基于阈值过滤的自适应无迹卡尔曼滤波在农业无人机组合导航中的应用  

Application of threshod filtering-based adaptive unscented Kalman filtering in agricultural drone integrated navigation

作　　者：唐思嘉 王其[2,3] 马云鹏 朱金印 TANG Sijia;WANG Qi;MA Yunpeng;ZHU Jinyin(College of Mechanical Engineering,Anhui Science and Technology University,Fengyang 233100,China;College of Computing,Nanjing University of Information Science and Technology,Nanjing 210044,China;College of Cyberspace Security,Nanjing University of Information Science and Technology,Nanjing 210044,China;Nanjing University of Information Science and Technology Nantong Research Institute,Nantong 226001,China)

机构地区：[1]安徽科技学院机械工程学院,安徽凤阳233100 [2]南京信息工程大学计算机学院,江苏南京210044 [3]南京信息工程大学网络空间安全学院,江苏南京210044 [4]南京信息工程大学南通研究院,江苏南通226001

出　　处：《安徽科技学院学报》2024年第6期94-103,共10页Journal of Anhui Science and Technology University

基　　金：江苏省自然科学基金青年项目(BK20160955)。

摘　　要：目的:针对无人机在进行喷药作业过程中系统模型会受到外界扰动引起滤波器噪声和干扰信号效果的问题,提出一种基于阈值过滤的自适应无迹卡尔曼滤波的方法。方法:通过设立阈值判断是否属于离群值,降低异常值的权重,减少因噪声干扰导致的系统模型误差。应用组合的信息融合方式,同时简化噪声估计过程。结果:通过MATLAB仿真实验证明,将UKF、AUKF、简化AUKF以及本算法进行比较,基于阈值过滤的自适应无迹卡尔曼算法在东向和北向速度的预测准确度上分别提高了57.9%和54.1%。结论:该方法在方位精度上有大幅提升,使得系统的鲁棒性更强。Objective:A threshold filtering based adaptive unscented Kalman filtering method was proposed to address the issue of filter noise and interference signal effects caused by external disturbances in the system model of drones during spraying operations.Methods:By setting a threshold to determine whether it belonged to outliers,the weight of outliers was reduced,and the system model error caused by noise interference was reduced.The information fusion method of combination was applied to simplify the noise estimation process.Results:The MATLAB simulation results showed that compared with UKF,AUKF,simplified AUKF and the algorithm in this paper,the adaptive unscented Kalman algorithm based on threshold processing had improved the prediction accuracy of eastbound and northbound speeds by 57.9%and 54.1%respectively.Conclusion:The method had greatly improved the position accuracy,making the system more robust.

关 键 词：喷药 阈值过滤 自适应无迹卡尔曼 噪声 松组合 

分 类 号：S852.3[农业科学—基础兽医学]