WO2020052213A1 - 一种迭代容积点无迹卡尔曼滤波方法 - Google Patents
一种迭代容积点无迹卡尔曼滤波方法 Download PDFInfo
- Publication number
- WO2020052213A1 WO2020052213A1 PCT/CN2019/079135 CN2019079135W WO2020052213A1 WO 2020052213 A1 WO2020052213 A1 WO 2020052213A1 CN 2019079135 W CN2019079135 W CN 2019079135W WO 2020052213 A1 WO2020052213 A1 WO 2020052213A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- points
- sigma
- unscented kalman
- kalman filter
- iterative
- Prior art date
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 30
- 230000008569 process Effects 0.000 claims description 27
- 238000013507 mapping Methods 0.000 claims description 11
- 238000012886 linear function Methods 0.000 claims description 5
- 238000003745 diagnosis Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 2
- 238000004891 communication Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/15—Correlation function computation including computation of convolution operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Definitions
- the invention belongs to the technical field of communication navigation, and relates to an iterative volume point unscented Kalman filtering method.
- Kalman filtering technology is an algorithm for optimally estimating the state of the system through the system's input and output observation data. It has important significance and has been well applied in many fields such as communication, navigation, guidance and control. In linear stationary systems with known noise, general linear Kalman filtering can be effectively used, but in non-linear systems, because the state transition matrix cannot be represented linearly, there are many improved Kalman filtering techniques. .
- Extended Kalman filter is to expand the nonlinear system function according to Taylor, and take its linear part to solve the state matrix. Although this method is less computationally expensive than other nonlinear filters, it cannot solve high-degree-of-freedom nonlinear strong System estimation problem.
- Volumetric Kalman Filtering and Unscented Kalman Filtering is a filtering technique that fits a line with points. It takes a series of points in a non-linear function and then fits the entire non-linear function by mapping the point set.
- the trace Kalman filtering process is shown in Figure 1.
- the volume points are characterized by the same weight, but the volume points cannot fit the statistical characteristics of strong nonlinear functions well.
- the sigma points of the untraced Kalman filter have weights during the filtering process. Becomes negative, which diverges the filtered results.
- the present invention discloses an iterative volume point unscented Kalman filtering method, which is improved on the original Kalman filtering method, and can be effectively applied to a system with high degree of freedom and strong nonlinearity.
- the invention firstly provides a new method for selecting sigma points; secondly, according to the requirements of the positive definiteness of weights in the filtering process, the weight coefficients of each sigma points are given, which is closer to the statistical characteristics of the state quantity, thereby solving the traditional unscented Karl
- the filter divergence problem caused by Mann filter due to the non-positive definiteness of the error covariance matrix again, the design flow of the general volume point Kalman filtering algorithm is given; finally, the parameter iteration method is used to diagnose the weight of the sigma points in the filter online Positive definiteness.
- the present invention provides the following technical solutions:
- An iterative volume point unscented Kalman filter algorithm includes the following steps:
- Re-determining the weighting coefficient of the sigma points Re-determine the weighting coefficients of the sigma points in the filtering process, calculate the average value of the state prediction value at the last moment online, and compare with the state amount calculated by the weighted average of the non-linear mapping of the sigma points To determine the positive definiteness of the weight coefficient, the average value of the state quantity calculated by the weighted average of the sigma points must be equal to the average value of the previous state prediction value, and ensure that the weights are positive definite throughout the filtering process;
- volume point unscented Kalman filtering algorithm Introducing the Kalman gain iteration coefficient into the volume point unscented Kalman filter, and detecting the positive definiteness of the weight in the filtering process in real time to avoid filtering divergence.
- step 1) selects the sigma points of the iterative volume point unscented Kalman filter algorithm in step 1):
- m is the average of the initial state quantities
- P is the error covariance matrix of the initial state quantities
- k is the empirical value
- ⁇ determines the degree of dispersion of the sigma points
- T is the transpose of the matrix
- E i is defined as:
- step 2 the specific steps of re-determining the weighting coefficient of the sigma point in step 2) include:
- step 3 specifically includes the following steps:
- X is a state quantity
- Q is a covariance matrix of noise
- u is an input
- F ( ⁇ k , u k ) is a state quantity
- ⁇ is the sigma points by mapping of the observation matrix
- H is the observation matrix
- W i m is the weight of observations
- R k is the measurement noise
- Y is the observed value
- K is the Kalman gain
- step 4) specifically includes the following steps:
- G is a proportional parameter
- the present invention has the following advantages and beneficial effects:
- the invention can be effectively applied to a system of high-degree-of-freedom and strong nonlinearity containing random noise, and cooperatively solves the calculation amount problem, the non-linear filter divergence problem, and the negative weight problem, and can effectively improve the estimation accuracy and real-time property of the state quantity. Will diverge the filtering results.
- the present invention can better fit the statistical characteristics of non-linear system functions, and can avoid the non-positive definiteness of the sigma point weights with respect to the untraced Kalman filter.
- Figure 1 is a flowchart of the unscented Kalman filtering algorithm.
- FIG. 2 is a flowchart of a method for iteratively calculating a volume point Kalman filter provided by the present invention.
- An iterative volume point Kalman filtering method includes the following steps:
- Step 1 Initialize the state initial value and covariance matrix of the nonlinear system, and add the volume points selected in the volume Kalman filter algorithm to the sigma points of the untraced Kalman filter algorithm to form a new sigma point set online calculation state
- the mean and covariance of the quantity include the following sub-steps:
- volume points are added to the 2n + 1 sigma points of the original unscented Kalman filter to form a new set of 4n + 1 sigma points.
- the average value of the state values calculated by weighting the sigma points must be equal to The average value of the state prediction value of the previous step.
- the new sigma point is defined as:
- m is the average of the initial state quantities
- P is the error covariance matrix of the initial state quantities
- k is an empirical value, usually 0
- ⁇ determines the degree of the sigma points, usually a small positive value
- T is a matrix Transpose
- E i is defined as:
- I is the identity matrix
- Step 2 Re-determine the weight coefficient of the sigma points in the filtering process, calculate the average value of the state prediction value at the last moment online, and compare with the state amount calculated by the sigma point nonlinear mapping weighted average, and make the sigma point weighted average
- the calculated state quantity average must be equal to the average value of the previous state prediction value.
- the positive definiteness of the weight coefficient is determined. The process of re-determining the weight coefficient is as follows:
- the weights are defined as:
- Step 3 Use the improved sigma point set in step 1 and step 2 to solve the state transition matrix, establish a prediction and update model of the nonlinear system, and optimally estimate the state quantity at this moment, including the following processes:
- the improved sigma points map non-linear functions, perform one-step prediction of the state quantity of the system, and estimate the covariance matrix in the prediction update.
- X is a state quantity
- Q is a covariance matrix of noise
- u is an input
- F ( ⁇ k , u k ) is a state quantity.
- Step 4 Calculate the covariance matrix and Kalman gain in the prediction update, and continue to update the state quantity at the next moment, which specifically includes the following process:
- ⁇ is the sigma points by mapping of the observation matrix
- H is the observation matrix
- W i m is the weight of observations
- R k is the measurement noise
- Y is the observed value
- K is the Kalman gain.
- Step 5 Iterative coefficients are introduced into the Kalman gain, the positive definiteness of the weight coefficients is diagnosed online, and the new state quantities in the iterative process are repeatedly calculated, including the following processes:
- G is a proportional parameter
- j is used for the number of times of all iteration processes, and is defined as follows:
- x is the state amount of the index during the iteration.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Filters That Use Time-Delay Elements (AREA)
- Feedback Control In General (AREA)
- Navigation (AREA)
Abstract
Description
Claims (6)
- 一种迭代容积点无迹卡尔曼滤波算法,其特征在于,包括如下步骤:1)迭代容积点无迹卡尔曼滤波算法sigma点的选取:将容积卡尔曼滤波算法中选取的容积点添加到无迹卡尔曼滤波算法的sigma点中,形成新的sigma点集在线计算状态量的均值和协方差;2)重新确定sigma点的加权系数:重新确定滤波过程中sigma点的权重系数,在线计算上一时刻状态预测值的平均值,并与通过sigma点非线性映射加权平均计算后的状态量相比,确定权重系数的正定性;3)给出容积点无迹卡尔曼滤波算法的流程:利用改进后的sigma点拟合非线性函数,并通过非线性映射后的统计特性更新最优估计的状态量和协方差矩阵;4)迭代计算容积点无迹卡尔曼滤波算法:在容积点无迹卡尔曼滤波中引入卡尔曼增益迭代系数,并实时检测滤波过程中权重的正定性。
- 根据权利要求1所述的迭代容积点无迹卡尔曼滤波算法,其特征在于,所述步骤1)中迭代容积点无迹卡尔曼滤波算法sigma点的选取的具体步骤包括:(1.1)在容积积分中,利用2n个等权球面点来积分计算∫f(y)dy,其中f(y)是任意非线性系统函数,其中,n是函数变量的个数;(1.2)在原有的无迹卡尔曼滤波的2n+1个sigma点中添加容积点,形成新的4n+1个sigma点集,通过对sigma点加权平均计算后的状态值平均值必须等于前一步状态预测值的平均值,新的sigma点定义为:其中,其中,m是初始状态量的均值;P是初始状态量的误差协方差矩阵;k为经验取值;α决定sigma点的散布程度;T是矩阵的转置;E i的定义为:
- 根据权利要求1所述的迭代容积点无迹卡尔曼滤波算法,其特征在于,所述步骤3)具体包括如下步骤:(3.1)利用所述步骤1)和2)中改进的sigma点进行非线性函数的映射,对系统进行状态量的一步预测,并估计预测更新中的协方差矩阵:其中,X是状态量,Q是噪声的协方差矩阵,u是输入,F(χ k,u k)是状态量;其中,Υ为sigma点通过观测矩阵的映射;H为观测矩阵;W i m为观测量的权重;R k为 观测噪声;y为观测值;K为卡尔曼增益;
- 用于在线诊断滤波过程中权重的正定性,并且在每一次迭代中计算j=j+1,G=ηG,直到迭代过程结束,其中,R为观测噪声协方差矩阵。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811058809.4A CN109388778A (zh) | 2018-09-11 | 2018-09-11 | 一种迭代容积点无迹卡尔曼滤波方法 |
CN201811058809.4 | 2018-09-11 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2020052213A1 true WO2020052213A1 (zh) | 2020-03-19 |
Family
ID=65418908
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2019/079135 WO2020052213A1 (zh) | 2018-09-11 | 2019-03-21 | 一种迭代容积点无迹卡尔曼滤波方法 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN109388778A (zh) |
WO (1) | WO2020052213A1 (zh) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109388778A (zh) * | 2018-09-11 | 2019-02-26 | 东南大学 | 一种迭代容积点无迹卡尔曼滤波方法 |
CN111238484B (zh) * | 2020-02-28 | 2022-04-12 | 上海航天控制技术研究所 | 一种基于球形无迹变换的环火轨道自主导航方法 |
CN111756353B (zh) * | 2020-06-12 | 2024-04-16 | 杭州电子科技大学 | 一种基于非线性融合滤波的液位仪噪声优化方法 |
CN113805075A (zh) * | 2021-09-15 | 2021-12-17 | 上海电机学院 | 一种基于bcrls-ukf的锂电池荷电状态估计方法 |
CN117216482B (zh) * | 2023-11-07 | 2024-03-01 | 南开大学 | 借助迭代策略增强卡尔曼滤波器抗野值干扰能力的方法 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050251328A1 (en) * | 2004-04-05 | 2005-11-10 | Merwe Rudolph V D | Navigation system applications of sigma-point Kalman filters for nonlinear estimation and sensor fusion |
CN105975747A (zh) * | 2016-04-27 | 2016-09-28 | 渤海大学 | 一种基于无迹卡尔曼滤波算法的cstr模型参数辨识方法 |
CN108225337A (zh) * | 2017-12-28 | 2018-06-29 | 西安电子科技大学 | 基于sr-ukf滤波的星敏感器和陀螺组合定姿方法 |
CN109388778A (zh) * | 2018-09-11 | 2019-02-26 | 东南大学 | 一种迭代容积点无迹卡尔曼滤波方法 |
-
2018
- 2018-09-11 CN CN201811058809.4A patent/CN109388778A/zh active Pending
-
2019
- 2019-03-21 WO PCT/CN2019/079135 patent/WO2020052213A1/zh active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050251328A1 (en) * | 2004-04-05 | 2005-11-10 | Merwe Rudolph V D | Navigation system applications of sigma-point Kalman filters for nonlinear estimation and sensor fusion |
CN105975747A (zh) * | 2016-04-27 | 2016-09-28 | 渤海大学 | 一种基于无迹卡尔曼滤波算法的cstr模型参数辨识方法 |
CN108225337A (zh) * | 2017-12-28 | 2018-06-29 | 西安电子科技大学 | 基于sr-ukf滤波的星敏感器和陀螺组合定姿方法 |
CN109388778A (zh) * | 2018-09-11 | 2019-02-26 | 东南大学 | 一种迭代容积点无迹卡尔曼滤波方法 |
Also Published As
Publication number | Publication date |
---|---|
CN109388778A (zh) | 2019-02-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020052213A1 (zh) | 一种迭代容积点无迹卡尔曼滤波方法 | |
CN110659722B (zh) | 基于AdaBoost-CBP神经网络的电动汽车锂离子电池健康状态估算方法 | |
CN107045490A (zh) | 一种非线性系统的状态估计方法 | |
CN111537903B (zh) | 一种基于hckf的电池soc估计方法 | |
Hariz et al. | Optimal rate of convergence for nonparametric change-point estimators for nonstationary sequences | |
CN114567288B (zh) | 基于变分贝叶斯的分布协同非线性系统状态估计方法 | |
Muntwiler et al. | Learning-based moving horizon estimation through differentiable convex optimization layers | |
CN116595897A (zh) | 一种基于消息传递的非线性动态系统状态估计方法和装置 | |
CN109341690B (zh) | 一种鲁棒高效的组合导航自适应数据融合方法 | |
CN110222299A (zh) | 针对双变量含有误差的直线拟合问题的方法和设备 | |
CN110110711B (zh) | 一种噪声信道下的迭代学习控制系统输入信号估计方法 | |
Mattila et al. | What did your adversary believeƒ Optimal filtering and smoothing in counter-adversarial autonomous systems | |
CN114614797B (zh) | 基于广义最大非对称相关熵准则的自适应滤波方法和系统 | |
Vladimirov et al. | Directly coupled observers for quantum harmonic oscillators with discounted mean square cost functionals and penalized back-action | |
CN114413448B (zh) | 空调所属建筑物一阶模型实时参数辨识方法 | |
CN107421543B (zh) | 一种基于状态扩维的隐函数量测模型滤波方法 | |
CN115498980A (zh) | 一种基于M估计的递归最小p阶自适应滤波定位方法 | |
CN112636719A (zh) | 数据丢失和信道噪声干扰下的ilc系统输入信号滤波方法 | |
CN110765608A (zh) | 一种用于微机电系统传感器的高精度交互两级估计算法 | |
CN116340766B (zh) | 基于滑动窗口的锂电池soc在线预测方法与相关设备 | |
CN113269309B (zh) | 适用于微弱信号提取的rbf神经网络改进ukf方法 | |
CN117236060A (zh) | 一种基于辅助模型的连续搅拌反应釜新型递推估计方法 | |
CN113472318B (zh) | 一种顾及观测模型误差的分级自适应滤波方法及系统 | |
Kolås et al. | State estimation IS the real challenge in NMPC | |
CN117150212A (zh) | 一种基于小样本历史数据的一元时间序列预测方法及设备 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 19860316 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: 19860316 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 13.01.2022) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 19860316 Country of ref document: EP Kind code of ref document: A1 |