CN111239725B - Dynamic self-adaptive multi-radar information fusion method - Google Patents

Dynamic self-adaptive multi-radar information fusion method Download PDF

Info

Publication number
CN111239725B
CN111239725B CN202010148975.4A CN202010148975A CN111239725B CN 111239725 B CN111239725 B CN 111239725B CN 202010148975 A CN202010148975 A CN 202010148975A CN 111239725 B CN111239725 B CN 111239725B
Authority
CN
China
Prior art keywords
track
fusion
period
data
max
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.)
Active
Application number
CN202010148975.4A
Other languages
Chinese (zh)
Other versions
CN111239725A (en
Inventor
张良成
王运锋
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.)
Chengdu Yunwei Technology Co ltd
Original Assignee
Chengdu Yunwei Technology Co ltd
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 Chengdu Yunwei Technology Co ltd filed Critical Chengdu Yunwei Technology Co ltd
Priority to CN202010148975.4A priority Critical patent/CN111239725B/en
Publication of CN111239725A publication Critical patent/CN111239725A/en
Application granted granted Critical
Publication of CN111239725B publication Critical patent/CN111239725B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/87—Combinations of radar systems, e.g. primary radar and secondary radar
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
    • G01S13/50—Systems of measurement based on relative movement of target
    • G01S13/58—Velocity or trajectory determination systems; Sense-of-movement determination systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a dynamic self-adaptive multi-radar information fusion method, which aims at effective dynamic self-adaptive weighted fusion of multi-radar track information, receives and analyzes real-time quality characteristics of radar data sources, further calculates quality weight factors of the data sources, and realizes dynamic self-adaptive data fusion according to the weight factors so as to achieve the purpose of fusing multi-source data into a more optimal result; meanwhile, in order to further improve the stability and accuracy of the fusion result, the sensor data is iteratively optimized by using a filter. The method of the invention shows better fusion effect in both actual scene and simulation scene, and greatly reduces the influence of error and misinformation of the radar sensor.

Description

Dynamic self-adaptive multi-radar information fusion method
Technical Field
The invention relates to the field of multi-sensor information fusion processing and application, in particular to a dynamic self-adaptive multi-radar information fusion method.
Background
The multi-radar information fusion is to make full use of multiple radar information sources and combine redundant or complementary information of the multiple radar information sources in space or time according to a certain processing standard to obtain a consistent explanation or description of a measured object, so that the information system has better performance relative to a system formed by various subsets contained in the information system. The purpose of multi-source information fusion is mainly two aspects: on one hand, aiming at the redundancy of multi-source information, eliminating noise and abnormal values in input information; on the other hand, for the complementarity of the multi-source information, valuable information related to practical application is obtained, and complete information description of the observed object is obtained to the maximum extent.
The multi-source information fusion method has various types, each method has advantages and disadvantages, and the methods have certain differences. The method can be divided into an estimation theory method, an uncertainty reasoning method and an artificial intelligence and mode identification method according to the mathematical basis of the multi-source information fusion algorithm. Information fusion can be divided into data layer fusion, feature layer fusion and decision layer fusion according to the layer where the information source is processed. When the fusion methods are classified according to the fusion levels, a weighted average value or a clustering algorithm is mostly adopted in the data layer information fusion; the feature layer information fusion algorithm comprises a neural network, a fuzzy theory, a D-S evidence theory and the like; algorithms commonly used for the decision-level information fusion include D-S evidence theory, Bayes inference and fuzzy theory.
The estimation theory method comprises a linear estimation technology and a non-linear estimation technology. Common linear estimation techniques include kalman filtering and weighted averaging; nonlinear estimation techniques such as Extended Kalman Filter (EKF), Strong Tracking Filter (STF), and lossless Kalman Filter (UKF), etc. Along with the deepening of the cognition degree of information fusion, the research of a nonlinear non-Gaussian system based on a random sampling technology also obtains a plurality of results.
The uncertain reasoning method aims to realize information processing according to uncertain information and is used for realizing identification, attribute judgment and the like of target identities. The uncertain reasoning method mainly comprises a subjective Bayesian method, a D-S evidence reasoning method, a DSmT method, a fuzzy mathematic theory method, a possibility reasoning method and the like. The subjective Bayesian method requires that possible decisions of the system are mutually independent, and is a high-efficiency information fusion method used in the early stage; the D-S evidence theory adopts a trust function as a measurement, so that the uncertain problem caused by the unknown can be better solved; the DSmT method is an extension of the traditional D-S theory, and by using trust functions in combination, any type of independent source can be expressed. Moreover, considering that the prior probability is not known under normal conditions in practical applications, the result is inaccurate when the set prior probability values do not match.
The artificial intelligence and mode method can process the imperfect data, and the imperfect information is finally integrated into more uniform and perfect information through continuous learning and induction in the information processing process. The main method comprises a rough set theory, a random theory, an information entropy theory, a gray system theory, a Bayesian network, a neural network, a genetic algorithm and the like.
Disclosure of Invention
The invention aims to solve the technical problem of providing a dynamic self-adaptive multi-radar information fusion method, which obtains better fusion effect in a complex actual scene, greatly overcomes the problems of false alarm, missing report and error of a sensor and improves the output precision.
In order to solve the technical problems, the invention adopts the technical scheme that:
a dynamic self-adaptive multi-radar information fusion method comprises the following steps:
step 1: receiving preprocessing: receiving track data sent by each radar source, and preprocessing the received original data, wherein the preprocessing comprises the following steps:
1) numerical value validity judgment
Figure BDA0002401770110000021
||pos-center||2Indicating the distance between the radar track position and the fusion center, and setting an allowable range for an epsilon finger;
2) numerical value out-of-range determination
Figure BDA0002401770110000031
is _ valid () is a value out-of-bounds overflow judgment, and when the received data exceeds the representation range of the data used by the computer, the overflow of the data is generated; the data overflow condition can occur when the data value exceeds the limit of the word length of the computer, and the overflow type is divided into positive overflow and negative overflow; positive overflow indicates that the data value is greater than the maximum value of the demonstration range of the computer table, and negative overflow indicates that the data value is less than the minimum value of the demonstration range of the computer table;
3) removing ambiguity conflicts, wherein if reports appear for multiple times in the same period and the same batch number of the radar source, only the latest report of the time is retained, and other ambiguity reports are abandoned;
4) forming internal track data, updating or creating the data according to the channel number and the batch number after the judgment of the previous step, and updating the existing internal track by using new data when the data are found out through searching of the channel number and the batch number; if no internal track meeting the conditions exists, establishing an internal track, performing fusion track association by taking the distance as a scale, regarding the fusion track as associable when the distance difference between the fusion track and any fusion track is less than a set association threshold lambda, and finally determining the association relation to follow the principle of selecting nearest neighbor, wherein the reference value of lambda is the product of the fusion period and the uniform velocity of the detection target; in the period, the internal track of new data is not received, and the state of the internal track is updated to be extrapolation waiting;
step 2: quality evaluation;
carrying out quality evaluation on the preprocessed fusion system internal track data to obtain quality factors Q (i) ═ Σ (f) (i), F (i), d (i), s (i)), and obtaining the weight value of the period reflecting radar source and data characteristics thereof on the basis
Figure BDA0002401770110000032
(i, j belongs to A), wherein A represents an internal track set which is related to the same fusion track and survives from different radar sources in the same processing period, i is the serial number of the internal track reported by the radar source to be evaluated, and j is the serial number of the track reported by all the radar sources in the set A;
the four sub-item weight calculation modes for forming the quality factor Q (i) in the period are as follows:
1) obtaining a frequency weight f (i) according to the updated frequency of the sensor
Figure BDA0002401770110000041
I.e. the cumulative update times count of the internal track iupdate(i) Divided by the number of fused system cycles countsysObtaining; countupdate(i) Initializing to 0 after the internal track is established and starting statistics, and receiving new track data, count, sent by the information source represented by the internal track each timeupdate(i) Increasing by 1; countsysCounting after the self-fusion system is started, and counting every time a system period is passedsysThe value is increased by 1;
2) evaluating the distance weight d (i) of the credibility of the data according to the spatial distance difference
Figure BDA0002401770110000042
d(i)∈[0,dmax]Wherein d ismaxRepresents the upper limit of the set distance weight, dmaxIn generalTaking the value as 1; dis (i) represents the Euclidean distance of the position after the internal track filtering and the associated fused track prediction position; k represents that the fusion system sets an outlier distance threshold, and the value of K refers to a correlation threshold lambda to ensure that K is less than or equal to lambda.
3) Fitting weight F (i) for evaluating the degree of spatial dispersion in the period by using a fitting curve
F(i)＝max(Rx,Ry),F(i)∈[0,1]，
Figure BDA0002401770110000043
SSE refers to the sum of the remaining squares, which is the sum of the squares of the differences between the radar reported position and the fitted estimated position range; SSR refers to the regression sum of squares, which is the sum of squared deviations of the regression values of the reported positions and the average value of the reported positions;
4) judging the stability weight s (i) of the course change degree according to the recent history
s(i)＝g(θ),s(i)∈[smin,smax]
Figure BDA0002401770110000044
Wherein, the course approximate included angle theta is the absolute value of the difference between the course of the last period and the latest course of the period, a is the set allowable maximum variation angle, smaxTo set the upper limit of the weight, sminSetting a weight lower limit;
and step 3: dynamic adaptive weighted fusion: carrying out weighted average on the preprocessed system internal track data in combination with the weight obtained by the quality evaluation to obtain a periodic preliminary prediction fusion result
Figure BDA0002401770110000045
Wherein the internal track is predicted at the current period and fused with the track position
Figure BDA0002401770110000046
The generation method of (a) is expressed as:
Figure BDA0002401770110000051
the estimated fusion track position w (i) of the internal track in the period is the filtering position of the internal track obtained after the filtering treatment is carried out on the newly reported track of the sensor in the period
Figure BDA0002401770110000052
Output position of latest merged track with last cycle
Figure BDA0002401770110000053
After taking balance and accepting the trade-off, the balance proportion is set ad, and the ad belongs to [0,1 ]](ii) a Last cycle newest fusion track output position
Figure BDA0002401770110000054
Obtaining and backfilling after overall evaluation correction and filtering in the previous fusion period;
after that, the overall period situation composed of the ratio of the actual weight sum to the desired weight sum is determined
Figure BDA0002401770110000055
Qbase＝∑(fmax,Fmax,dmax,smax)
Wherein Q isbaseIs the sum of the maximum values of the four weights, fmax,Fmax,dmax,smaxThe maximum value, count, of the weight of the four sub-items of the quality factor Q (i) respectivelyeffThe number of internal tracks in the non-extrapolation waiting state in the period is shown; qty shows the track quality of the full sensor in this period, the larger the value Qty is, the better the overall quality is; further estimating the data quality of each sensor in the period according to Qty, and performing overall estimation on the weighted average preliminary prediction fusion result to obtain the result of the fusion track position after estimation
Figure BDA0002401770110000056
As a result of optimized fusion, wherein
Figure BDA0002401770110000057
And predicting the position of the fused track based on the motion state in the current period.
Further, the method also comprises the step 4: filtering, correcting and backfilling;
the filtering is embedded in the dynamic self-adaptive weighting fusion process, and the filtering operation is carried out on the internal flight path and the fusion result by utilizing a standard KF filter; filtering for merged tracks
Figure BDA0002401770110000061
Figure BDA0002401770110000062
The final output result of the fusion track in the period is obtained;
backfilling is carried out after weighted average is finished, and the track result is fused in the period
Figure BDA0002401770110000063
Giving it an associated internal track
Figure BDA0002401770110000064
Compared with the prior art, the invention has the beneficial effects that: the method can process data of multiple homologous/heterologous sensors, simultaneously obtains better fusion effect in a complex actual scene, greatly overcomes the problems of false alarm, missing report and error of the sensors, and improves the output precision. The fusion system needs less manual intervention, the setting work is completed before the fusion system is started, and no additional intervention operation is needed in the operation process.
Drawings
FIG. 1 is a process flow diagram of a dynamic adaptive multi-radar information fusion method of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
The method comprises the steps of receiving preprocessing, quality evaluation, dynamic self-adaptive weighting fusion, filtering correction, backfilling and the like, and specifically comprises the following steps:
firstly, receiving preprocessing: and receiving track data sent by each radar source, and preprocessing the received original data.
1) Numerical value validity judgment
Figure BDA0002401770110000065
||pos-center||2The distance between the radar track position and the fusion center is indicated, the epsilon indicates a set allowable range, the epsilon value is flexible, the common epsilon is an air condition monitoring range, for example, an air condition center is responsible for monitoring the air condition in a square circle of 50km, and the epsilon can be set to 50000.
2) Numerical value out-of-range determination
Figure BDA0002401770110000071
is _ valid () is a value out-of-bounds overflow judgment, and when received data exceeds the representation range of data used by a computer, overflow of data occurs. The data overflow condition can occur when the data value exceeds the limit of the word length of the computer, and the overflow type is divided into positive overflow and negative overflow. Positive overflow indicates that the data value is greater than the maximum value of the computer table range, and negative overflow indicates that the data value is less than the minimum value of the computer table range.
3) And (4) eliminating ambiguity conflicts, wherein if multiple reports appear in the same batch number of the same radar source in the same period, only the latest report is kept, and other ambiguity reports are abandoned.
4) Forming internal track data, updating or creating the data according to the channel number and the batch number after the judgment of the previous step, and updating the existing internal track by using new data when the data are found out through searching of the channel number and the batch number; if no internal track meeting the conditions exists, establishing an internal track, performing fusion track association by taking the distance as a scale, regarding the fusion track as associable when the distance difference between the fusion track and any fusion track is less than a set association threshold lambda, and finally determining the association relation to follow the principle of selecting nearest neighbor, wherein the reference value of lambda is the product of the fusion period and the uniform velocity of the detection target; this period does not receive an already internal track of new data, updating its state to extrapolation wait.
And II, evaluating the quality.
Carrying out quality evaluation on the preprocessed fusion system internal track data to obtain quality factors Q (i) ═ Σ (f) (i), F (i), d (i), s (i)), and obtaining the weight value of the period reflecting radar source and data characteristics thereof on the basis
Figure BDA0002401770110000072
A represents the set of internal tracks from different radar sources surviving in the same processing cycle associated with the same fusion track.
The four sub-item weight calculation modes for forming the quality factor Q (i) in the period are as follows:
1) obtaining a frequency weight f (i) according to the updated frequency of the sensor
Figure BDA0002401770110000073
I.e. the cumulative update times count of the internal track iupdate(i) Divided by the number of fused system cycles countsysThus obtaining the product. countupdate(i) Initializing to 0 after the internal track is established and starting statistics, and receiving new track data, count, sent by the information source represented by the internal track each timeupdate(i) Increasing by 1; countsysCounting after the self-fusion system is started, and counting every time a system period is passedsysThe value is increased by 1.
2) Evaluating the distance weight d (i) of the credibility of the data according to the spatial distance difference
Figure BDA0002401770110000081
d(i)∈[0,dmax]Wherein d ismaxRepresents the upper limit of the set distance weight, dmaxUsually taken as 1; dis (i) represents the Euclidean distance of the position after the internal track filtering and the associated fused track prediction position; k represents that the fusion system sets an outlier distance threshold, and the value of K refers to a correlation threshold lambda to ensure that K is less than or equal to lambda.
3) Fitting weight F (i) for evaluating the degree of spatial dispersion in the period by using a fitting curve
F(i)＝max(Rx,Ry),F(i)∈[0,1]，
Figure BDA0002401770110000082
SSE refers to the sum of the remaining squares, which is the sum of the squares of the differences between the radar reported position and the fitted estimated position range, i.e., the sum of the squares of the error terms; SSR refers to the regression sum of squares, which is the sum of squared deviations of the reported position regression values and the reported position mean. The specific calculation method refers to a classical mathematical formula.
4) Judging the stability weight s (i) of the course change degree according to the recent history
s(i)＝g(θ),s(i)∈[smin,smax]
Figure BDA0002401770110000083
Wherein, the course approximate included angle theta is the absolute value of the difference between the course of the last period and the latest course of the period, a is the set allowable maximum variation angle, smaxTo set the upper limit of the weight, sminSetting a weight lower limit; general smaxTaking the value 2, sminValue 0, a value range [0,90 ]]。
Thirdly, dynamic self-adaptive weighted fusion, namely, carrying out weighted average on the preprocessed system internal track data in combination with the weight obtained by the quality evaluation to obtain a periodic preliminary prediction fusion result
Figure BDA0002401770110000084
Wherein the internal track is predicted at the current period and fused with the track position
Figure BDA0002401770110000085
The generation method of (a) is expressed as:
Figure BDA0002401770110000091
the estimated fusion track position w (i) of the internal track in the current period is obtained after filtering treatment by combining the new report track of the current period of the sensorPartial track filter position
Figure BDA0002401770110000092
Output position of latest merged track with last cycle
Figure BDA0002401770110000093
After taking balance and accepting the trade-off, the balance proportion is set ad, and the ad belongs to [0,1 ]]Wherein ad < 0.5 represents the estimated deviation to the report position, ad 0.5 represents the estimated deviation to the key point of the report and fusion position, and ad > 0.5 represents the estimated deviation to the latest position of the fusion track.
Last cycle newest fusion track output position
Figure BDA0002401770110000094
Obtaining and backfilling after overall evaluation correction and filtering in the previous fusion period; after that, the overall period situation composed of the ratio of the actual weight sum to the desired weight sum is determined
Figure BDA0002401770110000095
Qbase＝∑(fmax,Fmax,dmax,smax)
Wherein Q isbaseIs the sum of the four maximum values of the weights, counteffThe number of internal tracks in the non-extrapolation waiting state in the period is shown; qty shows the total sensor track quality at this period, the larger the value Qty, the better the overall quality. Further estimating the data quality of each sensor in the period according to Qty, and performing overall estimation on the weighted average preliminary prediction fusion result to obtain the result of the fusion track position after estimation
Figure BDA0002401770110000096
As a result of optimized fusion, wherein
Figure BDA0002401770110000097
And predicting the position of the fused track based on the motion state in the current period.
Fourthly, filtering correction and backfilling, wherein filtering is carried outAnd in the dynamic self-adaptive weighting fusion process, the standard KF filter is used for filtering the internal flight path and the fusion result. Filtering for merged tracks
Figure BDA0002401770110000101
The final output result of the fusion track in the period is obtained, and the filtering work is to ensure the stability of the internal track data and the fusion result so as to improve the processing effect of the method;
backfilling is carried out after weighted average is finished, and the track result is fused in the period
Figure BDA0002401770110000102
Giving it an associated internal track
Figure BDA0002401770110000103

Claims (2)

1. A dynamic self-adaptive multi-radar information fusion method is characterized by comprising the following steps:
step 1: receiving preprocessing: receiving track data sent by each radar source, and preprocessing the received original data, wherein the preprocessing comprises the following steps:
1) numerical value validity judgment
Figure FDA0003249762740000011
||pos-center||2Indicating the distance between the radar track position and the fusion center, and setting an allowable range for an epsilon finger;
2) numerical value out-of-range determination
Figure FDA0003249762740000012
is _ valid () is a value out-of-bounds overflow judgment, and when the received data exceeds the representation range of the data used by the computer, the overflow of the data is generated; the data overflow condition can occur when the data value exceeds the limit of the word length of the computer, and the overflow type is divided into positive overflow and negative overflow; positive overflow indicates that the data value is greater than the maximum value of the computer table demonstration range, and negative overflow indicates that the data value is less than the computer expression(iii) minimum value of range;
3) removing ambiguity conflicts, wherein if reports appear for multiple times in the same period and the same batch number of the radar source, only the latest report of the time is retained, and other ambiguity reports are abandoned;
4) forming internal track data, updating or creating the data according to the channel number and the batch number after the judgment of the previous step, and updating the existing internal track by using new data when the data are found out through searching of the channel number and the batch number; if no internal track meeting the conditions exists, establishing an internal track, performing fusion track association by taking the distance as a scale, regarding the fusion track as associable when the distance difference between the fusion track and any fusion track is less than a set association threshold lambda, and finally determining the association relation to follow the principle of selecting nearest neighbor, wherein the reference value of lambda is the product of the fusion period and the uniform velocity of the detection target; in the period, the internal track of new data is not received, and the state of the internal track is updated to be extrapolation waiting;
step 2: quality evaluation;
carrying out quality evaluation on the preprocessed fusion system internal track data to obtain quality factors Q (i) ═ Σ (f) (i), F (i), d (i), s (i)), and obtaining the weight value of the period reflecting radar source and data characteristics thereof on the basis
Figure FDA0003249762740000021
i, j belongs to A, A represents an internal track set which is related to the same fusion track and survives from different radar sources in the same processing period, i is the serial number of the internal track reported by the radar source to be evaluated, and j is the serial number of the track reported by all the radar sources in the set A;
the four sub-item weight calculation modes for forming the quality factor Q (i) in the period are as follows:
1) obtaining a frequency weight f (i) according to the updated frequency of the sensor
Figure FDA0003249762740000022
I.e. the cumulative update times count of the internal track iupdate(i) Divided by the number of fused system cycles countsysObtaining; countupdate(i) Initial after internal track establishmentChanging to 0 and starting statistics, and receiving new track data, count, sent by the information source represented by the internal track each timeupdate(i) Increasing by 1; countsysCounting after the self-fusion system is started, and counting every time a system period is passedsysThe value is increased by 1;
2) evaluating the distance weight d (i) of the credibility of the data according to the spatial distance difference
Figure FDA0003249762740000023
d(i)∈[0,dmax]Wherein d ismaxRepresents the upper limit of the set distance weight, dmaxUsually taken as 1; dis (i) represents the Euclidean distance of the position after the internal track filtering and the associated fused track prediction position; k represents that the fusion system sets an outlier distance threshold, and the value of K refers to a correlation threshold lambda to ensure that K is less than or equal to lambda;
3) fitting weight F (i) for evaluating the degree of spatial dispersion in the period by using a fitting curve
F(i)＝max(Rx,Ry),F(i)∈[0,1]，
Figure FDA0003249762740000024
SSE refers to the sum of the remaining squares, which is the sum of the squares of the differences between the radar reported position and the fitted estimated position range; SSR refers to the regression sum of squares, which is the sum of squared deviations of the regression values of the reported positions and the average value of the reported positions;
4) judging the stability weight s (i) of the course change degree according to the recent history
s(i)＝g(θ),s(i)∈[smin,smax]
Figure FDA0003249762740000025
Wherein, the course approximate included angle theta is the absolute value of the difference between the course of the last period and the latest course of the period, a is the set allowable maximum variation angle, smaxTo set the upper limit of the weight, sminSetting a weight lower limit;
and step 3: dynamic adaptive weighted fusion: carrying out weighted average on the preprocessed system internal track data in combination with the weight obtained by the quality evaluation to obtain a periodic preliminary prediction fusion result
Figure FDA0003249762740000031
Wherein the internal track is predicted at the current period and fused with the track position
Figure FDA0003249762740000032
The generation method of (a) is expressed as:
Figure FDA0003249762740000033
the estimated fusion track position w (i) of the internal track in the period is the filtering position of the internal track obtained after the filtering treatment is carried out on the newly reported track of the sensor in the period
Figure FDA0003249762740000034
Output position of latest merged track with last cycle
Figure FDA0003249762740000035
After taking balance and accepting the trade-off, the balance proportion is set ad, and the ad belongs to [0,1 ]](ii) a Last cycle newest fusion track output position
Figure FDA0003249762740000036
Obtaining and backfilling after overall evaluation correction and filtering in the previous fusion period;
after that, the overall period situation composed of the ratio of the actual weight sum to the desired weight sum is determined
Figure FDA0003249762740000037
i∈A,Qty∈[0,1]
Qbase＝∑(fmax,Fmax,dmax,smax)
Wherein Q isbaseIs the sum of the maximum values of the four weights, fmax,Fmax,dmax,smaxThe maximum value, count, of the weight of the four sub-items of the quality factor Q (i) respectivelyeffThe number of internal tracks in the non-extrapolation waiting state in the period is shown; qty shows the track quality of the full sensor in this period, the larger the value Qty is, the better the overall quality is; further estimating the data quality of each sensor in the period according to Qty, and performing overall estimation on the weighted average preliminary prediction fusion result to obtain the result of the fusion track position after estimation
Figure FDA0003249762740000041
As a result of optimized fusion, wherein
Figure FDA0003249762740000042
And predicting the position of the fused track based on the motion state in the current period.
2. The dynamic adaptive multi-radar information fusion method according to claim 1, further comprising the step of 4: filtering, correcting and backfilling;
the filtering is embedded in the dynamic self-adaptive weighting fusion process, and the filtering operation is carried out on the internal flight path and the fusion result by utilizing a standard KF filter; filtering for merged tracks
Figure FDA0003249762740000043
Figure FDA0003249762740000044
The final output result of the fusion track in the period is obtained;
backfilling is carried out after weighted average is finished, and the track result is fused in the period
Figure FDA0003249762740000045
Giving it an associated internal track
Figure FDA0003249762740000046
CN202010148975.4A 2020-03-05 2020-03-05 Dynamic self-adaptive multi-radar information fusion method Active CN111239725B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010148975.4A CN111239725B (en) 2020-03-05 2020-03-05 Dynamic self-adaptive multi-radar information fusion method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010148975.4A CN111239725B (en) 2020-03-05 2020-03-05 Dynamic self-adaptive multi-radar information fusion method

Publications (2)

Publication Number Publication Date
CN111239725A CN111239725A (en) 2020-06-05
CN111239725B true CN111239725B (en) 2022-02-15

Family

ID=70873315

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010148975.4A Active CN111239725B (en) 2020-03-05 2020-03-05 Dynamic self-adaptive multi-radar information fusion method

Country Status (1)

Country Link
CN (1) CN111239725B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112580727A (en) * 2020-12-22 2021-03-30 广西壮族自治区计量检测研究院 Detection data transmission method and device based on data fusion

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103926584A (en) * 2014-04-30 2014-07-16 电子科技大学 Space-frequency-polarization combined cooperation detection method
CN104280723A (en) * 2014-10-22 2015-01-14 四川大学 Method for processing newly built system track in radar clutter area
CN106054194A (en) * 2016-05-10 2016-10-26 南京信息工程大学 Spaceborne radar and ground-based radar reflectivity factor data three dimensional fusion method
CN107192998A (en) * 2017-04-06 2017-09-22 中国电子科技集团公司第二十八研究所 A kind of adapter distribution track data fusion method based on covariance target function
CN108140323A (en) * 2015-08-03 2018-06-08 大众汽车有限公司 For the method and apparatus of improved data fusion during environment measuring in motor vehicle
CN110596693A (en) * 2019-08-26 2019-12-20 杭州电子科技大学 Multi-sensor GMPHD self-adaptive fusion method with iterative updating

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7940206B2 (en) * 2005-04-20 2011-05-10 Accipiter Radar Technologies Inc. Low-cost, high-performance radar networks

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103926584A (en) * 2014-04-30 2014-07-16 电子科技大学 Space-frequency-polarization combined cooperation detection method
CN104280723A (en) * 2014-10-22 2015-01-14 四川大学 Method for processing newly built system track in radar clutter area
CN108140323A (en) * 2015-08-03 2018-06-08 大众汽车有限公司 For the method and apparatus of improved data fusion during environment measuring in motor vehicle
CN106054194A (en) * 2016-05-10 2016-10-26 南京信息工程大学 Spaceborne radar and ground-based radar reflectivity factor data three dimensional fusion method
CN107192998A (en) * 2017-04-06 2017-09-22 中国电子科技集团公司第二十八研究所 A kind of adapter distribution track data fusion method based on covariance target function
CN110596693A (en) * 2019-08-26 2019-12-20 杭州电子科技大学 Multi-sensor GMPHD self-adaptive fusion method with iterative updating

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
High-level Information Fusion:An Overview;PEK HUI FOO;《JOURNAL OF ADVANCES IN INFORMATION FUSION》;20131231;第01卷(第08期);第33-72页 *
一种多雷达航迹加权融合的权值动态分配算法;黄友澎等;《计算机应用》;20080901;第28卷(第09期);第2452-2454页 *
多源航迹信息融合主要技术研究;杨晓丹 等;《研究与开发》;20170331;第8-11页 *

Also Published As

Publication number Publication date
CN111239725A (en) 2020-06-05

Similar Documents

Publication Publication Date Title
CN107092582B (en) Online abnormal value detection and confidence evaluation method based on residual posterior
CN112946624B (en) Multi-target tracking method based on track management method
CN110109095B (en) Target feature assisted multi-source data association method
CN110927712B (en) Tracking method and device
CN111581832B (en) Improved gray Elman neural network hovercraft motion prediction method based on ARMA model correction
CN111239725B (en) Dynamic self-adaptive multi-radar information fusion method
Hung et al. Modified PSO Algorithm on Recurrent Fuzzy Neural Network for System Identification.
CN111917785A (en) Industrial internet security situation prediction method based on DE-GWO-SVR
CN109460608B (en) High and steep slope deformation prediction method based on fuzzy time sequence
CN110849372A (en) Underwater multi-target track association method based on EM clustering
CN116689503A (en) Strip steel full-length thickness prediction method based on memory function network
CN112381334A (en) Method for predicting deformation trend of high and steep slope based on multi-factor fuzzy time sequence
CN114444389B (en) Air attack target dynamic threat assessment method based on combined weighting and improved VIKOR
CN110011847B (en) Data source quality evaluation method under sensing cloud environment
CN105223559B (en) A kind of long-range radar track initiation method that can switch parallel
CN111142101B (en) Data association method
Juang et al. Automatic construction of feedforward/recurrent fuzzy systems by clustering-aided simplex particle swarm optimization
CN113255963A (en) Road surface use performance prediction method based on road element splitting and deep learning model LSTM
CN111340853B (en) Multi-sensor GMPHD self-adaptive fusion method based on OSPA iteration
CN107979606A (en) It is a kind of that there is adaptive distributed intelligence decision-making technique
CN112907975A (en) Detection method for abnormal parking based on millimeter wave radar and video
CN115932913B (en) Satellite positioning pseudo-range correction method and device
CN108922191B (en) Travel time calculation method based on soft set
CN111328015B (en) Wireless sensor network target tracking method based on Fisher information distance
Havangi Intelligent FastSLAM: An intelligent factorized solution to simultaneous localization and mapping

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