基于自动导航的旋翼式微小型无人机农作物遥感信息解析方法

Micro multirotor unmanned aerial vehicle (UAV) is characterized with compact, low-cost, and high flexibility and is a main platform for rapid access to information on field crops. However, due to the limitation of the stability of Micro UAV control system and the accuracy of automatic navigation system, the problem to be solved is how to adjust the aircraft position quickly and accurately so that the target is always in the line of sight of a camera or other monitoring system. It is a critical technology for Micro UAV platform to become widely used in precision agriculture. The objectives of this current project are to 1) establish a three-dimensional dynamic model of a hovering Micro UAV; 2) develop multi-sensor fusion algorithm and the extended Kalman filter algorithm for Micro UAV position and posture correction control; 3) establish automatic navigation position andattitude compensation inference rulethrough artificial neural network in order to provide accurate status information for Micro UAV control system. Based on the achievement of three objectives abovementioned, we propose an optimum selectionmethod of spectral wavelengths based on an improved particle swarm algorithm to improve analytical resolution of post-processed remotely-sensed spectral data.The effective implementation of this project will provide a fairly new platform to access remote sensing information based on an autonomousMicro UAV and data analysis method. It will accelerate the development of information technology in field crop monitoring.

旋翼式微小型无人机具有结构紧凑、成本低、灵活性高、获取数据快速等特点，是农田作物信息快速获取的主要平台。但受微小型无人机控制系统稳定性和机载自动导航系统精度的限制，如何快速准确地调整飞机位置、姿态使被测目标始终处于监测视野中是微小型无人机这一遥感平台能否被广泛应用在精确农业研究领域的关键技术。本项目以自动导航的旋翼式微小型无人机为研究对象，建立其定点悬停作业状态下的三自由度动力学模型；利用多传感器融合算法、扩展卡尔曼滤波算法实现对微小型无人机位置、姿态校正控制；采用人工神经网络建立自动导航位置、姿态补偿推理规则，为微小型无人机控制系统提供准确的状态信息。在此基础上提出一种基于改进粒子群算法的遥感信息波长优选方法，实现遥感光谱信息后处理数据解析精度。本项目的有效实施，将提供一种全新的基于自动导航的微小型无人机遥感信息获取平台及数据解析方法，将对农田作物监测信息化发展有积极的推动作用。

本项目以自动导航的微小型无人机为研究对象，建立其悬停作业状态下的三自由度随机非线性连续系统模型，在INS/GPS组合导航系统中获得多传感器观测值，利用多传感器融合算法、扩展卡尔曼滤波方程估计将传感器数据融合以计算无人机准确状态，实现对微小型无人机位置、姿态校正控制。.基于以上理论基础研究，项目搭建了微小型无人机测试平台进行了飞行测试试验，实时采集在不同侧风速干扰状态下无人机偏航角速度、俯仰角角速度、翻滚角速度、升力、推力、阻力及旋翼转动惯量等信息。开发了基于LabVIEW程序环境的无人机位姿检测软件，以图形化的形式直观显示无人机自主飞行状态和扩展卡尔曼滤波算法优化效果，将扩展卡尔曼滤波多传感器融合算法载入机载控制器实时检测无人机位姿信息提供理论依据。.最后，本项目与吉林大学地球探测科学与技术学院地理遥感实验室开展合作研究，借助微小型无人机遥感平台获取了作物归一化植被指数（NDVI）并构建了一种新型组合滤波算法Hyb-F (Hybrid filtering Method Based on Grubbs-introduced S-G and AG Algorithms)，将局部Savitzky-Golay（S-G）滤波和全局Asymmetric Gaussian （AG）拟合算法相结合，并在各自过程中引入格拉布斯Grubbs检测算法。该组合算法优点在于分步剔除序列局部噪声和全局噪声，构建具有精细细部特征，同时反映作物植被变化的NDVI时间序列。该研究解决了微小型无人机应用于精准农业领域的控制难题，实现了位置、姿态校正控制，提升了我国精准农业航空科技创新能力。.