摘要
针对GPS与微机电系统组合导航系统存在高非线性的缺陷,特别是GPS失效期间同卡尔曼滤波方位估计相关的不确定性的影响,提出混合误差模型。使用支持向量机对卡尔曼滤波输出进行建模,利用基于自回归的快速正交搜索对非线性方位误差进行建模,针对GPS/简化的惯性传感器系统利用混合支持向量机-快速正交搜索方法同传统卡尔曼滤波和增强卡尔曼滤波-快速正交搜索方法进行比较,结果表明,混合支持向量机-快速正交搜索方法优于卡尔曼滤波算法和增强卡尔曼滤波-快速正交搜索算法。
To reduce the influence of the uncertainty related to the Kalman filter(KF)azimuth estimation,a hybrid error model was proposed for GPS/MEMS integrated navigation system during GPS outages.Firstly,the support vector machine(SVM)was used to modeling the KF output.Secondly,the fast orthogonal search(FOS)based on autoregressive(AR)was employed to modeling the nonlinear azimuth error.Finally,the GPS/reduced inertial sensor system(RISS)system with the hybrid SVM-FOS method was compared with the traditional KF and the enhanced KF-FOS method.The results showed that the hybrid SVM-FOS method is better than the KF algorithm and the enhanced KF-FOS algorithm.
作者
庞玺斌
梁成程
张闯
PANG Xi-bin;LIANG Cheng-cheng;ZHANG Chuang(College of Navigation, Dalian Maritime University, Dalian Liaoning 116026, China)
出处
《船海工程》
北大核心
2020年第5期127-132,共6页
Ship & Ocean Engineering
基金
国家自然科学基金(51879027)。
关键词
快速正交搜索
支持向量机
简化的惯性传感器系统
GPS
fast orthogonal search(FOS)
support vector machine(SVM)
reduced inertial sensor system(RISS)
GPS
