智能汽车SINS/GNSS动态组合导航观测模型系统误差自适应估计方法

Adaptive Estimation Method of Observational Model Systematic Errors for Intelligent Vehicle SINS/GNSS Dynamic Integrated Navigation

为了满足当前汽车智能化发展对导航定位精度的更高需求,解决车辆非线性动态组合导航定位过程中,因观测模型偏差导致的定位精度下降问题,结合自适应滤波原理,提出了一种新的非线性动态组合导航系统观测模型系统误差自适应估计方法,用于抑制观测模型系统误差对滤波估计精度的影响,克服现有误差估计方法的缺陷和不足,提升车辆动态组合导航定位精度。首先,利用预测观测协方差矩阵和新息信息构造自适应因子,提出了一种自适应无迹卡尔曼滤波算法(AUKF),通过自适应调整状态预测和观测信息的权重抑制异常观测对动态导航系统的影响。随后,在考虑观测模型系统误差影响的情况下,对非线性导航系统模型进行改进,并在此基础上将提出的AUKF扩展用于估计和补偿观测模型系统误差,以提高系统状态估计精度。最后,将提出的自适应估计方法应用于搭载SINS/GPS非线性组合定位系统的车辆在环山路的跑车试验中,分别从考虑和不考虑观测模型系统误差影响2个方面对算法性能进行验证,并与其他滤波方法进行对比分析。试验结果表明:与现有方法相比,提出的方法不仅能够控制异常观测对非线性导航系统的影响,而且能够有效地估计和补偿观测模型系统误差,因此具有较好的滤波解算精度;定位精度小于0.46 m,即使在系统误差干扰情况下,定位精度仍然达到0.50 m,在抑制观测模型系统误差干扰方面显著优于其他滤波方法,且算法简单易行,在精度和实时性方面达到了更佳的平衡。

To meet the higher demand for navigation and positioning accuracy in the current development of vehicle intelligence and solve the problem of reduced accuracy caused by observational model bias in the nonlinear dynamic integrated navigation and positioning of intelligent vehicles,a novel adaptive estimation method of observational model systematic errors for nonlinear dynamic integrated navigation systems was proposed based on the principle of adaptive filtering.This method can be used to suppress the effects of observational model systematic errors on filtering estimation accuracy and to overcome the defects and deficiencies of existing error estimation methods,thereby improving the positioning accuracy of vehicle dynamic integrated navigation.First,an adaptive unscented Kalman filtering(AUKF)algorithm was developed by constructing an adaptive factor using innovation information and the covariance matrices of predicted observations.Then,the effects of abnormal observations on the dynamic navigation system were suppressed by adaptively adjusting the weight of the state prediction and observational information.Based on the effects of observational model systematic errors,the nonlinear navigation system model was improved and the proposed AUKF was extended to estimate and compensate for observational model systematic errors and thereby improve the accuracy of system state estimation.Finally,the proposed adaptive estimation method was applied to a road test of a vehicle equipped with a strap-down inertial navigation system(SINS)/global positioning system(GPS)nonlinear integrated positioning system on a mountain road,and its performance was verified in two respects:①when considering and not considering the effects of observational model systematic errors,and②when compared with other filtering methods.Experimental results demonstrate that when compared with existing methods,the proposed method not only can control the effects of abnormal observations on nonlinear navigation systems,it can also effectively estimate and compensate for observational model systematic errors.Accordingly,this method has good filtering accuracy,showing a positioning accuracy of less than 0.46 m.Even in the case of systematic error interference,the positioning accuracy still reaches 0.50 m.The proposed method is superior to other filtering methods in suppressing the interference of observational model systematic errors.The algorithm is simple and easy to implement,achieving a better balance between accuracy and real-time performance.