北斗导航型接收终端简化型稳健平方根容积卡尔曼滤波
发布时间：2018-10-29 12:11
【摘要】：本文围绕着降低动态定位中算法复杂度、减少计算量、提升计算效率;解决动态定位模型不匹配、使用单一模型误差较大;解决动态定位中状态噪声和量测噪声为非高斯白噪声的影响三类问题展开研究。主要内容和创新点如下:1.基于卫星导航的载体状态方程为线性,且稳健平方根容积卡尔曼(Square root Cubature Kalman Filtering—SCKF)在状态更新时其容积点经状态转移矩阵传递后的加权和为零,则可使用标准KF算法进行状态更新,量测更新过程仍采用SCKF;本文提出了简化型稳健平方根容积卡尔曼算法(Simplified SCKF,简称SSCKF)。该算法旨在解决动态导航定位计算量大、效率低的问题。仿真及实测数据表明SSCKF与SCKF精度相当,而解算时间较SCKF算法降低25%左右,能有效地降低算法复杂度,提升算法效率。2.基于SSCKF,结合变维交互多模思想,本文提出了简化型稳健平方根容积卡尔曼变维交互多模算法。该算法针对常规交互多模模型集覆盖不全面及模型数目过多导致的模型竞争等问题,将不同维数的模型交互,如匀速模型和匀加速模型,同时进行并行滤波,并由二者的量测残差计算出相应的似然函数,更新两种模型滤波结果所占的权重,将最终加权和作为整个变维模型的结果输出;下一时刻子模型的状态输入值不采用其自身上一时刻滤波结果,而是采用变维交互模型整体输出结果乘以维数转换矩阵得到的值,这样就能保证各时刻的状态输入值的准确性。3.针对动态导航过程中,状态噪声和量测噪声一般呈现非高斯白噪声的特点,本文提出了简化型SCKF高斯和算法。分析了动态导航中伪距测量值噪声的峭度值和相关系数,得出其呈现非高斯特性的结论;若仍将实际运动中的非高斯噪声强制当成高斯白噪声来处理,则会对滤波精度造成影响。采用多个高斯白噪声作为子高斯项,利用其加权和来近似表示非高斯白噪声,同时限制各时刻子高斯项总数目以确保各时刻解算效率。跑车实测数据验证,该算法能够有效抑制非高斯白噪声的影响,提升算法的稳定性和滤波精度。
[Abstract]:This paper focuses on reducing the complexity of the algorithm in dynamic location, reducing the computational complexity, improving the computational efficiency, solving the mismatch of the dynamic location model, and using a single model with large error. To solve the problem that the state noise and measurement noise are non-Gao Si white noise in dynamic positioning, three kinds of problems are studied. The main contents and innovations are as follows: 1. The carrier state equation based on satellite navigation is linear, and the weighted sum of the volume points transferred by the state transfer matrix is zero when the robust square root volume (Square root Cubature Kalman Filtering-SCKF) is updated. The standard KF algorithm can be used to update the state, and the SCKF; is still used in the measurement update process. In this paper, a simplified robust square-root volume Kalman algorithm (Simplified SCKF, for SSCKF). Is proposed. This algorithm aims to solve the problem of large amount of computation and low efficiency in dynamic navigation. The simulation and measured data show that the precision of SSCKF is equal to that of SCKF, and the solution time is about 25% lower than that of SCKF algorithm, which can effectively reduce the complexity of the algorithm and improve the efficiency of the algorithm. 2. Based on SSCKF, and variable dimensional interactive multimode theory, a simplified robust square root volume variable dimension interactive multimode algorithm is proposed. In order to solve the problem of model competition caused by incomplete coverage of conventional interactive multimode model set and excessive number of models, the model with different dimensions, such as uniform model and uniform acceleration model, is filtered in parallel at the same time. The corresponding likelihood function is calculated from the measurement residuals of the two models, the weight of the filtering results of the two models is updated, and the final weighted sum is taken as the output of the whole variable-dimensional model. The state input value of the next sub-model does not use its own filtering result at the last moment, but the global output of the variable dimensional interactive model is multiplied by the value obtained by the dimension conversion matrix. This ensures the accuracy of the state input values at each time. 3. In view of the fact that the state noise and the measurement noise generally present non-Gao Si white noise in the dynamic navigation process, this paper presents a simplified SCKF Gao Si and algorithm. The kurtosis and correlation coefficient of pseudo-range measurement noise in dynamic navigation are analyzed. If the non-Gao Si noise in the actual motion is still forced to be treated as Gao Si white noise, the filtering accuracy will be affected. Several Gao Si white noises are used as the subterms of Gao Si, and the weighted sum of them is used to approximate the non-Gao Si white noise, and at the same time to limit the total number of sub-Gao Si items at each time to ensure the efficiency of the solution at each time. The experimental data show that the algorithm can effectively suppress the influence of non-Gao Si white noise and improve the stability and filtering accuracy of the algorithm.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN967.1;TN713
本文编号：2297664
[Abstract]:This paper focuses on reducing the complexity of the algorithm in dynamic location, reducing the computational complexity, improving the computational efficiency, solving the mismatch of the dynamic location model, and using a single model with large error. To solve the problem that the state noise and measurement noise are non-Gao Si white noise in dynamic positioning, three kinds of problems are studied. The main contents and innovations are as follows: 1. The carrier state equation based on satellite navigation is linear, and the weighted sum of the volume points transferred by the state transfer matrix is zero when the robust square root volume (Square root Cubature Kalman Filtering-SCKF) is updated. The standard KF algorithm can be used to update the state, and the SCKF; is still used in the measurement update process. In this paper, a simplified robust square-root volume Kalman algorithm (Simplified SCKF, for SSCKF). Is proposed. This algorithm aims to solve the problem of large amount of computation and low efficiency in dynamic navigation. The simulation and measured data show that the precision of SSCKF is equal to that of SCKF, and the solution time is about 25% lower than that of SCKF algorithm, which can effectively reduce the complexity of the algorithm and improve the efficiency of the algorithm. 2. Based on SSCKF, and variable dimensional interactive multimode theory, a simplified robust square root volume variable dimension interactive multimode algorithm is proposed. In order to solve the problem of model competition caused by incomplete coverage of conventional interactive multimode model set and excessive number of models, the model with different dimensions, such as uniform model and uniform acceleration model, is filtered in parallel at the same time. The corresponding likelihood function is calculated from the measurement residuals of the two models, the weight of the filtering results of the two models is updated, and the final weighted sum is taken as the output of the whole variable-dimensional model. The state input value of the next sub-model does not use its own filtering result at the last moment, but the global output of the variable dimensional interactive model is multiplied by the value obtained by the dimension conversion matrix. This ensures the accuracy of the state input values at each time. 3. In view of the fact that the state noise and the measurement noise generally present non-Gao Si white noise in the dynamic navigation process, this paper presents a simplified SCKF Gao Si and algorithm. The kurtosis and correlation coefficient of pseudo-range measurement noise in dynamic navigation are analyzed. If the non-Gao Si noise in the actual motion is still forced to be treated as Gao Si white noise, the filtering accuracy will be affected. Several Gao Si white noises are used as the subterms of Gao Si, and the weighted sum of them is used to approximate the non-Gao Si white noise, and at the same time to limit the total number of sub-Gao Si items at each time to ensure the efficiency of the solution at each time. The experimental data show that the algorithm can effectively suppress the influence of non-Gao Si white noise and improve the stability and filtering accuracy of the algorithm.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN967.1;TN713
【参考文献】
相关期刊论文 前10条
1 张凯;单甘霖;;基于高斯和与SCKF的非线性非高斯滤波算法[J];仪器仪表学报;2014年11期
2 李小民;杜占龙;郑宗贵;张国荣;毛琼;;基于STSCKF和改进型χ~2检验的模拟电路故障辨识[J];仪器仪表学报;2014年10期
3 王树磊;魏瑞轩;关旭宁;;容积法则辅助的交互式多模型滤波算法[J];控制与决策;2014年09期
4 张鑫春;郭承军;;均方根嵌入式容积卡尔曼滤波[J];控制理论与应用;2013年09期
5 姜伟;吕泽均;蓝瑶;;基于变维交互作用的IMM-CKF算法[J];计算机应用与软件;2013年05期
6 王硕;刘丽;;基于机动转弯目标的自适应交互式多模型算法[J];计算机仿真;2013年04期
7 孙枫;唐李军;;Cubature卡尔曼滤波与Unscented卡尔曼滤波估计精度比较[J];控制与决策;2013年02期
8 孙枫;唐李军;;Cubature卡尔曼滤波-卡尔曼滤波算法[J];控制与决策;2012年10期
9 李振华;宁磊;徐胜男;;基于均差滤波与高斯和的非线性非高斯系统滤波算法[J];控制与决策;2012年01期
10 孙枫;唐李军;;Cubature粒子滤波[J];系统工程与电子技术;2011年11期
相关博士学位论文 前3条
1 何可可;非线性非高斯条件下贝叶斯滤波若干问题研究[D];南京理工大学;2012年
2 曹轶之;非高斯/非线性滤波算法研究及其在GPS动态定位中的应用[D];解放军信息工程大学;2012年
3 唐李军;Cubature卡尔曼滤波及其在导航中的应用研究[D];哈尔滨工程大学;2012年
相关硕士学位论文 前3条
1 李家森;北斗/INS组合导航信息融合滤波算法研究[D];哈尔滨工程大学;2013年
2 陈勇;信息融合技术在组合导航系统中的应用研究[D];南京理工大学;2007年
3 彭竞;非线性滤波技术在卫星导航系统中的应用研究[D];国防科学技术大学;2005年
，本文编号：2297664
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2297664.html
