和功耗的限制，本系统不采用传统的多普勒测速仪作为外部传感器，而提出来以惯性导航系统为核心，加入此偏振光矫正信息，并以水下航行器的运动模型辅助的方法。本文研究工作包括以下几个方面：首先根据各种传感器的制约性和AUV工作环境的特殊性，本文提出了偏振光/DVL/SINS的组合模式，利用偏振光相机代替传统的磁罗盘传感器来测量AUV的航向角，并用以来作为组合导航的辅助信息。并对此设计了联邦卡尔曼滤波器，用于对系统进行滤波估计，将滤波结果反馈给组合导航系统，用以减少组合导航的导航误差，同时运用仿真实验对此系统进行了验证。其次指出由于受到DVL工作条件的限制，引入现阶段研究的微小型水下航行器运动学模型，对微小型自主水下航行器进行水中受力分析，建立了水下航行器的运动学模型。由于海洋环境复杂无法十分准确的描述水下航行器所受到的力，所以该力学模型需要在特定的工作环境下才可应用。其中要求航行器处于无漩涡和界流的环境中，基于这个动力学模型辅助和偏振光相机测量航向信息辅助的前提下，建立了水下航行器的组合导航模型，并且利用传统的卡尔曼滤波对模型进行滤波估计，将滤波结果反馈给组合导航系统，用以减小导航误差。由于海洋环境复杂，水下航行器很容易受到海底漩涡或界流的影响，因此本文对原有水下航行器运动学模型进行改进，在其受力中加入漩涡或界流对其的影响，这样就会在动力学模型辅助的输出信息中存在无法确定的测量噪声，因此传统的卡尔曼滤波就无法满足系统需要，本文提出运用自适应卡尔曼滤波对模型中的不确定性进行估计方法以满足对导航系统的稳定性的要求。针对导航系统中测量噪声的不确定性，将分别用Sage-Husa自适应卡尔曼滤波、多渐消因子自适应卡尔曼滤波、模糊自适应卡尔曼滤波对其进行滤波计算，分别对其进行仿真。其中模糊自适应卡尔曼滤波分别利用T-S模糊模型和Mamdani模型实现模糊规则。同时为了解决模糊控制器计算量大，计算法则程度高的问题，提出了运用数学函数对模糊过程进行拟合，通过运用拟合函数代替模糊控制器的方法来提高系统的运算速度。更多还原

【Abstract】 Micro autonomous underwater vehicle has become a significant tool for variescountries to make their marine environment exploration since the20th century.Therefore, autonomous navigation system has become the key technologies ofautonomous underwater vehicle exploration mission. Due to the limitation of detectorvolume, power consumption and working environment, some traditional methods ofnavigation such as GPS could not be applied. So the micro-mechanical measurementunit takes the prior position to the device for autonomous underwater vehiclenavigation. However, the exterior correction should be used to aid the navigation errordue to the characteristics of error accumulation caused by inertial navigation system.Because of the constraints of size and power consumption, the presented externalsensor was not chose as Doppler-velocimetry, The proposed method is used inertialnavigation system as the core, adding the compass correction information andunderwater vehicle motion model-assisted methods. This paper includes the followingaspects:Firstly, according to the special nature of the constraints of the various sensorsand AUV work environment, this paper presents polarization/DVL/SINS combinationmode, the camera instead of the traditional use of polarized light sensors to measurethe magnetic compass heading angle AUV, and since as a combination withnavigation auxiliary information. And this design federal Kalman filter for filteringthe system estimates will filter the results back to the integrated navigation system toreduce the navigational error of navigation, while the use of simulation of this systemwas verified.Secondly, due to the limitations noted DVL working conditions, the introductionof micro and small underwater vehicle research stage to the kinematic model of themicro and small AUV carried water stress analysis, kinematic model AUV. Due to thecomplexity of the marine environment, the force suffered could not be evacuatedaccurately. The model of the force suffered needs to be applied in a particular workenvironment. In which the vehicle should be under the environment without aircraftand swirl flow. The model was established on secondary kinetic model combined withmagnetic compass heading information. And the conventional Kalman filter forfiltering was presented, including the results of the feedback filter integratednavigation system to reduce the navigational error.Due to the complexity of the Marine environment, underwater vehicle is easilyaffected by sea whirlpool or boundary flow, therefore，We improved the originalkinematics model of underwater vehicle，take the vortex or boundary flow’s impact into account. This will in the output of the auxiliary information exists in the uncertaindynamic model of the measurement noise which made the traditional Kalman filterunusable. To satisfy the stability of the navigation system, We proposethe adaptivekalman filter estimate the uncertainty model.We use Sage-Husa adaptive kalman filter and adaptive kalman filter fadingfactor, the fuzzy adaptive kalman filter to filter operation and simulation aiming at theuncertainty of measurement noise in the navigation system，Using the T-S fuzzymodel and Mamdani model to achieve fuzzy adaptive kalman filtering。We come upwith the mathematical function of fuzzy process fitting to solve the problem that largeamount of calculation of the fuzzy controller and the algorithm of high degree, weusing the method of fitting function instead of a fuzzy controller to improve theoperation speed of the system simultaneously.更多还原