WO2021063136A1 - 一种数据驱动的高精度组合导航数据融合方法 - Google Patents

一种数据驱动的高精度组合导航数据融合方法 Download PDF

Info

Publication number
WO2021063136A1
WO2021063136A1 PCT/CN2020/111697 CN2020111697W WO2021063136A1 WO 2021063136 A1 WO2021063136 A1 WO 2021063136A1 CN 2020111697 W CN2020111697 W CN 2020111697W WO 2021063136 A1 WO2021063136 A1 WO 2021063136A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
integrated navigation
sampling point
matrix
fusion method
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/111697
Other languages
English (en)
French (fr)
Inventor
崔冰波
龚辰
魏新华
卢泽民
李晋阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Priority to GB2105557.9A priority Critical patent/GB2591954B/en
Publication of WO2021063136A1 publication Critical patent/WO2021063136A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/005Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 with correlation of navigation data from several sources, e.g. map or contour matching
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/10Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
    • G01C21/12Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
    • G01C21/16Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
    • G01C21/183Compensation of inertial measurements, e.g. for temperature effects
    • G01C21/188Compensation of inertial measurements, e.g. for temperature effects for accumulated errors, e.g. by coupling inertial systems with absolute positioning systems
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/10Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
    • G01C21/12Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
    • G01C21/16Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01CMEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20Instruments for performing navigational calculations

Definitions

  • the invention belongs to the field of integrated navigation and multi-source data fusion, and specifically relates to a data-driven high-precision integrated navigation data method.
  • Navigation and positioning are widely used in the fields of national defense, industry and agriculture, such as satellite navigation, inertial navigation, visual navigation and LiDAR. Since a single navigation system is often unable to handle navigation problems in complex environments, the application needs to be based on multiple sources. Integrated navigation system with sensors. In particular, new navigation and positioning applications represented by unmanned driving have extremely demanding requirements on the robustness and intelligence of the navigation system, and multi-source integrated navigation has become the preferred solution.
  • the nonlinear filtering technology represented by extended Kalman filter (EKF) is widely used in the field of integrated navigation, and the combined filtering of the nonlinear filtering model can be satisfied by linearizing the model function at the model prediction point.
  • EKF extended Kalman filter
  • EKF Error Kalman Filter
  • CKF volumetric Kalman filter
  • the estimation of CKF is often too optimistic, that is, its variance value is too small.
  • the prior art does not consider the influence of the nonlinearity of the measurement model on the filter update, and assumes that the second moment of the random variable in the Gaussian approximation process can be accurately matched, ignoring the influence of the system uncertainty on the state estimation model.
  • the present invention proposes a data-driven high-precision integrated navigation data fusion method, which approximates model errors based on IMU (Inertial Measurement Unit) raw data, realizes accurate estimation of integrated system navigation parameters, and improves The robustness of the CKF combined filtering algorithm is improved.
  • IMU Inertial Measurement Unit
  • a data-driven high-precision integrated navigation data fusion method When the integrated navigation system is working normally, the navigation parameters and sampling points are recursively updated based on the data of multi-source heterogeneous sensors; when the integrated navigation system is interfered, the sampling point error propagation model The integrated navigation system is provided with continuous auxiliary measurement updates, and the sampling point error propagation model uses an extreme learning machine combined with the original data of the inertial measurement unit to update the sampling points.
  • the input of the extreme learning machine is the priori prediction distribution information of the state model, the output angle increment of the sky gyro and the travel direction ratio, and the output is the posterior sampling point error matrix.
  • the prior prediction distribution information includes a sampling point prediction error matrix with
  • state prior distribution and likelihood function are calculated by using Gaussian process integral moment matching GPQMT.
  • the input parameter frequency of the extreme learning machine is asynchronous data
  • the output parameter frequency can be selected as any one of the input parameter frequencies.
  • H ⁇
  • ( ⁇ 1 ,..., ⁇ M ) is the weight connecting the hidden layer nodes with the network output
  • ( ⁇ 1 ... ⁇ N ) is the sample Output variables:
  • the training process of the extreme learning machine keeps the randomly generated initial input weights and biases unchanged.
  • the present invention provides a data-driven high-precision integrated navigation data fusion method, which has the following beneficial effects compared with the prior art:
  • the present invention proposes a sampling point prediction error generation method based on Gaussian process integral moment matching (GPQMT), which improves the sampling point generation during the training process of the sampling point error propagation model The quality of the state prior distribution and the variance estimation accuracy of the likelihood function are further improved.
  • GPQMT Gaussian process integral moment matching
  • the present invention proposes a data-driven transformation of the sampling point error matrix to update the sampling points, which improves the efficiency of non-linear measurement updates and improves the state posterior distribution Update frequency.
  • the present invention obtains the sampling point prediction error matrix with The output angle increment of the sky gyro in the integrated navigation system and the specific force of the carrier travel direction are used as the input variables of the sampling point error propagation model, which realizes the direct coupling of the sampling point update and the unobservable state quantity, and the system model moment is directly based on the sensor data
  • the approximation of the matching error improves the robustness of the parameterized model of the integrated navigation system.
  • Figure 1 is a flow chart of data-driven sampling point update of the present invention
  • Fig. 2 is a flowchart of sampling point error conversion based on ELM of the present invention.
  • a data-driven high-precision integrated navigation data fusion method When the integrated navigation system is working normally, the navigation parameters and sampling points are recursively updated based on the data of multi-source heterogeneous sensors; when the integrated navigation system is interfered, the Gaussian process is used to achieve the combination
  • the process of the data-driven transformation of the sampling point error matrix includes: After the integrated navigation system is powered on, it is performed at time t k ⁇ t n Fitting of the sampling point error propagation model; when t k > t n , the model error is approximated based on the original IMU data, and the sampling point error matrix is predicted.
  • Step (1) sampling point prediction error matrix calculation
  • the volume point vector ⁇ k is initialized by the CKF sampling point generation method, and the integrated navigation system function f(x k-1 ) and the measurement function h(x k ) Is calculated to obtain the sampling point prediction matrix with Then, get the state prior distribution based on GPQMT
  • the kernel function of the a-th dimension state variable propagated through f(x k-1 );
  • the off-diagonal element of the posterior variance is calculated
  • the variance of the system noise w k is Q k , then That is, the diagonal elements of the posterior variance are generated, and the posterior variance generated based on GPQMT can be obtained by comprehensive formula (5).
  • the kernel function of the a-th dimension state variable of h(x k) Is the hyperparameter of the kernel function, which represents the signal variance of the a-th dimension state variable.
  • the estimated value of the posterior moment predicted by h(x k) can be obtained, and then the measurement The variance of noise R k can be obtained.
  • Step (2) training and prediction of sampling point error propagation model
  • the predicted covariance matrix of the joint probability distribution of navigation parameters are P k
  • IMU Inertial measurement unit
  • the sampling point error matrix is predicted, and the steps are shown by the dotted line in Figure 2: 1) As the input variable of the prediction model, the posterior propagation error matrix of the sampling point at time t k is predicted The subscript s2 indicates that the current error matrix is the posterior information of the prediction phase; 2) The mean and variance of the posterior distribution of the state variable calculated by CKF are P k
  • ELM extreme learning machine
  • ⁇ i is the network output weight
  • ⁇ i is the input weight connecting the input variable and the hidden layer node
  • b i is the bias
  • the ELM training process keeps the randomly generated initial input weights and biases unchanged.
  • H + is the generalized inverse of matrix H .
  • the output angle increment of the sky gyro in the integrated navigation system and the specific force of the carrier travel direction are used as the input variables of the sampling point error propagation model, which realizes the direct coupling between the sampling point update and the unobservable state quantity, and the system model is directly based on the sensor data (System function f(x k-1 ) and measurement function h(x k ))
  • System function f(x k-1 ) and measurement function h(x k ) The approximation of the moment matching error improves the robustness of the parameterized model of the integrated navigation system.

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Navigation (AREA)
  • Gyroscopes (AREA)

Abstract

一种数据驱动的高精度组合导航数据融合方法,采用高斯过程实现组合导航系统导航参数后验分布的精确估计,并实现采样点误差阵的数据驱动变换,具体为:组合导航系统上电后,在时刻t k<t n时进行采样点误差传播模型的拟合;当时刻t k>t n时,基于IMU原始数据进行模型误差的逼近,进行采样点误差矩阵的预测。基于采样点预测误差阵,将组合导航系统中的天向陀螺输出角增量、载体行进方向比力作为采样点误差传播模型的输入变量,实现了采样点更新,直接基于传感器数据进行系统模型矩匹配误差的逼近。由此实现了组合系统导航参数的精确估计,改善了组合导航系统参数化模型的鲁棒性。

Description

一种数据驱动的高精度组合导航数据融合方法 技术领域
本发明属于组合导航及多源数据融合领域,具体涉及一种数据驱动的高精度组合导航数据方法。
背景技术
导航定位在国防、工业及农业领域上都有广泛应用,如卫星导航、惯性导航、视觉导航以及LiDAR等,由于单一的导航系统往往无法处理复杂环境下的导航问题,应用中需要建立基于多源传感器的组合导航系统。尤其是以无人驾驶为代表的新型导航定位应用,其对导航系统的鲁棒性和智能化要求极为苛刻,多源组合导航成为优选方案。以扩展卡尔曼滤波(EKF)为代表的非线性滤波技术在组合导航领域应用广泛,通过在模型预测点处将模型函数线性化能够满足非线性滤波模型的组合滤波。然而受系统和量测不确定性影响,EKF在矩匹配的精度、量测更新的效率以及鲁棒性等方面均无法满足实际组合导航系统的需求。以无迹卡尔曼滤波(UKF)为代表的确定性采样点逼近策略可以较EKF获得更好的矩匹配精度和收敛速度。对于状态维数较高的组合导航滤波问题,容积卡尔曼滤波(CKF)较UKF有更好的稳定性。
由于采用有限的确定性采样点,无法匹配状态模型的整体概率分布函数,CKF的估计常常过于乐观即其方差值偏小。现有技术均没有考虑量测模型非线性对滤波更新的影响,且均假设高斯逼近过程中随机变量的二阶矩可以精确匹配,忽略了系统不确定性对状态估计模型的影响。
发明内容
为了克服现有技术的不足,本发明提出了一种数据驱动的高精度组合导航数据融合方法,基于IMU(惯性测量单元)原始数据进行模型误差的逼近,实现组合系统导航参数的精确估计,改善了CKF组合滤波算法的鲁棒性。
为实现上述目的,本发明采用的技术方案为:
一种数据驱动的高精度组合导航数据融合方法,组合导航系统正常工作时,基于多源异构传感器数据进行导航参数和采样点递推更新;当组合导航系统受到干扰时,采样点误差传播模型给组合导航系统提供连续的辅助量测更新,所述采样点误差传播模型采用极限学习机结合惯性测量单元的原始数据进行采样点更新。
进一步,所述极限学习机的输入为状态模型的先验预测分布信息、天向陀螺输出角增量和行进方向比力,输出为后验采样点误差阵。
更进一步,所述先验预测分布信息包括采样点预测误差阵
Figure PCTCN2020111697-appb-000001
Figure PCTCN2020111697-appb-000002
更进一步,所述
Figure PCTCN2020111697-appb-000003
其中
Figure PCTCN2020111697-appb-000004
为系统函数的采样点预测矩阵,
Figure PCTCN2020111697-appb-000005
为状态先验分布的均值。
更进一步,所述
Figure PCTCN2020111697-appb-000006
其中
Figure PCTCN2020111697-appb-000007
为量测函数的采样点预测矩阵,
Figure PCTCN2020111697-appb-000008
为似然函数的均值。
更进一步,所述状态先验分布和似然函数采用高斯过程积分矩匹配GPQMT计算得到。
进一步,所述极限学习机的输入参数频率是异步数据,输出参数频率可选为输入参数频率的任意一种。
进一步,所述采样点误差传播模型的训练过程为:设
Figure PCTCN2020111697-appb-000009
为输入变量,γ=1,…,2n,
Figure PCTCN2020111697-appb-000010
为输出变量;其中
Figure PCTCN2020111697-appb-000011
为k时刻天向陀螺输出角增量,
Figure PCTCN2020111697-appb-000012
为k时刻载体行进方向比力输出,
Figure PCTCN2020111697-appb-000013
均为采样点预测误差阵,n为组合导航系统状态维数;
采样点误差传播模型采用如下形式:
Figure PCTCN2020111697-appb-000014
1≤j≤N,其中ρ i为网络输出权值,N=2n为CKF采样点个数,φ i为连接输入变量和隐层节点的输入权值,b i为偏置,M为隐层节点的个数;上式写成矩阵形式有Hρ=Π,其中ρ=(ρ 1,…,ρ M)为连接隐层节点与网络输出的权值,Π=(Π 1 … Π N)为样本输出变量:
Figure PCTCN2020111697-appb-000015
极限学习机训练过程保持随机产生的初始输入权值和偏置不变,未知的网络输出权值ρ通过求解最小均方误差下的解ρ=H +Π得到,其中H +为矩阵H的广义逆。
本发明提出了一种数据驱动的高精度组合导航数据融合方法,相比现有技术,具有以下有益效果:
(1)本发明为克服高斯滤波模型方差估计精度不高的缺陷,提出一种基于高斯过程积分矩匹配(GPQMT)的采样点预测误差生成方法,提高了采样点误差传播模型训练过程采样点生成的质量,进而改善了状态先验分布和似然函数方差估计精度。
(2)本发明为改善组合导航系统滤波的输出连续性和稳定性,提出采样点误差阵的数据驱动变换进行采样点的更新,改善了非线性量测更新的效率并提高了状态后验分布更新频率。
(3)本发明得到采样点预测误差阵
Figure PCTCN2020111697-appb-000016
Figure PCTCN2020111697-appb-000017
将组合导航系统中天向陀螺输出角增 量、载体行进方向比力作为采样点误差传播模型的输入变量,实现了采样点更新与不可观测状态量的直接耦合,直接基于传感器数据进行系统模型矩匹配误差的逼近,改善了组合导航系统参数化模型的鲁棒性。
附图说明
图1为本发明数据驱动的采样点更新流程图;
图2为本发明基于ELM的采样点误差变换流程图。
具体实施方式
下面结合附图对本发明作更进一步的说明。
一种数据驱动的高精度组合导航数据融合方法,组合导航系统正常工作时,基于多源异构传感器数据进行导航参数和采样点递推更新;当组合导航系统受到干扰时,采用高斯过程实现组合导航系统导航参数后验分布的精确估计,并实现采样点误差阵的数据驱动变换,采样点误差阵的数据驱动变换的过程包括:组合导航系统上电后,在时刻t k<t n时进行采样点误差传播模型的拟合;当时刻t k>t n时,基于IMU原始数据进行模型误差的逼近,进行采样点误差矩阵的预测。
具体过程如下:
步骤(1),采样点预测误差阵计算
如图1所示,在t k<t n时,首先,采用CKF的采样点生成方法初始化容积点向量Θ k,对组合导航系统函数f(x k-1)和量测函数h(x k)的函数值进行计算,分别得到采样点预测矩阵
Figure PCTCN2020111697-appb-000018
Figure PCTCN2020111697-appb-000019
然后,基于GPQMT得到状态先验分布
Figure PCTCN2020111697-appb-000020
和似然函数
Figure PCTCN2020111697-appb-000021
Figure PCTCN2020111697-appb-000022
P k|k-1表示状态先验分布的均值与方差,
Figure PCTCN2020111697-appb-000023
表示似然函数的均值与方差,其中Z k-1=z 1:k-1为组合系统直至k-1时刻的量测序列;最后,计算得到采样点预测误差阵
Figure PCTCN2020111697-appb-000024
Figure PCTCN2020111697-appb-000025
所述GPQMT的实现过程如下:设组合导航系统的系统模型包括系统函数x k=f(x k-1)+w k和量测函数z k=h(x k)+v k,w k、v k分别为系统噪声和量测噪声,将上述方程定义为统一的形式y i=g(x i)+ε i,GPQMT基于已知数据集D={X:=[x 1,…,x N] T,Y:=[y 1,…,y N]}推理出函数g的后验分布,其中
Figure PCTCN2020111697-appb-000026
表示由n维实数空间到1维实数空间的映射,且
Figure PCTCN2020111697-appb-000027
σ ε为模型噪声标准差;设
Figure PCTCN2020111697-appb-000028
为统计特性已知的高斯随机变量,然后基于函数g的后验分布进行 g(x *)函数值的预测;X:=[x 1,…,x N] T为采样点集合,y:=[y 1,…,y N]为对应的函数g实例化值集合,N=2n为CKF采样点个数。下面以系统函数f(x k-1)的逼近为例详述函数后验分布的估计过程,设训练维数索引为a,a=1,…,N,则有对第a维变量有均值:
Figure PCTCN2020111697-appb-000029
Figure PCTCN2020111697-appb-000030
其中
Figure PCTCN2020111697-appb-000031
i和j表示组合导航系统训练数据集的点索引,i≠j,i,j=1,…,N,I为n维单位阵,y a是高斯过程训练的第a维变量期望目标,
Figure PCTCN2020111697-appb-000032
是拟合系统方程产生的噪声方差,
Figure PCTCN2020111697-appb-000033
经f(x k-1)传播的第a维状态变量的核函数;
Figure PCTCN2020111697-appb-000034
是核函数的超参数,表示第a维状态变量的信号方差,
Figure PCTCN2020111697-appb-000035
其中l f,i为长度缩放因子,i=1,…,N。针对f(x k-1)不同的状态维数选择相同的核函数参数,即
Figure PCTCN2020111697-appb-000036
类似的,第a维状态变量方差有:
Figure PCTCN2020111697-appb-000037
Figure PCTCN2020111697-appb-000038
其中
Figure PCTCN2020111697-appb-000039
当计算第a维状态变量与第b维状态变量的方差时,即计算后验方差的非对角线元素有在;
Figure PCTCN2020111697-appb-000040
设系统噪声w k的方差为Q k,则有
Figure PCTCN2020111697-appb-000041
即产生后验方差对角线元素,综合式(5)可得基于GPQMT生成的后验方差。
类似的,构造h(x k)的第a维状态变量的核函数
Figure PCTCN2020111697-appb-000042
是核函数的超参数,表示第a维状态变量的信号方差,基于式(1)、(3)和(5)可得经h(x k)预测的后验矩估计值,再考虑量测噪声的方差R k可得
Figure PCTCN2020111697-appb-000043
针对h(x k)不同的状态维数选择相同的核函数参数,即
Figure PCTCN2020111697-appb-000044
而在预测f(x k-1)和h(x k)后验分布时,选择不同的超参数集合θ f:={k ff,l f,i}、θ h:={k hh,l h,i}。
步骤(2),采样点误差传播模型的训练与预测
首先,基于CKF滤波框架、容积点向量Θ k计算导航参数联合概率分布的预测协方差阵
Figure PCTCN2020111697-appb-000045
状态后验分布的均值和方差为
Figure PCTCN2020111697-appb-000046
P k|k,并计算后验采样点误差阵
Figure PCTCN2020111697-appb-000047
如下:
Figure PCTCN2020111697-appb-000048
以惯性测量单元(IMU)天向陀螺输出角增量
Figure PCTCN2020111697-appb-000049
载体行进方向比力输出
Figure PCTCN2020111697-appb-000050
为输入变量,以
Figure PCTCN2020111697-appb-000051
为期望输出进行采样点误差阵传递函数Τ(Ξ)的拟合,其中下标s1表示当前误差阵为训练阶段后验信息。其次,当时刻t k>t N时,进行采样点误差矩阵的预测,其步骤如图2虚线所示:1)以
Figure PCTCN2020111697-appb-000052
为预测模型的输入变量,预测得到t k时刻采样点后验传播误差阵
Figure PCTCN2020111697-appb-000053
其中下标s2表示当前误差阵为预测阶段后验信息;2)采用CKF计算得到状态变量后验分布的均值和方差为
Figure PCTCN2020111697-appb-000054
P k|k,以IMU输出频率200Hz进行后验采样点误差阵
Figure PCTCN2020111697-appb-000055
的预测,并对5Hz输出的后验分布
Figure PCTCN2020111697-appb-000056
进行校正;3)将
Figure PCTCN2020111697-appb-000057
Figure PCTCN2020111697-appb-000058
相加得到t k+1时刻的采样点
Figure PCTCN2020111697-appb-000059
进而完成下一滤波周期采样点的初始化与预测采样点误差阵的计算。
采样点误差传播模型(即传递函数Τ(Ξ))的训练和预测过程均采用极限学习机(ELM)实现,其内容为:设
Figure PCTCN2020111697-appb-000060
为输入变量,γ=1,…,2n,
Figure PCTCN2020111697-appb-000061
为输出变量即后验采样点误差阵。设学习样本的个数为N=2n,即CKF采样点的个数,则有M个隐层节点的单隐藏层前馈神经网络(极限学习机)可表示为:
Figure PCTCN2020111697-appb-000062
其中ρ i为网络输出权值,φ i为连接输入变量和隐层节点的输入权值,b i为偏置;上式写成紧凑的矩阵形式有Hρ=Π,其中ρ=(ρ 1,…,ρ M)为连接隐层节点与网络输出的权值,Π=(Π 1,…,Π N)为样本输出变量:
Figure PCTCN2020111697-appb-000063
ELM训练过程保持随机产生的初始输入权值和偏置不变,未知的网络输出权值ρ可通过求解最小均方误差下的解ρ=H +Π得到,其中H +为矩阵H的广义逆。训练过程完成网络输出 权值的计算后,在模型工作在预测模式时,输入变量Ξ可得到预测的采样点误差阵输出
Figure PCTCN2020111697-appb-000064
γ=1,…,2n。
基于
Figure PCTCN2020111697-appb-000065
Figure PCTCN2020111697-appb-000066
将组合导航系统中的天向陀螺输出角增量、载体行进方向比力作为采样点误差传播模型的输入变量,实现了采样点更新与不可观测状态量的直接耦合,直接基于传感器数据进行系统模型(系统函数f(x k-1)和量测函数h(x k))矩匹配误差的逼近,改善了组合导航系统参数化模型的鲁棒性。
以上所述仅是本发明的优选实施方式,应当指出:对于本技术领域的普通技术人员来说,在不脱离本发明原理的前提下,还可以做出若干改进和润饰,这些改进和润饰也应视为本发明的保护范围。

Claims (8)

  1. 一种数据驱动的高精度组合导航数据融合方法,其特征在于:组合导航系统正常工作时,基于多源异构传感器数据进行导航参数和采样点递推更新;当组合导航系统受到干扰时,采样点误差传播模型给组合导航系统提供连续的辅助量测更新,所述采样点误差传播模型采用极限学习机结合惯性测量单元的原始数据进行采样点更新。
  2. 根据权利要求1所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述极限学习机的输入为状态模型的先验预测分布信息、天向陀螺输出角增量和行进方向比力,输出为后验采样点误差阵。
  3. 根据权利要求2所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述先验预测分布信息包括采样点预测误差阵
    Figure PCTCN2020111697-appb-100001
    Figure PCTCN2020111697-appb-100002
  4. 根据权利要求3所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述
    Figure PCTCN2020111697-appb-100003
    其中
    Figure PCTCN2020111697-appb-100004
    为系统函数的采样点预测矩阵,
    Figure PCTCN2020111697-appb-100005
    为状态先验分布的均值。
  5. 根据权利要求3所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述
    Figure PCTCN2020111697-appb-100006
    其中
    Figure PCTCN2020111697-appb-100007
    为量测函数的采样点预测矩阵,
    Figure PCTCN2020111697-appb-100008
    为似然函数的均值。
  6. 根据权利要求4或5所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述状态先验分布和似然函数采用高斯过程积分矩匹配GPQMT计算得到。
  7. 根据权利要求2所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述极限学习机的输入参数频率是异步数据,输出参数频率可选为输入参数频率的任意一种。
  8. 根据权利要求1所述的数据驱动的高精度组合导航数据融合方法,其特征在于:所述采样点误差传播模型的训练过程为:设
    Figure PCTCN2020111697-appb-100009
    为输入变量,γ=1,…,2n,
    Figure PCTCN2020111697-appb-100010
    为输出变量;其中
    Figure PCTCN2020111697-appb-100011
    为k时刻天向陀螺输出角增量,
    Figure PCTCN2020111697-appb-100012
    为k时刻载体行进方向比力输出,
    Figure PCTCN2020111697-appb-100013
    均为采样点预测误差阵,n为组合导航系统状态维数;
    采样点误差传播模型采用如下形式:
    Figure PCTCN2020111697-appb-100014
    其中ρ i为网络输出权值,N=2n为CKF采样点个数,φ i为连接输入变量和隐层节点的输入权值,b i为偏置,M为隐层节点的个数;上式写成矩阵形式有Hρ=Π,其中ρ=(ρ 1,…,ρ M)为连接隐层节点与网络输出的权值,Π=(Π 1 … Π N)为样本输出变量:
    Figure PCTCN2020111697-appb-100015
    极限学习机训练过程保持随机产生的初始输入权值和偏置不变,未知的网络输出权值ρ通过求解最小均方误差下的解ρ=H +Π得到,其中H +为矩阵H的广义逆。
PCT/CN2020/111697 2019-09-30 2020-08-27 一种数据驱动的高精度组合导航数据融合方法 Ceased WO2021063136A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
GB2105557.9A GB2591954B (en) 2019-09-30 2020-08-27 Data-driven high-precision integrated navigation data fusion method

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910943387.7 2019-09-30
CN201910943387.7A CN110702095B (zh) 2019-09-30 2019-09-30 一种数据驱动的高精度组合导航数据融合方法

Publications (1)

Publication Number Publication Date
WO2021063136A1 true WO2021063136A1 (zh) 2021-04-08

Family

ID=69197438

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/111697 Ceased WO2021063136A1 (zh) 2019-09-30 2020-08-27 一种数据驱动的高精度组合导航数据融合方法

Country Status (3)

Country Link
CN (1) CN110702095B (zh)
GB (1) GB2591954B (zh)
WO (1) WO2021063136A1 (zh)

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113581244A (zh) * 2021-07-20 2021-11-02 南京富岛信息工程有限公司 一种基于信息融合的智能铁鞋股道识别系统及方法
CN113722961A (zh) * 2021-09-01 2021-11-30 浙江大学 一种基于广义协同高斯过程模型的结构不确定性量化解析方法
CN114018250A (zh) * 2021-10-18 2022-02-08 杭州鸿泉物联网技术股份有限公司 惯性导航方法、电子设备、存储介质和计算机程序产品
CN114037017A (zh) * 2021-11-25 2022-02-11 西安电子科技大学 基于误差分布拟合的数据融合方法
CN114719850A (zh) * 2022-03-09 2022-07-08 北京理工大学 多层感知和容积卡尔曼滤波的姿态估计融合方法及系统
CN115112121A (zh) * 2022-03-28 2022-09-27 浙江德清知路导航科技有限公司 基于数据与模型结合的多源融合定位方法、系统及终端
CN115127554A (zh) * 2022-08-31 2022-09-30 中国人民解放军国防科技大学 一种基于多源视觉辅助的无人机自主导航方法与系统
CN115294745A (zh) * 2022-05-23 2022-11-04 电子科技大学 一种基于神经网络与数据差分的锂电池热失控分层预警方法
CN115469547A (zh) * 2022-09-21 2022-12-13 温州大学 一种逆变器的自适应控制系统
CN115523926A (zh) * 2022-10-25 2022-12-27 中国星网网络应用有限公司 一种融合状态信息确定方法、装置、电子设备及存储介质
CN116105731A (zh) * 2023-04-07 2023-05-12 中国人民解放军国防科技大学 稀疏测距条件下的导航方法、装置、计算机设备及介质
CN116523388A (zh) * 2023-04-17 2023-08-01 无锡雪浪数制科技有限公司 一种基于工业互联网平台的数据驱动的质量建模方法
CN116562124A (zh) * 2023-03-31 2023-08-08 中山大学 高速飞行器天线罩烧蚀对电磁性能影响的预测方法和装置
CN116975205A (zh) * 2023-08-04 2023-10-31 中国科学院空天信息创新研究院 一种太空环境数据与导航时频领域数据的关联分析方法
CN118428217A (zh) * 2024-04-29 2024-08-02 国网宁夏电力有限公司电力科学研究院 一种变压器铁心及夹件多点接地缺陷预测方法、介质及系统
CN119413162A (zh) * 2024-12-23 2025-02-11 珠海魔方芯智联科技有限公司 Imu数据与导航信号融合处理方法、装置、导航设备和介质
CN119714253A (zh) * 2024-11-15 2025-03-28 厦门大学 惯导、视觉与先验信息动态融合的定位方法
CN120820146A (zh) * 2025-09-15 2025-10-21 山东黄河水利工程质量检测中心 组合导航式水下作业平台高精度定位方法
CN121115058A (zh) * 2025-11-13 2025-12-12 北京安心易维科技有限公司 一种基于多源融合的高鲁棒性卫星时间快速估计方法

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110702095B (zh) * 2019-09-30 2022-09-16 江苏大学 一种数据驱动的高精度组合导航数据融合方法
CN113048984B (zh) * 2021-04-01 2023-10-03 江苏科技大学 一种用于水下无人机器人集群的动态定位信息融合方法
CN115014325B (zh) * 2022-06-01 2025-06-20 齐鲁工业大学(山东省科学院) 一种基于鲁棒容积点更新框架的组合导航数据融合方法
CN117268381B (zh) * 2023-11-17 2024-02-02 北京开运联合信息技术集团股份有限公司 一种航天器状态的判断方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105180935A (zh) * 2015-10-30 2015-12-23 东南大学 一种适用于gnss微弱信号的组合导航数据融合方法
CN107390246A (zh) * 2017-07-06 2017-11-24 电子科技大学 一种基于遗传神经网络的gps/ins组合导航方法
CN110702095A (zh) * 2019-09-30 2020-01-17 江苏大学 一种数据驱动的高精度组合导航数据融合方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7289906B2 (en) * 2004-04-05 2007-10-30 Oregon Health & Science University Navigation system applications of sigma-point Kalman filters for nonlinear estimation and sensor fusion
CN107621264B (zh) * 2017-09-30 2021-01-19 华南理工大学 车载微惯性/卫星组合导航系统的自适应卡尔曼滤波方法
CN108759838A (zh) * 2018-05-23 2018-11-06 安徽科技学院 基于秩卡尔曼滤波器的移动机器人多传感器信息融合方法
CN109284677B (zh) * 2018-08-16 2022-06-03 昆明理工大学 一种贝叶斯滤波目标跟踪算法
CN109341690B (zh) * 2018-09-25 2022-03-22 江苏大学 一种鲁棒高效的组合导航自适应数据融合方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105180935A (zh) * 2015-10-30 2015-12-23 东南大学 一种适用于gnss微弱信号的组合导航数据融合方法
CN107390246A (zh) * 2017-07-06 2017-11-24 电子科技大学 一种基于遗传神经网络的gps/ins组合导航方法
CN110702095A (zh) * 2019-09-30 2020-01-17 江苏大学 一种数据驱动的高精度组合导航数据融合方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
E. S. ABDOLKARIMI, ET AL.: "A wavelet-extreme learning machine for low-cost INS/GPS navigation system in high-speed applications.", GPS SOLUTIONS., vol. 22, no. 15, 15 November 2017 (2017-11-15), XP036401531, ISSN: 1080-5370, DOI: 10.1007/s10291-017-0682-x *
LI DENGAO; JIA XUAN; ZHAO JUMIN: "A Novel Hybrid Fusion Algorithm for Low-Cost GPS/INS Integrated Navigation System During GPS Outages", IEEE ACCESS, IEEE, USA, vol. 8, 13 March 2020 (2020-03-13), USA, pages 53984 - 53996, XP011780309, DOI: 10.1109/ACCESS.2020.2981015 *
WEN, YALI: "The Low-Cost Integrated Navigation System for Small Rotor UAV", CHINESE MASTER'S THESES FULL-TEXT DATABASE, ENGINEERING SCIENCE II, no. 08, 15 August 2016 (2016-08-15), pages 1 - 86, XP055798773, ISSN: 1674-0246 *

Cited By (28)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113581244A (zh) * 2021-07-20 2021-11-02 南京富岛信息工程有限公司 一种基于信息融合的智能铁鞋股道识别系统及方法
CN113581244B (zh) * 2021-07-20 2024-04-26 南京富岛信息工程有限公司 一种基于信息融合的智能铁鞋股道识别系统及方法
CN113722961A (zh) * 2021-09-01 2021-11-30 浙江大学 一种基于广义协同高斯过程模型的结构不确定性量化解析方法
CN113722961B (zh) * 2021-09-01 2024-02-13 浙江大学 一种基于广义协同高斯过程模型的结构不确定性量化解析方法
CN114018250A (zh) * 2021-10-18 2022-02-08 杭州鸿泉物联网技术股份有限公司 惯性导航方法、电子设备、存储介质和计算机程序产品
CN114018250B (zh) * 2021-10-18 2024-05-03 杭州鸿泉物联网技术股份有限公司 惯性导航方法、电子设备、存储介质和计算机程序产品
CN114037017B (zh) * 2021-11-25 2022-10-21 西安电子科技大学 基于误差分布拟合的数据融合方法
CN114037017A (zh) * 2021-11-25 2022-02-11 西安电子科技大学 基于误差分布拟合的数据融合方法
CN114719850A (zh) * 2022-03-09 2022-07-08 北京理工大学 多层感知和容积卡尔曼滤波的姿态估计融合方法及系统
CN115112121A (zh) * 2022-03-28 2022-09-27 浙江德清知路导航科技有限公司 基于数据与模型结合的多源融合定位方法、系统及终端
CN115112121B (zh) * 2022-03-28 2023-03-03 浙江德清知路导航科技有限公司 基于数据与模型结合的多源融合定位方法、系统及终端
CN115294745A (zh) * 2022-05-23 2022-11-04 电子科技大学 一种基于神经网络与数据差分的锂电池热失控分层预警方法
CN115294745B (zh) * 2022-05-23 2023-09-29 电子科技大学 一种基于神经网络与数据差分的锂电池热失控分层预警方法
CN115127554A (zh) * 2022-08-31 2022-09-30 中国人民解放军国防科技大学 一种基于多源视觉辅助的无人机自主导航方法与系统
CN115469547A (zh) * 2022-09-21 2022-12-13 温州大学 一种逆变器的自适应控制系统
CN115523926A (zh) * 2022-10-25 2022-12-27 中国星网网络应用有限公司 一种融合状态信息确定方法、装置、电子设备及存储介质
CN116562124A (zh) * 2023-03-31 2023-08-08 中山大学 高速飞行器天线罩烧蚀对电磁性能影响的预测方法和装置
CN116562124B (zh) * 2023-03-31 2024-02-06 中山大学 高速飞行器天线罩烧蚀对电磁性能影响的预测方法和装置
CN116105731A (zh) * 2023-04-07 2023-05-12 中国人民解放军国防科技大学 稀疏测距条件下的导航方法、装置、计算机设备及介质
CN116523388B (zh) * 2023-04-17 2023-11-10 无锡雪浪数制科技有限公司 一种基于工业互联网平台的数据驱动的质量建模方法
CN116523388A (zh) * 2023-04-17 2023-08-01 无锡雪浪数制科技有限公司 一种基于工业互联网平台的数据驱动的质量建模方法
CN116975205A (zh) * 2023-08-04 2023-10-31 中国科学院空天信息创新研究院 一种太空环境数据与导航时频领域数据的关联分析方法
CN118428217A (zh) * 2024-04-29 2024-08-02 国网宁夏电力有限公司电力科学研究院 一种变压器铁心及夹件多点接地缺陷预测方法、介质及系统
CN119714253A (zh) * 2024-11-15 2025-03-28 厦门大学 惯导、视觉与先验信息动态融合的定位方法
CN119413162A (zh) * 2024-12-23 2025-02-11 珠海魔方芯智联科技有限公司 Imu数据与导航信号融合处理方法、装置、导航设备和介质
CN119413162B (zh) * 2024-12-23 2025-04-18 珠海魔方芯智联科技有限公司 Imu数据与导航信号融合处理方法、装置、导航设备和介质
CN120820146A (zh) * 2025-09-15 2025-10-21 山东黄河水利工程质量检测中心 组合导航式水下作业平台高精度定位方法
CN121115058A (zh) * 2025-11-13 2025-12-12 北京安心易维科技有限公司 一种基于多源融合的高鲁棒性卫星时间快速估计方法

Also Published As

Publication number Publication date
GB202105557D0 (en) 2021-06-02
CN110702095B (zh) 2022-09-16
GB2591954B (en) 2022-02-09
CN110702095A (zh) 2020-01-17
GB2591954A (en) 2021-08-11

Similar Documents

Publication Publication Date Title
CN110702095B (zh) 一种数据驱动的高精度组合导航数据融合方法
CN109211276B (zh) 基于gpr与改进的srckf的sins初始对准方法
CN109459019B (zh) 一种基于级联自适应鲁棒联邦滤波的车载导航计算方法
CN103063212B (zh) 一种基于非线性映射自适应混合Kalman/H∞滤波器的组合导航方法
Gao et al. Distributed state fusion using sparse-grid quadrature filter with application to INS/CNS/GNSS integration
CN110610513B (zh) 自主移动机器人视觉slam的不变性中心差分滤波器方法
CN107727097B (zh) 基于机载分布式位置姿态测量系统的信息融合方法和装置
CN112798021A (zh) 基于激光多普勒测速仪的惯导系统行进间初始对准方法
CN108255791A (zh) 基于分布式传感器一致性的机动目标跟踪方法
CN109115228B (zh) 一种基于加权最小二乘容积卡尔曼滤波的目标定位方法
CN111256710A (zh) 一种地图匹配方法及系统
Wang et al. A compensation method for gyroscope random drift based on unscented Kalman filter and support vector regression optimized by adaptive beetle antennae search algorithm
CN108871365B (zh) 一种航向约束下的状态估计方法及系统
Shaukat et al. Underwater vehicle positioning by correntropy-based fuzzy multi-sensor fusion
CN109341690B (zh) 一种鲁棒高效的组合导航自适应数据融合方法
CN119394301B (zh) 一种LiDAR-IMU耦合里程计方法、系统、电子设备和车辆
CN112083457A (zh) 一种神经网络优化的imm卫星定位导航方法
Ye et al. Federated learning-enabled cooperative localization in multi-agent system
CN103968839B (zh) 一种基于蜂群算法改进ckf的单点重力匹配方法
CN115546258A (zh) 一种新型交互式多模型状态估计方法及系统
Si et al. A Review of Nonlinear Filtering Algorithms in Integrated Navigation Systems
CN111310110B (zh) 一种高维耦合不确定系统混合状态估计方法
CN109253727A (zh) 一种基于改进迭代容积粒子滤波算法的定位方法
CN106871905A (zh) 一种非理想条件下高斯滤波替代框架组合导航方法
Gao et al. SLAM estimation method for uncertain model noise parameters

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 202105557

Country of ref document: GB

Kind code of ref document: A

Free format text: PCT FILING DATE = 20200827

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20871032

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20871032

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20871032

Country of ref document: EP

Kind code of ref document: A1