CN111382876A - Method and system for acquiring initial value of ground fire transfer orbit design based on evolutionary algorithm - Google Patents

Method and system for acquiring initial value of ground fire transfer orbit design based on evolutionary algorithm Download PDF

Info

Publication number
CN111382876A
CN111382876A CN202010130385.9A CN202010130385A CN111382876A CN 111382876 A CN111382876 A CN 111382876A CN 202010130385 A CN202010130385 A CN 202010130385A CN 111382876 A CN111382876 A CN 111382876A
Authority
CN
China
Prior art keywords
fire transfer
initial value
orbit
ground fire
track
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
CN202010130385.9A
Other languages
Chinese (zh)
Other versions
CN111382876B (en
Inventor
朱庆华
刘宇
黄韵弘
鲁启东
张玉花
冯建军
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.)
Shanghai Aerospace Control Technology Institute
Original Assignee
Shanghai Aerospace Control Technology Institute
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 Shanghai Aerospace Control Technology Institute filed Critical Shanghai Aerospace Control Technology Institute
Priority to CN202010130385.9A priority Critical patent/CN111382876B/en
Publication of CN111382876A publication Critical patent/CN111382876A/en
Application granted granted Critical
Publication of CN111382876B publication Critical patent/CN111382876B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method and a system for quickly acquiring an initial value of a ground fire transfer orbit design based on an evolutionary algorithm. Before the orbit is accurately transferred, the orbit initial value of the dominant direction is rapidly calculated based on the evolutionary algorithm by directly utilizing the previously established orbit database, so that the follow-up more efficient search can be supported. The method mainly comprises 3 steps: 1) according to the GTOP database, establishing a learning sample library by using the characteristic value of the initial value sensitive matrix of the model as a sample label; 2) constructing a track optimization support vector machine (TO _ SVM), and learning a ground fire transfer track learning sample library; 3) and (4) operating the TO _ SVM aiming at the ground fire transfer orbit TO be designed, and giving initial orbit design values (the emission date, the arrival date and the residual velocity of the emission orbit) of the dominant direction c for subsequent accurate design.

Description

Method and system for acquiring initial value of ground fire transfer orbit design based on evolutionary algorithm
Technical Field
The invention relates to a method and a system for quickly acquiring an initial value of a ground fire transfer orbit design based on an evolutionary algorithm. The method can quickly calculate the initial value of the orbit in the dominant direction before accurately transferring the orbit by adopting an indirect method and an evolutionary algorithm so as to carry out more efficient search in the following process.
Background
Because the relative positions and speeds of the earth and the mars are constantly changed, and the carrying capacity of the rocket is limited, the earth-fire transfer orbit parameters (the launching time, the arrival time, the launching energy, the arrival braking energy and the like) of the mars detector need to be optimally designed. Generally, firstly traversing all possible combinations of emission time and arrival time, selecting a proper orbit from all possible orbits by adopting a conic section splicing method as a design initial value, and then optimizing by adopting an accurate dynamic model to obtain a final accurate ground fire transfer orbit solution. The method for searching the design initial value by traversing and adopting the conical curve splicing method has large calculation amount and cannot be applied to the spacecraft.
Disclosure of Invention
The invention aims to: the method can quickly provide a plurality of orbit design initial values (emission date and arrival date) in the dominant direction for subsequent accurate design and use under the condition of no traversal. The method can overcome the defect that the traditional method has no directional evolution, thereby greatly improving the calculation efficiency of the accurate transfer trajectory of the ground fire.
In order to solve the technical problems, the invention is realized by the following technical scheme:
an initial value rapid acquisition method for ground fire transfer orbit design based on an evolutionary algorithm comprises the following steps:
the method comprises the following steps: according to the GTOP database, adopting the characteristic value of the model initial value sensitive matrix as a sample label to establish an earth-fire transfer orbit learning sample library;
step two: constructing a track optimization support vector machine TO _ SVM, and learning a ground fire transfer track learning sample library;
step three: and (3) aiming at the ground fire transfer orbit TO be designed, operating the orbit optimization support vector machine TO _ SVM in the step two, and giving initial orbit design values of the dominant direction for subsequent use, wherein the initial orbit design values comprise a launch date and an arrival date.
Further, the GTOP database refers to an international trajectory optimization competition.
Further, according to the date of transmission DdDate of arrival AdEmission energy Cd 3And to braking energy Ca 3Constructing sample points X as sample featuresiI.e., the following formula:
Figure BDA0002395637080000021
wherein l is the number of samples.
Further, using sample point XiAnd the characteristic value A of the model initial value sensitivity matrixiThe ground fire transfer orbit learning sample library is constructed as follows:
Figure BDA0002395637080000022
further, the established learning sample library is utilized
Figure BDA0002395637080000023
Solving for the optimal solution (omega) of the quadratic convex optimization process*,b*) And obtaining a track optimization classification decision function:
f(x)＝sgn((ω*)Tφ(x)+b*)
the track optimization classification decision function is a trained track optimization support vector machine TO _ SVM;
wherein sgn is a sign function, and f (x) has a value range of [0, 1 ].
Further, for the ground fire transfer orbit TO be designed in the third step, the orbit optimization support vector machine TO _ SVM in the second step is operated TO provide an initial orbit design value in the dominant direction, which specifically comprises the following steps:
(1) ground fire transfer track designed for needsGenerating a set of sample points
Figure BDA0002395637080000024
(2) Using the trained support vector machine TO _ SVM, for
Figure BDA0002395637080000031
Performing label classification to obtain 1 sample point set
Figure BDA0002395637080000032
I.e. the initial value of the track design for the dominant direction.
The ground fire transfer track to be designed has a transmitting date interval [ Cd1,Cd2]Arrival date interval [ Ca1,Ca2]。
Further, the set of sample points
Figure BDA0002395637080000033
In particular, wherein
Xj＝[Dd(j),Da(j)]。
Further, the invention also provides a system for acquiring the initial value of the ground fire transfer track design, which comprises:
a sample library establishing module: according to the GTOP database, adopting the characteristic value of the model initial value sensitive matrix as a sample label to establish an earth-fire transfer orbit learning sample library;
the support vector machine establishing module comprises: constructing a track optimization support vector machine TO _ SVM, and learning a ground fire transfer track learning sample library;
the track design initial value determining module: and (3) operating an orbit optimization support vector machine TO _ SVM aiming at the ground fire transfer orbit TO be designed, and giving initial orbit design values of the dominant direction for subsequent use, wherein the initial orbit design values comprise a launch date and an arrival date.
Compared with the prior art, the method adopted by the invention has the advantages and beneficial effects that:
(1) before the track is accurately transferred, the track initial value of the dominant direction is rapidly calculated based on the evolutionary algorithm by directly utilizing the track database established previously, and the follow-up more efficient search can be supported.
(2) The method can overcome the defect that the traditional method has no directional evolution, thereby greatly improving the calculation efficiency of the accurate transfer trajectory of the ground fire.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm can rapidly calculate the initial value of the orbit in the dominant direction before the indirect method and the evolutionary algorithm are adopted to accurately transfer the orbit, so that more efficient search can be performed subsequently.
As shown in fig. 1, the steps are as follows:
step one: according to the GTOP database, adopting the characteristic value of the model initial value sensitive matrix as a sample label to establish an earth-fire transfer orbit learning sample library;
according to the disclosed data of each track of international track optimization competition (GTOP), according to the transmission date DdDate of arrival AdEmission energy Cd 3And to braking energy Ca 3These 4 parameters are used as sample features to construct a sample point XiI.e., the following formula:
Figure BDA0002395637080000041
wherein l is the number of samples.
Then, calculating the characteristic value A of the model initial value sensitivity matrix of each sample pointi。
Using sample point XiAnd the characteristic value A of the model initial value sensitivity matrixiThe ground fire transfer orbit learning sample library is constructed as follows:
Figure BDA0002395637080000042
step two:constructing a track optimization support vector machine (TO _ SVM), and learning a ground fire transfer track learning sample library;
a Support Vector Machine (SVM) method which is one of evolution algorithms is a Machine learning method aiming at the problem of small sample classification, and is provided according to the principle of structure risk minimization in a statistical learning theory. The SVM is essentially a convex quadratic optimization problem optimization method.
The trajectory optimization problem is essentially a non-linear optimization problem. Therefore, the secondary convex optimization process is as follows:
Figure BDA0002395637080000051
s.t.yi(ωTφ(xi)+b)≥1-ξi，i＝1，2，...，l，
ξ＝(ξ1，ξ2，...，ξl)T≥0l，
Figure BDA0002395637080000052
wherein ξ ═ (ξ)1，ξ2，...，ξl) For non-negative relaxation variables, each component corresponds to the degree of misdivision of a sample point, C is a penalty parameter, phi (x)i) The function is transferred for the sample point. This requires the construction of a classification hyperplane in the high-dimensional feature space for such an optimization problem to arrive at an orbit optimization classification decision function. Utilizing the learning sample library established in the step one
Figure BDA0002395637080000053
To find the optimal solution (omega) of the above equation*，b*) An orbit optimization classification decision function is obtained:
f(x)＝sgn((ω*)Tφ(x)+b*)
the above formula is the orbit optimization support vector machine (TO _ SVM) obtained after training, where sgn is a sign function, and the value range of f (x) is [0, 1 ]. The method can be used for optimizing the initial value of the track without the sample label in the next step.
Step three:and (4) operating the SVM in the step two aiming at the ground fire transfer orbit (the launching date interval and the arrival date interval) needing to be designed, and selecting the orbit design initial value (the launching date and the arrival date) in the dominant direction for subsequent accurate design.
The specific process is as follows:
1) aiming at ground fire transfer track (emission date interval C)d1，Cd2]Arrival date interval [ Ca1，Ca2]) Generating a set of sample points
Figure BDA0002395637080000054
Wherein
Xj＝[Dd(j)，Da(j)]
2) Using the SVM model trained in step 2, to
Figure BDA0002395637080000055
Performing label classification into a sample point set of 1 (i.e. f (x) ═ 1)
Figure BDA0002395637080000061
I.e. the initial value (launch date, arrival date) for the trajectory design for the dominant direction.
The invention can rapidly provide a plurality of initial values (emission date and arrival date) of the track design in the dominant direction for subsequent accurate design without traversing. The method can overcome the defect that the traditional method has no directional evolution, thereby greatly improving the calculation efficiency of the accurate transfer trajectory of the ground fire.

Claims (10)

1. An evolution algorithm-based method for quickly acquiring an initial value of a ground fire transfer orbit design is characterized by comprising the following steps:
the method comprises the following steps: according to the GTOP database, adopting the characteristic value of the model initial value sensitive matrix as a sample label to establish an earth-fire transfer orbit learning sample library;
step two: constructing a track optimization support vector machine TO _ SVM, and learning a ground fire transfer track learning sample library;
step three: and (3) aiming at the ground fire transfer orbit TO be designed, operating the orbit optimization support vector machine TO _ SVM in the step two, and giving initial orbit design values of the dominant direction for subsequent use, wherein the initial orbit design values comprise a launch date and an arrival date.
2. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 1, is characterized in that: the GTOP database refers to an international track optimization competition.
3. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 1, is characterized in that: according to the date of transmission DdDate of arrival AdEmission energy Cd 3And to braking energy Ca 3Constructing sample points X as sample featuresiI.e., the following formula:
Figure FDA0002395637070000011
wherein l is the number of samples.
4. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 3, is characterized in that: using sample point XiAnd the characteristic value A of the model initial value sensitivity matrixiThe ground fire transfer orbit learning sample library is constructed as follows:
Figure FDA0002395637070000012
5. according to claimThe method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm is characterized by comprising the following steps of: using established learning sample libraries
Figure FDA0002395637070000013
Solving for the optimal solution (omega) of the quadratic convex optimization process*,b*) And obtaining a track optimization classification decision function:
f(x)＝sgn((ω*)Tφ(x)+b*)
the track optimization classification decision function is a trained track optimization support vector machine TO _ SVM;
wherein sgn is a sign function, and f (x) has a value range of [0, 1 ].
6. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 3, is characterized in that: and step three, operating the track optimization support vector machine TO _ SVM in the step two aiming at the ground fire transfer track TO be designed, and giving an initial value of track design in the dominant direction, specifically:
(1) generating a sample point set aiming at the ground fire transfer orbit needing to be designed
Figure FDA0002395637070000021
(2) Using the trained support vector machine TO _ SVM, for
Figure FDA0002395637070000022
Performing label classification to obtain 1 sample point set
Figure FDA0002395637070000023
I.e. the initial value of the track design for the dominant direction.
7. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 6, wherein: the ground fire transfer track to be designed is used for launching the dayPeriod interval [ Cd1,Cd2]Arrival date interval [ Ca1,Ca2]。
8. The method for rapidly acquiring the initial value of the ground fire transfer orbit design based on the evolutionary algorithm as claimed in claim 6, wherein: the set of sample points
Figure FDA0002395637070000024
In particular, wherein
Xj＝[Dd(j),Da(j)]。
9. A fire transfer orbit design initial value acquisition system realized by the fire transfer orbit design initial value quick acquisition method according to claim 1, which is characterized by comprising the following steps:
a sample library establishing module: according to the GTOP database, adopting the characteristic value of the model initial value sensitive matrix as a sample label to establish an earth-fire transfer orbit learning sample library;
the support vector machine establishing module comprises: constructing a track optimization support vector machine TO _ SVM, and learning a ground fire transfer track learning sample library;
the track design initial value determining module: and (3) operating an orbit optimization support vector machine TO _ SVM aiming at the ground fire transfer orbit TO be designed, and giving initial orbit design values of the dominant direction for subsequent use, wherein the initial orbit design values comprise a launch date and an arrival date.
10. The fire transfer track design initial value acquisition system according to claim 9, wherein:
using sample point XiAnd the characteristic value A of the model initial value sensitivity matrixiThe ground fire transfer orbit learning sample library is constructed as follows:
Figure FDA0002395637070000031
wherein, according to the emission date DdDate of arrival AdEmission energy Cd 3And to braking energy Ca 3Constructing sample points X as sample featuresiI.e., the following formula:
Figure FDA0002395637070000032
wherein l is the number of samples.
CN202010130385.9A 2020-02-28 2020-02-28 Ground fire transfer orbit design initial value acquisition method and system based on evolutionary algorithm Active CN111382876B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010130385.9A CN111382876B (en) 2020-02-28 2020-02-28 Ground fire transfer orbit design initial value acquisition method and system based on evolutionary algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010130385.9A CN111382876B (en) 2020-02-28 2020-02-28 Ground fire transfer orbit design initial value acquisition method and system based on evolutionary algorithm

Publications (2)

Publication Number Publication Date
CN111382876A true CN111382876A (en) 2020-07-07
CN111382876B CN111382876B (en) 2023-09-29

Family

ID=71217085

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010130385.9A Active CN111382876B (en) 2020-02-28 2020-02-28 Ground fire transfer orbit design initial value acquisition method and system based on evolutionary algorithm

Country Status (1)

Country Link
CN (1) CN111382876B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114137993A (en) * 2021-11-01 2022-03-04 北京空间飞行器总体设计部 Method for searching ground fire short transfer orbit launching window with deep space maneuver

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103112600A (en) * 2013-03-04 2013-05-22 北京理工大学 Interplanetary transfer orbit design method
CN105631095A (en) * 2015-12-18 2016-06-01 中国人民解放军国防科学技术大学 Search method for multi-constrained earth-moon transfer orbit cluster with equal launch intervals
CN107391813A (en) * 2017-07-03 2017-11-24 北京航空航天大学 A kind of energetic optimum moon based on the moon high ladder ground transfer orbit design method
CN107844462A (en) * 2017-10-26 2018-03-27 北京理工大学 A kind of interplanetary continuous thrust transfer orbit appraisal procedure

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103112600A (en) * 2013-03-04 2013-05-22 北京理工大学 Interplanetary transfer orbit design method
CN105631095A (en) * 2015-12-18 2016-06-01 中国人民解放军国防科学技术大学 Search method for multi-constrained earth-moon transfer orbit cluster with equal launch intervals
CN107391813A (en) * 2017-07-03 2017-11-24 北京航空航天大学 A kind of energetic optimum moon based on the moon high ladder ground transfer orbit design method
CN107844462A (en) * 2017-10-26 2018-03-27 北京理工大学 A kind of interplanetary continuous thrust transfer orbit appraisal procedure

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
尚海滨等: "基于形状的行星际小推力转移轨道初始设计方法" *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114137993A (en) * 2021-11-01 2022-03-04 北京空间飞行器总体设计部 Method for searching ground fire short transfer orbit launching window with deep space maneuver
CN114137993B (en) * 2021-11-01 2022-07-29 北京空间飞行器总体设计部 Ground fire short transfer orbit launching window searching method with deep space maneuver

Also Published As

Publication number Publication date
CN111382876B (en) 2023-09-29

Similar Documents

Publication Publication Date Title
CN111897353B (en) Aircraft maneuvering trajectory prediction method based on GRU
CN106156401B (en) Multi-combination classifier based data driving system state model online identification method
CN109711714A (en) Product quality prediction technique is assembled in manufacture based on shot and long term memory network in parallel
CN108596327B (en) Seismic velocity spectrum artificial intelligence picking method based on deep learning
CN108090621B (en) Short-term wind speed prediction method and system based on staged overall optimization
Lizia et al. Application of high order expansions of two-point boundary value problems to astrodynamics
Liu et al. A fault diagnosis intelligent algorithm based on improved BP neural network
CN113255432B (en) Turbine vibration fault diagnosis method based on deep neural network and manifold alignment
CN111079856B (en) Multi-period intermittent process soft measurement modeling method based on CSJITL-RVM
Alumäe et al. Improved Multilingual Training of Stacked Neural Network Acoustic Models for Low Resource Languages.
CN111191368B (en) Continuous low-thrust interplanetary transfer orbit optimization method and device
CN110609229A (en) Wind driven generator blade imbalance fault detection method based on deep learning
CN114219181A (en) Wind power probability prediction method based on transfer learning
CN111382876A (en) Method and system for acquiring initial value of ground fire transfer orbit design based on evolutionary algorithm
CN108399272B (en) Neural network decision-making method for artificial intelligence programmer to write digital aircraft code
CN118378659B (en) Oxygen leakage risk prediction optimization method for oxygen therapy hole joint
CN112763215A (en) Multi-working-condition online fault diagnosis method based on modular federal deep learning
Wang et al. Deep learning based missile trajectory prediction
CN109325288A (en) A kind of Solid Launch Vehicle population parameter based on uncertainty optimization determines method and system
CN114329315B (en) Static aeroelastic rapid solving method based on dynamic modal decomposition technology
CN113673704B (en) Relational network reasoning optimization method based on software and hardware cooperative acceleration
Viavattene et al. Artificial neural network design for tours of multiple asteroids
Zhu et al. Fast accessibility evaluation of the main-belt asteroids manned exploration mission based on a learning method
Xie et al. Feasible Low-thrust Trajectory Identification via a Deep Neural Network Classifier
CN115470883A (en) Neural network output post-processing method of multi-model mechanism

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