CN112861262A - Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier - Google Patents

Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier Download PDF

Info

Publication number
CN112861262A
CN112861262A CN202110181883.0A CN202110181883A CN112861262A CN 112861262 A CN112861262 A CN 112861262A CN 202110181883 A CN202110181883 A CN 202110181883A CN 112861262 A CN112861262 A CN 112861262A
Authority
CN
China
Prior art keywords
trend
self
function
basic
combination
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.)
Pending
Application number
CN202110181883.0A
Other languages
Chinese (zh)
Inventor
沈凯
刘庭欣
邓志红
汪进文
付梦印
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.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
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 Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN202110181883.0A priority Critical patent/CN112861262A/en
Publication of CN112861262A publication Critical patent/CN112861262A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • 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
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computational Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Optimization (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Geometry (AREA)
  • Algebra (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • Databases & Information Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Evolutionary Biology (AREA)
  • Operations Research (AREA)
  • Automation & Control Theory (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Computing Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The self-organizing data-driven modeling method for trajectory prediction of highly dynamic carriers of the present disclosure, S1: selecting a basic trend term function set according to the trajectory of the high dynamic carrier; s2: combining the basic trend term functions pairwise to obtain an initial model according to a K-G polynomial; s3: the basic trend item functions are crossed and combined according to the initial module to generate a current layer inductive trend item function and a current layer memory trend item function; s4: utilizing self-organizing preferred judgment criteria to summarize the current layer and screen the memory trend item function; s5: repeating S3 and S4 to obtain a set of induction and memory trend item functions when the iteration times are met; s6: and (4) performing intersection, combination and preferential judgment on the trend item functions in the induction and memory trend item function set and the initial basic trend item function set to screen out an optimal model. The GNSS navigation data can be predicted when the GNSS is unlocked, the trajectory of the high dynamic carrier is obtained by combining the GNSS navigation data with the INS, and the problems of trajectory information loss and positioning failure caused by GNSS unlocking are solved.

Description

Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier
Technical Field
The disclosure belongs to the technical field of navigation and positioning, and particularly relates to a self-organizing data-driven modeling method for trajectory prediction of a high dynamic carrier.
Background
In recent years, with the continuous emergence of high-speed and high-dynamic motion carriers (such as high-speed aircrafts like cannonballs) and the like, the high-dynamic carriers are required to have the capabilities of real-time precise positioning and continuous robust navigation, so that the research on the navigation and positioning of the carriers in a high-dynamic environment is of great significance. The GNSS/INS integrated navigation mode is the most common navigation mode currently because it can provide all-weather and continuous navigation information. Then, under complex environments such as high dynamics and strong interference, a large doppler frequency is generated on a GNSS carrier, which easily causes tracking lock losing of GNSS signals, and the GNSS signals are easily shielded, interfered and deceived, so that a navigation positioning system has risks of discontinuity, unreliability or unreliability, and even completely loses positioning capability.
At present, trajectory prediction is mainly carried out by establishing a motion model of a high-dynamic carrier, but different carrier motion models are different, and some carriers are difficult to model, so that a set of universal trajectory prediction method is difficult to establish.
Therefore, a self-organizing data driven modeling method is provided, a specific motion model does not need to be established, GNSS navigation data can be predicted through the model prediction method in the GNSS unlocking time, and the trajectory track of the high dynamic carrier is obtained after the GNSS navigation data is combined with the INS, so that the problems of trajectory track information loss and positioning failure caused by GNSS unlocking can be compensated in a short time.
Disclosure of Invention
In view of the above, the present disclosure provides a self-organizing data-driven modeling method for trajectory prediction of a high dynamic carrier, which solves the problem that in a complex, open, and dynamic scene, the high dynamic carrier is affected by various active and passive random interferences inside and outside, and particularly, navigation and positioning of the high dynamic carrier are discontinuous and unreliable under the condition of GNSS lock loss.
According to an aspect of the present disclosure, there is provided a self-organizing data-driven modeling method for trajectory prediction of highly dynamic carriers, the method comprising:
s1: selecting a basic trend function set of the self-organizing data according to trajectory prediction of the high dynamic carrier, wherein the basic trend function set is an initial basic trend function set;
s2: combining the basic trend term functions pairwise according to a K-G polynomial to obtain the self-organization data driven initial model;
s3: the basic trend item functions are crossed and combined according to the initial model to generate a generalized trend item function combination of the current layer and a memory trend item function combination of the current layer;
s4: screening the inductive trend term function combination and the memory trend term function combination of the current layer by utilizing a self-organization preferred judgment composite criterion to obtain a basic trend term function set of the next layer;
s5: taking the next layer of basic trend item function set obtained in the step S4 as the basic trend item function in the step S3, repeating the step S3 and the step S4, and obtaining a generalized trend item function combination set and a memory trend item function combination set when the iteration times are met;
s6: and performing intersection, self-combination and self-organization preferred judgment composite criterion evaluation screening on all trend item functions in the induction trend item function set, the memory trend item function set and the initial basic trend item function set to obtain the self-organization data-driven optimal model.
In one possible implementation, the method further includes: and calculating a set of parameters with minimum deviation by using the input original training data as the parameters of the basic trend term function of the self-organization data.
In one possible implementation, the set of basic trend term functions includes a constant trend term, a power polynomial trend term, a trigonometric function trend term, and an exponential function trend term.
In one possible implementation, the generating the generalized trend term function combination of the current layer by crossing and linearly self-combining the basic trend term functions includes:
and performing intersection, iteration and linear combination on the basic trend term functions to generate the inductive trend term function combination of the current layer.
In one possible implementation, the self-organizing preferential decision composite criterion includes: a regularity criterion, a balance criterion, and a simple row criterion;
the form of the self-organizing preferential judgment composite criterion is as follows:
Figure BDA0002942359170000031
wherein i is a positive integer, SECiIndicating the ith criterion used, ciExpress SECiWeight occupied in the self-organizing preferential decision composite criterion CSEC, eiIndicating the use criterion SECiThe resulting prediction error value is calculated.
In one possible implementation, the screening the generalized trend term function combination and the memory trend term function combination of the current layer by using a self-organizing preferred decision composite criterion includes:
and calculating the competitiveness of the trend item function in the inductive trend item function combination and the memory trend item function combination of the current layer by utilizing the self-organization preferred judgment composite criterion, reserving the trend item function with strong competitiveness, and eliminating the trend item function with weak competitiveness.
In one possible implementation, the K-G polynomial is of the form:
Figure BDA0002942359170000032
wherein x isiIs a variable of a K-G polynomial, ai…ai*j*…*mIs the coefficient of K-G polynomial, i and m are positive integers, and i is 1 … m.
The self-organizing data driven modeling method disclosed by the invention comprises the following steps of S1: selecting a basic trend function set of the self-organizing data according to trajectory prediction of the high dynamic carrier, wherein the basic trend function set is an initial basic trend function set; s2: combining the basic trend term functions pairwise according to a K-G polynomial to obtain the self-organization data driven initial model; s3: the basic trend item functions are crossed and combined according to the initial model to generate a generalized trend item function combination of the current layer and a memory trend item function combination of the current layer; s4: screening the inductive trend term function combination and the memory trend term function combination of the current layer by utilizing a self-organization preferred judgment composite criterion to obtain a basic trend term function set of the next layer; s5: taking the next layer of basic trend item function set obtained in the step S4 as the basic trend item function in the step S3, repeating the step S3 and the step S4, and obtaining a generalized trend item function combination set and a memory trend item function combination set when the iteration times are met; s6: and performing intersection, self-combination and self-organization preferred judgment composite criterion evaluation screening on all trend item functions in the induction trend item function set, the memory trend item function set and the initial basic trend item function set to obtain the self-organization data-driven optimal model. Without establishing a specific motion model, GNSS navigation data can be predicted within the GNSS unlocking time, and the trajectory of the high dynamic carrier is obtained after the GNSS navigation data is combined with the INS, so that the problems of trajectory information loss and positioning failure caused by GNSS unlocking can be compensated within a short time.
Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.
Fig. 1 shows a flow diagram of a self-organizing data-driven modeling method for ballistic prediction of highly dynamic carriers according to an embodiment of the present disclosure;
figure 2 shows a schematic diagram of an ad-hoc data-driven modeling network structure for ballistic prediction of highly dynamic carriers according to an embodiment of the present disclosure;
FIG. 3 shows an X-plot of ballistic predictions for a highly dynamic vehicle based on a self-organizing data-driven model according to an embodiment of the present disclosure;
fig. 4 shows a Y-plot of ballistic predictions for a highly dynamic carrier based on a self-organizing data-driven model according to an embodiment of the disclosure.
Detailed Description
Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers can indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.
Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present disclosure.
The basic idea of the self-organizing data-driven modeling method is as follows: the model structure is selected in a self-organizing manner and model parameter estimation is carried out, modeling work can be completed in a self-organizing manner only by giving input and output of the model, GNSS navigation data can be predicted under the condition that GNSS signals are unlocked, a virtual redundant sensor is formed, the virtual GNSS position and speed signals acceptable in an error allowable range are provided, combined navigation is formed by the virtual GNSS position and speed signals and the position and speed signals calculated by SINS, and the function of predicting the trajectory of a carrier is achieved.
Fig. 1 shows a flow diagram of a self-organizing data-driven modeling method for ballistic prediction of highly dynamic carriers according to an embodiment of the present disclosure. As shown in fig. 1, the method may include:
s1: selecting a basic trend term function set of the self-organizing data according to the trajectory prediction of the high dynamic carrier, wherein the basic trend term function set is an initial basic trend term function set.
Wherein the basic trend term function set may include a constant trend term, a power polynomial trend term, a trigonometric function trend term, and an exponential function trend term.
The mathematical description of the dynamic object can be represented by a trend term function f (t, c), and f (t, c) can be approximated by a combination of some basic functions.
For example, a power polynomial trend term
Figure BDA0002942359170000051
Can be used to build models
Figure BDA0002942359170000052
The trigonometric function trend term may include a cosine function and a sine function, which may be expressed as:
Figure BDA0002942359170000053
Figure BDA0002942359170000054
the models of formula (4) can be combined from formulas (2) and (3):
Figure BDA0002942359170000055
trend term of exponential function
Figure BDA0002942359170000056
Can be used for forming combined model
Figure BDA0002942359170000057
The basic trend function terms can be arbitrarily combined into the basic trend function terms required by the self-organizing data-driven modeling, and for example, the basic trend function terms can be as follows: constant trend term a, first power trend term k1T, the quadratic power trend term k2*(p2*t)2Trend term k of sine function3*sin(p3T), exponential function trend term
Figure BDA0002942359170000058
And the like, all of the variables are natural numbers and are not defined one by one.
In addition, the method can also utilize the input original training data to calculate a set of parameters with the minimum deviation as the parameters of the basic trend term function of the self-organizing data. For example, the parameters a, k in each basic trend term function can be performed by using the input original training data1,k2,p2,k3,p3,k4,p4And (4) determining. And substituting a certain amount of randomly generated parameters into the basic trend term function, and selecting a group of parameters with the minimum calculation deviation by using the original training data as the parameters of the basic trend term function.
S2: and combining the basic trend term functions pairwise according to a K-G polynomial to obtain the self-organization data driven initial model.
Wherein the K-G (Kolmogorov-Gabor) polynomial is in the form of:
Figure BDA0002942359170000061
wherein x isiIs a variable of K-G polynomial (basic trend term function in step S1), ai…ai*j*…*mIs the coefficient of K-G polynomial, i and m are positive integers, and i is 1 … m.
For example, assume that there are two basic trend term functions x1And x2As input, the initial model driven by the self-organizing data is obtained as follows:
Figure BDA0002942359170000062
s3: and intersecting and combining the basic trend term functions according to the initial model to generate an inductive trend term function combination of the current layer and a memory trend term function combination of the current layer.
In one example, the basic trend term functions are crossed, iterated, and linearly combined to generate a generalized trend term function combination for the current layer.
For example, if there are two basic trend term functions f1(t) and f2(t), then the basic trend term function f1(t) and f2(t) the following combination modes can be obtained by performing intersection and iteration:
y1 *(t)＝f1(t) formula (7), y2 *(t)＝f2(t) formula (8),
y3 *(t)＝f1(f2(t)) formula (9), y4 *(t)＝f2(f1(t)) formula (10),
and (3) linearly self-combining the formulas (7), (8), (9) and (10) to obtain a generalized trend term function combination: y is*(t)＝y1 *(t)+y2 *(t)+y3 *(t)+y4 *(t) formula (11).
The memory method is obtained by self-combination according to the initial model in step S2, i.e. the K-G polynomial combination form, and the memory trend term function combination is:
Figure BDA0002942359170000071
s4: and screening the inductive trend term function combination and the memory trend term function combination of the current layer by utilizing a self-organization preferred judgment composite criterion to obtain a basic trend term function set of the next layer.
For example, if m represents the number of trend function of the current layer combination, then a single self-combination based on the trend function of the current layer will generate m (m-1)/2 combination terms, and similarly, if each layer contains m trend function numbers, then each self-combination will generate m (m-1)/2 combination terms. If the optimal combination trend item is obtained by screening only at the last layer, and the combination item in the middle process is not preferentially judged, the calculated amount is rapidly increased, and the algorithm is not favorably realized, so that the self-organization preferred judgment composite criterion is required to be used for screening after each self-combination.
The self-organizing preferential decision composite criterion may include a regularity criterion, a balance criterion, a simple row criterion, and the like, which are not limited herein.
Among the common regularity criteria are:
Figure BDA0002942359170000072
in the formula, nbIs the number of samples, yiRepresenting true value, yi(B) An output prediction value representing a regularity criterion.
The mean absolute error criterion is:
Figure BDA0002942359170000073
the unbiased criterion is:
Figure BDA0002942359170000074
the form of the self-organizing preferred decision composite criterion is as follows:
Figure BDA0002942359170000075
wherein the content of the first and second substances,i is a positive integer, SECiIndicating the ith criterion used, ciExpress SECiWeight occupied in the self-organizing preferential decision composite criterion CSEC, eiIndicating the use criterion SECiThe resulting prediction error value is calculated.
And calculating the input inductive trend term function combination and the memory trend term function combination of the current layer by utilizing a self-organizing preferred judgment composite criterion, and selecting a basic trend term function set with smaller matching error for the next layer of input trend function term combination.
S5: and (5) taking the basic trend term function set of the next layer obtained in the step (S4) as the basic trend term function of the step (S3), repeating the step (S3) and the step (S4), and obtaining a generalized trend term function combination set and a memory trend term function combination set when the iteration number is met. For example, a network structure of induction and memory in step 3 and a self-organizing preferred decision composite criterion in step 4 may be used to generate a structure network with multiple layers of induction and memory, and as the network structure increases, the structure of the combination trend item is more complex, and the amount of calculation increases, but the effect of self-combination increases insignificantly, and the multiple layers of network structures need to be induced and trained.
S6: and performing intersection, self-combination and self-organization preferred judgment composite criterion evaluation screening on all trend item functions in the induction trend item function set, the memory trend item function set and the initial basic trend item function set to obtain the self-organization data-driven optimal model.
Fig. 2 shows a schematic structural diagram of an ad-hoc data-driven modeling network for ballistic prediction of highly dynamic carriers according to an embodiment of the present disclosure.
As shown in fig. 2, in the last layer of the "induction + memorization" multilayer network structure, an optimal trend term set obtained by the "memorization" combination, and an initial basic trend term are obtained by respectively utilizing the "induction" combination, self-combination is performed on the network structure of the "induction + memorization" for the last time, and an optimal set of trend terms is obtained by utilizing a self-service person preferred decision composite criterion, so that a final self-organization data-driven optimal model is generated, that is, the output data of the last layer of the model is fitted into the self-organization data-driven optimal model.
Application example:
by utilizing the self-organizing data driven modeling method, the trajectory of a high dynamic carrier is predicted when the GNSS signal is unlocked. Firstly, according to the reference trajectory of the high dynamic carrier, the motion of the high dynamic carrier is simplified into particle plane motion, namely the change rule of the centroid coordinate (x, y), the centroid speed v and the inclination angle theta of the high dynamic carrier at the time t. One variable of the five quantities is selected as an independent variable, and the other four variables can be used as functions of the independent variable, so that the equation of the mass center motion of the high-dynamic carrier is obtained as follows:
Figure BDA0002942359170000091
wherein the content of the first and second substances,
Figure BDA0002942359170000092
and obtaining the reference trajectory of the high dynamic carrier, namely the true value of the coordinates x and y by combining a Runge Kutta method with Matlab programming. And then predicting the trajectory of the missing part of the GNSS signal by using the proposed self-organizing data driven model.
FIG. 3 shows an X-plot of ballistic predictions for a highly dynamic vehicle based on a self-organizing data-driven model according to an embodiment of the present disclosure; fig. 4 shows a Y-plot of ballistic predictions for a highly dynamic carrier based on a self-organizing data-driven model according to an embodiment of the disclosure.
For example, the basic trend term function selects a primary trend term, a secondary trend term, a sine function trend term and an exponential function trend term, the network is summarized to be 2, namely the calculation times of the induction combination and the memory combination are two layers, and the number of the trend terms of each layer of the network is not more than 9.
The self-organizing preferential judgment composite criterion is formed by utilizing a normative criterion, an average absolute error criterion and an unbiased criterion, and the corresponding weights of the self-organizing preferential judgment composite criterion are as follows: c. C1＝0.8,c2＝0.15,c3The results of the first self-combination were evaluated and the basic trend term function was screened to 0.05.
Wherein, the number of the training samples of the model accounts for 70% of the total number of the samples, and the number of the testing samples accounts for 30% of the total number of the samples. The prediction results of the reference trajectory of the high dynamic carrier are shown in fig. 3 and 4, and it can be known from comparison between the actual values of the X and y coordinates shown in fig. 3 and 4 and the predicted values of the model that within 60.75s, the trajectory of 18.25s is predicted, the maximum error of the X coordinate is 14.75m, and the average error is 7.23 m; the maximum error of the Y coordinate is 5.34m, the average error of the Y coordinate is 2.74m, and therefore ballistic prediction of a high-dynamic carrier is achieved.
The self-organization data driving modeling method adopts an organization form that input variables are combined with each other pairwise, and selects a model with a smaller matching error for combination of a next layer through a self-organization preferred judgment composite criterion. The self-organizing data-driven modeling method is characterized in that pre-designed trend items are utilized for combination, and the combination process is divided into two modes of induction and memory combination. Because the types and the number of the selected trend items are different, the forms of the mathematical models covered by the combined trend items obtained after combination are also different, theoretically, the more the trend items are selected in the algorithm, the more the combination types are, the better the fitting effect on the models is, but the calculation amount required to be considered is increased. How to reasonably select the type of trend term and the preferential decision criterion has a direct relationship to the accuracy of modeling and prediction and the amount of computation.
Having described embodiments of the present disclosure, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims (7)

1. A self-organizing data-driven modeling method for ballistic prediction of highly dynamic carriers, the method comprising:
s1: selecting a basic trend function set of the self-organizing data according to trajectory prediction of the high dynamic carrier, wherein the basic trend function set is an initial basic trend function set;
s2: combining the basic trend term functions pairwise according to a K-G polynomial to obtain the self-organization data driven initial model;
s3: the basic trend item functions are crossed and combined according to the initial model to generate a generalized trend item function combination of the current layer and a memory trend item function combination of the current layer;
s4: screening the inductive trend term function combination and the memory trend term function combination of the current layer by utilizing a self-organization preferred judgment composite criterion to obtain a basic trend term function set of the next layer;
s5: taking the next layer of basic trend item function set obtained in the step S4 as the basic trend item function in the step S3, repeating the step S3 and the step S4, and obtaining a generalized trend item function combination set and a memory trend item function combination set when the iteration times are met;
s6: and performing intersection, self-combination and self-organization preferred judgment composite criterion evaluation screening on all trend item functions in the induction trend item function set, the memory trend item function set and the initial basic trend item function set to obtain the self-organization data-driven optimal model.
2. The self-organizing data-driven modeling method of claim 1, further comprising: and calculating a set of parameters with minimum deviation by using the input original training data as the parameters of the basic trend term function of the self-organization data.
3. The self-organizing data-driven modeling method of claim 1, wherein the set of basic trend term functions includes a constant trend term, a power polynomial trend term, a trigonometric function trend term, and an exponential function trend term.
4. The self-organizing data-driven modeling method of claim 3, wherein the intersecting and linearly self-combining the basic trend term functions to generate a generalized trend term function combination of a current layer comprises:
and performing intersection, iteration and linear combination on the basic trend term functions to generate the inductive trend term function combination of the current layer.
5. The self-organizing data-driven modeling method of claim 1, wherein the self-organizing preferential decision composite criterion comprises: a regularity criterion, a balance criterion, and a simple row criterion;
the form of the self-organizing preferential judgment composite criterion is as follows:
CSEC＝c1SEC1+c2SEC2+…+ciSECi+…+cnSECn，
Figure FDA0002942359160000021
wherein i is a positive integer, SECiIndicating the ith criterion used, ciExpress SECiWeight occupied in the self-organizing preferential decision composite criterion CSEC, eiIndicating the use criterion SECiThe resulting prediction error value is calculated.
6. The self-organizing data-driven modeling method of claim 5, wherein the filtering of the inductive trend term function combination and the memory trend term function combination of the current layer by using a self-organizing preferred decision composite criterion comprises:
and calculating the competitiveness of the trend item function in the inductive trend item function combination and the memory trend item function combination of the current layer by utilizing the self-organization preferred judgment composite criterion, reserving the trend item function with strong competitiveness, and eliminating the trend item function with weak competitiveness.
7. The self-organizing data-driven modeling method of claim 1, wherein the K-G polynomial is of the form:
Figure FDA0002942359160000022
wherein x isiIs a variable of a K-G polynomial, ai…ai*j*…*mIs the coefficient of K-G polynomial, i and m are positive integers, and i is 1 … m.
CN202110181883.0A 2021-02-08 2021-02-08 Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier Pending CN112861262A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110181883.0A CN112861262A (en) 2021-02-08 2021-02-08 Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110181883.0A CN112861262A (en) 2021-02-08 2021-02-08 Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier

Publications (1)

Publication Number Publication Date
CN112861262A true CN112861262A (en) 2021-05-28

Family

ID=75989559

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110181883.0A Pending CN112861262A (en) 2021-02-08 2021-02-08 Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier

Country Status (1)

Country Link
CN (1) CN112861262A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001236341A (en) * 2000-02-23 2001-08-31 Denso Corp Estimation model structuring device, information processor, and recording medium
CN106529713A (en) * 2016-11-03 2017-03-22 华北电力大学（保定） Grey GMDH network combination model-based wind speed prediction method and system
CN111062516A (en) * 2019-11-20 2020-04-24 广西电网有限责任公司 Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing
CN111931287A (en) * 2020-07-06 2020-11-13 北京电子工程总体研究所 Near space hypersonic target trajectory prediction method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001236341A (en) * 2000-02-23 2001-08-31 Denso Corp Estimation model structuring device, information processor, and recording medium
CN106529713A (en) * 2016-11-03 2017-03-22 华北电力大学（保定） Grey GMDH network combination model-based wind speed prediction method and system
CN111062516A (en) * 2019-11-20 2020-04-24 广西电网有限责任公司 Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing
CN111931287A (en) * 2020-07-06 2020-11-13 北京电子工程总体研究所 Near space hypersonic target trajectory prediction method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
KAI SHEN 等: ""Algorithms of Constructing Models for Compensating Navigation Systems of Unmanned Aerial Vehicles"", 《2016 ICRAE》 *
庞晓波 等: ""变量组合（GMDH）方法"", 《经济预测与决策》 *

Similar Documents

Publication Publication Date Title
Gilholm et al. Poisson models for extended target and group tracking
Eberhardt et al. Density trees for efficient nonlinear state estimation
CN104103070A (en) Landing point selecting method based on optical images
CN109918718A (en) Unmanned plane during flying method for evaluating quality, device, storage medium and electronic equipment
Taha et al. Improving INS/GPS positioning accuracy during GPS outages using fuzzy logic
Lee et al. Radial basis function network‐based available measurement classification of interferometric radar altimeter for terrain‐aided navigation
Wolek et al. The Orbiting Dubins Traveling Salesman Problem: planning inspection tours for a minehunting AUV
CN114386466B (en) Parallel hybrid clustering method for candidate signal mining in pulsar search
Ma et al. Multi-AUV collaborative operation based on time-varying navigation map and dynamic grid model
Nayak et al. Fractal dimension-based generalized box-counting technique with application to grayscale images
CN113030940B (en) Multi-star convex type extended target tracking method under turning maneuver
CN105021199A (en) LS (Least square)-based multi- model adaptive state estimation method and system
CN112861262A (en) Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier
Soysal et al. Data fusion in a multistatic radar network using covariance intersection and particle filtering
Gallo et al. Fuzzy B-splines: A surface model encapsulating uncertainty
CN116026325A (en) Navigation method and related device based on neural process and Kalman filtering
Gong et al. Active Suspension Control Based on Estimated Road Class for Off‐Road Vehicle
CN113628254A (en) Target track determination method based on mobile platform and related equipment
CN109084751B (en) High-energy-efficiency satellite attitude determination algorithm based on box particle filter
You et al. UAV-Pose: A Dual Capture Network Algorithm for Low Altitude UAV Attitude Detection and Tracking
Rahmati et al. Maneuvering target tracking method based on unknown but bounded uncertainties
CN110472700A (en) A kind of parameter fill method and device based on convolutional neural networks
CN117647997B (en) Knowledge bidirectional migration unmanned aerial vehicle collaborative track local re-planning method and system
CN117930871B (en) Rotor flow unmanned aerial vehicle group real-time conflict resolution method
Sun et al. Terminal trajectory planning for synthetic aperture radar imaging guidance based on chronological iterative search framework

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
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210528