【作者】 朱敏；

【导师】 陈熙源；

【作者基本信息】 东南大学 ， 仪器科学与技术， 2023， 硕士

【摘要】 高分辨率航空对地观测系统在灾害监测、资源勘察和军事侦察等领域发挥着重要作用。但观测平台受复杂内外部因素影响,遥感载荷无法保证匀速直线运动,成像质量下降。分布式位姿测量系统可以通过分布式传递对准提供多节点高精度信息,辅助遥感载荷实现运动误差的精确补偿。对此,本文深入研究基于柔性机翼形变分析的分布式传递对准方法。主要研究内容和成果如下:(1)分析了机翼形变引起的形变角、角速度和杆臂变化,建立基于柔性机翼形变分析的传递对准模型。首先从形变测量的角度,使用符合真实情况的模拟机翼,并布置FBG(Fiber Bragg Grating)传感器网络测量机翼形变。经地面实验验证,以高精度主子IMU(Inertial Measurement Unit)测得形变角为基准,不同载荷下机翼尾部FBG计算的形变角误差均不超过0.04°。其次从模型优化的角度,将误差角和动态杆臂引入状态变量,并使用FBG计算的机翼形变角修正量测。通过U型对地成像阶段的仿真实验验证,相比于传统模型,俯仰角、横滚角和航向角分别提高了78.5%、65.1%、88.6%。(2)提出了量测噪声自适应算法、多重因子自适应算法结合的改进鲁棒滤波算法。在量测噪声统计特性发生变化时,使用量测噪声自适应算法,并对滤波器退化、非正定以及新息失真等问题提出解决方案。在系统发生状态突变时,设计多重因子自适应算法,在遗忘因子自适应的基础上增加多通道精确补偿。通过在仿真实验中加入量测噪声和状态突变,验证了算法可有效适应两种变化。经基于实际飞行数据的半物理仿真实验验证,在爬升阶段,所提出算法的姿态估计精度在0.01°以内;在U型成像阶段,相比于传统算法,姿态和位置估计精度分别提高了83.9%、80.7%。(3)提出了基于全局融合的分布式传递对准算法。首先针对传感器故障问题,引入残差卡方检测及隔离量测异常值;其次针对子节点位姿精度较差问题,结合机翼子节点空间相关性以及分布滤波,提出基于全局估计的分布滤波算法,以修正当前时刻低精度子节点姿态和位置估计结果。经基于实际飞行数据的半物理仿真实验验证,相比于故障隔离前,隔离后的姿态精度有一定提高;基于全局估计的分布滤波算法相比于前期改进的自适应滤波算法,低精度子节点的姿态和位置估计精度分别提高了42.5%、50.1%。综上所述,本文以提高分布式位姿测量系统精度为目标,对机翼形变引起的影响展开研究,将FBG阵列辅助IMU测量的方法用于分布式传递对准,并引入自适应滤波算法和信息融合对此进行改进,在不同飞行阶段均可较优估计子节点运动参数信息。更多还原

【Abstract】 High-resolution aerial earth observation system plays an important role in disaster monitoring,resource exploration and military reconnaissance.However,the observation platform is affected by complex internal and external factors,and the remote sensing load cannot guarantee uniform linear motion,so the imaging quality is degraded.Distributed position and orientation system can provide multi-nodes high-precision information through distributed transfer alignment,and assist remote sensing load to realize accurate compensation of motion error.In this thesis,the distributed transfer alignment method based on flexible wing deformation analysis is deeply studied.The main research contents and achievements are as follows:(1)The deformation angle,angular velocity and lever arm changes caused by wing deformation are analyzed,and a transfer alignment model based on flexible wing deformation analysis is established.Firstly,from the perspective of deformation measurement,FBG(Fiber Bragg Grating)sensor network is arranged on the realistic simulated wing to measure the deformation of the wing.Ground experiment shows that the error of deformation angle calculated by FBG at the wing tail is less than 0.04°,using the high-precision IMU(Intrinsic Measurement Unit)measured deformation angle as the benchmark.Secondly,from the perspective of model optimization,the error angle and dynamic lever arm are taken as state variables,and the wing deformation angle calculated by FBG is used to correct the measurement.The simulation experiment in earth observation phase shows that compared to traditional models,the pitch angle,roll angle,and heading angle are increased by 78.5%,65.1%,and 88.6%,respectively.(2)An improved robust filtering algorithm combining measurement noise adaptive algorithm and multi-factor adaptive algorithm is proposed.Aiming at the unknown statistical characteristics of measurement noise,the improved adaptive measurement noise algorithm is proposed,which focuses on the filter degradation,non-positive definite matrix and innovation distortion.Aiming at the problem of state mutation,the multi-factor adaptive algorithm is used to compensate accurately in multiple channels,which is improved on the basis of forgetting factor adaptive algorithm.Simulation experiments with added measurement noise and state mutation verify that the improved adaptive algorithm can effectively adapt to these changes.The semi-physical simulation based on real flight data shows that compared with the traditional algorithm,the attitude estimation accuracy is within 0.01° during the climbing phase,and the attitude and position estimation accuracy are improved by 83.9% and 80.7% respectively during the U-shaped imaging phase.(3)A distributed transfer alignment algorithm based on global fusion is proposed.Aiming at the problem of sensor faults,residual chi-square detection and measurement isolation are carried out.Aiming at the problem of poor accuracy of sub-nodes,combined with the spatial correlation of wing sub-nodes and distributed filtering,a distributed filtering algorithm based on global estimation is proposed to correct the low-precision attitude and position estimation results of the sub-nodes.The semi-physical simulation based on real flight data shows that the attitude accuracy is improved after fault isolation,and compared to the improved adaptive algorithm,the distributed filtering algorithm based on global estimation improves the attitude and position estimation accuracy of low-precision sub-node by 42.5% and 50.1%,respectively.To sum up,in order to improve the accuracy of the distributed position and orientation system,this thesis studies the influence caused by wing deformation,applies the method of FBG array assisted IMU measurement to distributed transfer alignment,and improves it by combining adaptive filtering algorithm and information fusion,which can better estimate the motion parameter information of sub-nodes in different flight stages.更多还原