CN109582914A - The relevant parallel type with bias system of noise merges estimation method - Google Patents

The relevant parallel type with bias system of noise merges estimation method Download PDF

Info

Publication number
CN109582914A
CN109582914A CN201910079209.4A CN201910079209A CN109582914A CN 109582914 A CN109582914 A CN 109582914A CN 201910079209 A CN201910079209 A CN 201910079209A CN 109582914 A CN109582914 A CN 109582914A
Authority
CN
China
Prior art keywords
noise
deviation
state
estimation
relevant
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.)
Granted
Application number
CN201910079209.4A
Other languages
Chinese (zh)
Other versions
CN109582914B (en
Inventor
葛泉波
王宏
张建朝
牛竹云
何美光
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.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi 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 Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN201910079209.4A priority Critical patent/CN109582914B/en
Publication of CN109582914A publication Critical patent/CN109582914A/en
Application granted granted Critical
Publication of CN109582914B publication Critical patent/CN109582914B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)

Abstract

The present invention relates to a kind of relevant parallel types with bias system of noise to merge estimation method.For the filtering problem for the multi-sensor measurement system that there is the dynamic deviation and correlated noise for influencing system mode and measurement, the invention proposes a kind of band deviation system method of estimation based on decorrelation technique and parallel type Multi-sensor Fusion thinking, by introducing decorrelation technique, re-establish a kind of incoherent equivalent model of noise, simultaneously, based on the relevant two-stage Kalman filter device of noise, multiple zero deflection filters and multiple deviation filters are merged respectively, finally, zero deflection state value and deviation after meromixis are merged, obtain the optimal estimation value of system mode.The present invention solves the problems, such as that the correlation of process noise and measurement noise in estimation causes filtering accuracy to decline.

Description

The relevant parallel type with bias system of noise merges estimation method
Technical field
It is the invention belongs to filter estimation field, in particular to a kind of to be based on noise decorrelation technique and block form fusion structure The estimation method with bias system.
Background technique
When dynamic deviation has and influence systematic procedure or measurement, the system as estimation basis increases deviation side Journey needs to estimate deviation to obtain more accurately filtering estimated value.
In face of the state estimation problem containing dynamic deviation, common method is that deviation is synthesized to a new state with state, Estimated, although this method is readily appreciated that, but its calculating is related to the calculating of higher dimensional matrix, computationally intensive.On this basis, The problem that is introduced as of two-stage Kalman filter device proposes a kind of solution, and the thinking of the estimation method is as follows: being added The adaptive covariance matrix of noise and transition matrix concept, using topology and transition matrix by the filtering of enhanced situation Device procedure decomposition is unbiased filter and deviation filter two parts, is compensated with deviation filter to unbiased filter, Obtain the estimated value of system mode.
Compared to single-sensor, multi-sensor Information Fusion System is advantageous in performance.It is carried out using single sensor When state estimation, measurement data source is single and is easy to appear problem, this precision for allowing for filtering estimation can not be guaranteed；Make When carrying out state estimation with multiple sensors, measurement data source is numerous, these measurement data are less likely while going wrong, The precision of estimation is guaranteed.Therefore, emphasis of the present invention is to there are the relevant estimation problems with bias system of noise.
Summary of the invention
In order to cope with noise correlation and dynamic deviation situation above-mentioned, present invention introduces noise decorrelation techniques, obtain Obtain the two-stage Kalman filter device under noise correlated condition.Based on above-mentioned filter, respectively to multiple zero deflection state filterings Device and deviation filter carry out parallel type fusion, propose noise relevant parallel type two-stage Kalman filter fusion estimation side Method.
The present invention can be generally divided into five parts.First part is that system model is established；Second part introduces decorrelation Technology re-establishes a kind of incoherent equivalent model of noise；Multiple parts two are obtained according to metric data in Part III Stage Kalman filter；Part IV carries out parallel type fusion to zero deflection filter and deviation filter respectively；Finally, will Two fusion results are combined, and obtain the estimated value of system mode,.
Beneficial effects of the present invention: can be handled noise correlation, relatively single two-stage Kalman filter device, this Invention can obtain the more fine estimation to zero deflection state and deviation.
Detailed description of the invention
Fig. 1 is the method for the present invention recursive process.
Fig. 2 is step 4 in the present invention and 5 detailed process.
Specific embodiment
Below in conjunction with attached drawing, the invention will be further described.
As shown in Figure 1, the present invention the following steps are included:
Step 1. system modelling
Considering a kind of common band deviation multisensor syste is model, and systematic procedure noise statistics are it is known that deposit It is described as follows in the state equation, deviation equation and measurement equation of the system of correlated noise:
In formula, k indicates time series；xk,bkAnd yi,kRespectively system n ties up state vector, m ties up bias vector and i-th The p of sensor ties up observation vector；And vi,kRespectively system mode noise vector, system deviation noise vector and i-th The measurement noise vector of sensor；Ak+1,k∈Rn×nFor state-transition matrix；Ci,k∈Rp×nFor the state observation of i-th of sensor Matrix.Process noise, deviation noise and measurement noise are zero mean Gaussian white noise sequences: vi,k~N (0, Vi,k), and
Step 2. introduces decorrelation technique, re-establishes a kind of incoherent equivalent model of noise
For the band deviation multisensor syste that step 1 provides, there are correlation, nothings with each measurement noise for process noise Method directly uses two-stage Kalman filter device.Therefore, noise decorrelation technique is introduced, equivalent change is carried out to system state equation It changes, obtains and the measurement incoherent new system mode noise of noise.Reconstruction process is as described below: firstly, in system mode side Journey adds N number of formula for being zero:
It takesThen new state-noise and deviation noise, measurement noise are uncorrelated, i.e.,
The model of original system can be rewritten as
Step 3. obtains filtering the relevant two stages Kalman of i-th of noise of system mode according to i-th of measurement equation The estimated information of wave device, specifically:
Zero deflection state, the prediction of deviation, estimated value are obtained by the relevant two-stage Kalman filter device of i-th of noise And its covariance matrix, estimating to the relevant two-stage Kalman filter device of i-th of noise of system mode is obtained by combination Count information.
Step 4. is based on the relevant two-stage Kalman filter device of noise, joined simultaneously in multi-sensor information fusion mode This fusion mode of line (before combination, respectively merges multiple zero deflection filters and multiple deviation filters)
In the linear system being made of multisensor, each part zero deflection filter i, which does zero deflection state, to be come from Oneself estimated value；Meanwhile partial deviations filter i makes the estimated value of oneself to deviation.N number of part zero deflection filter is pressed It is merged according to block form mode, obtains better zero deflection state estimation；N number of partial deviations filter according to same Mode is merged, and better estimation of deviation value is obtained.
Based on the estimated information of multiple zero deflection filters, obtain the fused state estimation of zero deflection filter and its Covariance matrix is respectively as follows:
In formula,The fused state estimation of zero deflection filter and its association side for i-th filter Poor matrix.
Based on the estimated information of multiple deviation filters, the fused estimation of deviation value of deviation filter and its association side are obtained Poor matrix, is respectively as follows:
In formula, bi,k+1/k,The fused state estimation of deviation filter and its association side for i-th filter Poor matrix.
Step 5. is combined two estimated values by linear combination, obtains the estimated information of system mode xk+1/k+1,
In formula, Vk+1It is fusion factor.Fig. 2 is the detailed process of step 4 and five.

Claims (1)

1. the relevant parallel type with bias system of noise merges estimation method, it is characterised in that method includes the following steps:
Step 1. system modelling
Considering a kind of common band deviation multisensor syste is model, and systematic procedure noise statistics are it is known that there are phases State equation, deviation equation and the measurement equation for closing the system of noise are described as follows:
In formula, k indicates time series；xk,bkAnd yi,kRespectively system n ties up state vector, m dimension bias vector and i-th of biography and passes The p of sensor ties up observation vector；And vi,kRespectively system mode noise vector, system deviation noise vector and i-th of biography The measurement noise vector of sensor；Ak+1,k∈Rn×nFor state-transition matrix；Ci,k∈Rp×nFor the state observation square of i-th of sensor Battle array；Process noise, deviation noise and measurement noise are zero mean Gaussian white noise sequences:vi,k~N (0, Vi,k), and
Step 2. introduces decorrelation technique, re-establishes a kind of incoherent equivalent model of noise
Due to state-noise with measure noise there are correlation, system state equation needs equivalent transformation, the following institute of reconstruction process It states:
Firstly, adding N number of formula for being zero in system state equation:
It takesThen new state-noise and deviation noise, measurement noise are uncorrelated, i.e.,
Original system equivalent model is
Step 3. is obtained according to i-th of measurement equation to the relevant two-stage Kalman filter device of i-th of noise of system mode Estimated value, specifically:
By the relevant two-stage Kalman filter device of i-th of noise obtain zero deflection state, the prediction of deviation, estimated value and its Covariance matrix obtains the estimated value to the relevant two-stage Kalman filter device of i-th of noise of system mode by combination；
Step 4. is based on the relevant two-stage Kalman filter device of noise, and parallel type is added to multi-sensor information fusion mode and melts Syntype；
Based on the filtering estimated information of multiple zero deflection filters, obtain the fused state estimation of zero deflection filter and its Covariance matrix is respectively as follows:
In formula,For the fused state estimation of zero deflection filter and its covariance square of i-th of filter Battle array；
Based on the filtering estimated information of multiple deviation filters, the fused estimation of deviation value of deviation filter and its association side are obtained Poor matrix is respectively as follows:
In formula, bi,k+1/k,For the fused state estimation of deviation filter and its covariance square of i-th of filter Battle array；
Fused zero deflection state estimation and estimation of deviation value respectively are combined by step 5., obtain estimating for system mode Count information xk+1/k+1,
CN201910079209.4A 2019-01-28 2019-01-28 Parallel fusion estimation method of noise-related deviation system Active CN109582914B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910079209.4A CN109582914B (en) 2019-01-28 2019-01-28 Parallel fusion estimation method of noise-related deviation system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910079209.4A CN109582914B (en) 2019-01-28 2019-01-28 Parallel fusion estimation method of noise-related deviation system

Publications (2)

Publication Number Publication Date
CN109582914A true CN109582914A (en) 2019-04-05
CN109582914B CN109582914B (en) 2023-08-18

Family

ID=65918356

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910079209.4A Active CN109582914B (en) 2019-01-28 2019-01-28 Parallel fusion estimation method of noise-related deviation system

Country Status (1)

Country Link
CN (1) CN109582914B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110110711A (en) * 2019-06-06 2019-08-09 郑州轻工业学院 A kind of iterative learning control systems input signal estimation method under noisy communication channel
CN110209998A (en) * 2019-06-25 2019-09-06 北京信息科技大学 Optimal sequential-type under non-ideal communication channel merges estimation method

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014215822A (en) * 2013-04-25 2014-11-17 日本電信電話株式会社 State estimating apparatus, method, and program

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014215822A (en) * 2013-04-25 2014-11-17 日本電信電話株式会社 State estimating apparatus, method, and program

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
丁维福;秦超英;郝慧娟;: "基于两阶段卡尔曼滤波的多传感器信息融合" *
王艳艳;刘开周;封锡盛;: "基于强跟踪平方根容积卡尔曼滤波的纯方位目标运动分析方法" *
石章松,王树宗,刘忠: "用于纯方位目标跟踪的修正极坐标自适应卡尔曼滤波算法" *
陈兴林;李鹏;宋申民;: "基于模糊自适应卡尔曼滤波算法的多传感器信息融合" *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110110711A (en) * 2019-06-06 2019-08-09 郑州轻工业学院 A kind of iterative learning control systems input signal estimation method under noisy communication channel
CN110110711B (en) * 2019-06-06 2021-06-04 郑州轻工业学院 Iterative learning control system input signal estimation method under noise channel
CN110209998A (en) * 2019-06-25 2019-09-06 北京信息科技大学 Optimal sequential-type under non-ideal communication channel merges estimation method
CN110209998B (en) * 2019-06-25 2022-04-01 北京信息科技大学 Optimal sequential fusion estimation method under non-ideal channel

Also Published As

Publication number Publication date
CN109582914B (en) 2023-08-18

Similar Documents

Publication Publication Date Title
Sui et al. Craft: Cross-attentional flow transformer for robust optical flow
JP3612360B2 (en) Motion estimation method using moving object segmentation method
CN111814719B (en) Skeleton behavior recognition method based on 3D space-time diagram convolution
US10268901B2 (en) Quasi-parametric optical flow estimation
KR101634562B1 (en) Method for producing high definition video from low definition video
CN109543143A (en) The Multi-sensor Fusion estimation method of non-linear belt bias system
CN110501696B (en) Radar target tracking method based on Doppler measurement adaptive processing
CN110289989B (en) Distributed state estimation method based on cubature Kalman filtering algorithm
CN105100807A (en) Motion vector post-processing based frame rate up-conversion method
CN109582914A (en) The relevant parallel type with bias system of noise merges estimation method
CN109887008B (en) Method, device and equipment for parallax stereo matching based on forward and backward smoothing and O (1) complexity
CN111583300B (en) Target tracking method based on enrichment target morphological change update template
CN107729291A (en) A kind of Identification of Bilinear Model Parameters method based on decomposition technique
CN103684349B (en) A kind of kalman filter method based on recursion Estimates on Covariance Matrix
CN105550997A (en) Three-dimensional matching image denoising method based on multiple transform domains
CN111623703A (en) Novel Kalman filtering-based Beidou deformation monitoring real-time processing method
CN109871508A (en) There are the centralized two stages kalman estimate methods of correlated measurement noise
CN110569706A (en) Deep integration target tracking algorithm based on time and space network
US20130176300A1 (en) Disparity maps in uniform areas
CN111260131A (en) Short-term traffic flow prediction method and device
US20060204104A1 (en) Image processing method, image processing apparatus, program and recording medium
Zheng On least‐squares identification of stochastic linear systems with noisy input–output data
CN111191694A (en) Image stereo matching method
CN113255541B (en) Process parameter denoising method of self-adaptive flow industrial process based on eigenmode function recombination signal relative entropy
CN115100444A (en) Image mismatching filtering method and image matching device thereof

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant