摘要
为了提高船用单轴旋转捷联惯性导航系统(RSINS)初始对准的精度和快速性,针对传统的EKF滤波线性化误差和单传感器精度不高的问题,设计了一种基于自适应交互多模(AIMM)算法的SCNS/RSINS紧组合对准方法。该算法将自适应滤波器与交互多模型相结合,利用了两个合理构建状态模型和量测模型的平行滤波来实现对实际模态的覆盖:滤波1应用姿态四元数算法建立了状态方程的模型,量测量为RSINS与SCNS之间的姿态四元数误差;滤波2的根据SCNS/RSINS的误差特性构建了状态方程模型,量测量为RSINS与SCNS位置和航向误差,然后应用自适应IMM算法将两个平行滤波的估计值进行数据融合。在某种程度上,因状态噪声和量测噪声的不确定性,EKF的性能会被降低,而通过模型转换机制,IMM可用于选择一个合理的模型自动计算器来自适应地调整对准过程中噪声的协方差矩阵,因此该算法可以有效地解决SCNS/RSINS组合导航系统的初始对准问题。仿真结果表明:与EKF算法相比,基于自适应IMM算法的SCNS/RSINS组合对准方法的估计精度和对准快速能力都得到了改善,其中对方位陀螺漂移的估计时间缩短了至少40%。
In order to improve the accuracy and rapidity of initial alignment for shipboard single-axis Rotation Strapdown Inertial Navigation System（RSINS）, an Adaptive Interacting Multiple Model（AIMM） alignment method is applied into Strapdown Celestial Navigation System（SCNS）/RSINS tight integrated navigation system. The AIMM algorithm combines the IMM algorithm with the improved Sage-Husa adaptive filtering algorithm, which can achieve the coverage of real situation through a few sub-models. The complementary navigation information is provided by different kinds of sensors, which can overcome such problems as the single sensor＇s low accuracy and traditional extended Kalman filter＇s linearization errors, so the accuracy of the SCNS/RSINS integrated navigation system can be significantly improved. The proposed AIMM alignment algorithm utilizes two parallel filters, in which the state and observation models are established reasonably： the first filter is based on attitude quaternion algorithm, its measurement equation is formed from the difference between SCNS＇s and RSINS＇s attitude quaternion; and the second filter is based on the errors characteristics of SCNS and RSINS, its measurement equation is formed from the difference between SCNS＇s and RSINS＇s position and heading. Then an Adaptive Interacting Multiple Model Filter（AIMMF） algorithm is applied to process the above two parallel filters＇ data. To some extent, the performance of EKF will be degraded due to the uncertainty of the process and measurement noises. By means of a model switching mechanism, the IMM can be utilized for selecting an appropriate model automatically and calculating the process noise covariance in alignment phase. The resulting sensor fusion algorithm can effectively solve the alignment problem for the SCNS/RSINS integrated navigation system. Finally, the proposed AIMMF algorithm is testified by simulations, which show that the estimation accuracy and alignment rapid capacity are improved compared with those of th
作者
周凌峰
董燕琴
赵汪洋
赵小明
屈原津
侯志宁
ZHOU Ling-feng DONG Yan-qin ZHAO Wang-yang ZHAO Xiao-ming QU Yuan-jin HOU Zhi-ning(College of Automation, Harbin Engineering University, Harbin 150001, China Equipment Academy of the Rockets Force, Beijing 100092, China Tianjin Navigation Instrument Research Institute, Tianjin 300131, China)
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2016年第4期464-472,共9页
Journal of Chinese Inertial Technology
基金
国家重大科学仪器设备开发专项(2013YQ310799)
