CN106871908B

CN106871908B - Optimization selection strategy for satellite natural landmark library

Info

Publication number: CN106871908B
Application number: CN201611216394.XA
Authority: CN
Inventors: 吴云华; 葛林林; 陈志明; 华冰; 张泽中; 江春
Original assignee: Nanjing University of Aeronautics and Astronautics
Current assignee: Nanjing University of Aeronautics and Astronautics
Priority date: 2016-12-26
Filing date: 2016-12-26
Publication date: 2020-04-07
Anticipated expiration: 2036-12-26
Also published as: CN106871908A

Abstract

The invention discloses an optimization selection strategy of an on-board natural landmark library, and relates to the technical field of autonomous navigation of spacecraft. The scheme uses Analytic Hierarchy Process (AHP) to divide the model into upper and lower layers. The upper layer optimization model is combined with satellite overhead prediction technology to analyze the landmark coverage; the lower layer optimization model uses N _d invariant algorithm (DPNIA) to optimize the landmark library. The invention solves the problems of low matching efficiency of natural landmark database and limited storage space on the satellite when using images for autonomous navigation in the prior art, thereby greatly improving the real-time performance of autonomous navigation based on natural landmarks, and reducing the cost of satellite navigation. The storage space requirements of the landmark database lay the foundation for the application of autonomous navigation technology based on natural landmarks.

Description

Optimization selection strategy for satellite natural landmark library

The technical field is as follows:

the invention relates to an optimization selection strategy of a satellite natural landmark library, which is a solution for performing optimization selection on the natural landmark library by combining spacecraft over-the-top prediction and a DPNIA algorithm, and belongs to the technical field of autonomous navigation.

Background art:

with the increase in the number of on-orbit spacecraft, spacecraft navigation technology has transitioned from traditional radar station-based navigation to an era dominated by autonomous navigation. The spacecraft autonomous navigation technology is a technology for supporting autonomous operation of spaceflight by acquiring necessary observation information in the modes of autonomous measurement of a sensor on a satellite and the like under the condition that a spacecraft cannot be supported by a ground measurement and control system for a long time, and then processing the observation information by a satellite-borne processor to continuously determine and forecast parameters of a spacecraft orbit and the like.

Compared with the traditional navigation mode, the autonomous navigation has the advantages of strong concealment, high autonomy and the like, has important significance in military affairs, and becomes the main trend of the development of navigation technologies of various countries.

Due to the rapid development of technology, various autonomous navigation schemes have emerged. The method mainly comprises the following steps: astronomical navigation, global navigation satellite system, pulsar autonomous navigation, joint orbit determination navigation based on space relative measurement, quantum positioning navigation, autonomous navigation based on geomagnetic field, autonomous navigation based on natural polarized light and the like. These autonomous navigation schemes have advantages and disadvantages, for example, astronomical navigation based on "starlight direction + earth center vector" has advantages of low cost, mature technology and good reliability, but it is difficult to accurately determine the position of the earth edge in fact because the earth edge appears blurred due to factors such as atmospheric coverage, brightness of the earth edge gradually becoming dark with altitude, and the like. Therefore, the accuracy of the geocentric vector obtained by direct measurement by the earth sensor is low, so that the accuracy of the autonomous navigation method is limited. For another example, pulsar autonomous navigation has the advantage of wide application airspace, but the navigation accuracy needs to be further improved due to the influence of factors such as a star catalogue error, a time measurement error, a pulsar number and an initial position error. Different autonomous navigation methods have their own application places under different requirements at different periods.

Natural landmarks are terrain features with sharp contour features, including shorelines, islands, rivers, lakes, etc. The natural landmark has wide application prospect due to the advantages of no restriction of national boundaries, all-weather observation, easy identification and the like.

In view of the above advantages of natural landmarks, autonomous navigation of spacecraft based on natural landmarks is expected to become a next-generation new autonomous navigation solution. The autonomous navigation based on the natural landmarks is to use the landmarks as reference objects, use imaging equipment on satellites to obtain ground image information, use an image matching technology to match the obtained images with a natural landmark library established in advance, and use the position information of the landmarks in the landmark library to solve the orbit of the spacecraft when the matching is successful, thereby realizing the autonomous navigation of the spacecraft. The number of natural landmarks with salient features is huge, and all the natural landmarks cannot be stored on the satellite due to the limitation of the storage space on the satellite. Therefore, the optimization selection of the satellite natural landmark library has important significance.

The invention content is as follows:

the invention aims to provide an optimization selection strategy of a satellite natural landmark library, and aims to solve the problems that in the prior art, the matching efficiency is low and the satellite storage space is limited when a spacecraft utilizes the natural landmark library to conduct autonomous navigation.

The invention adopts the following technical scheme: an optimization selection strategy for a satellite natural landmark library comprises the following processes

Dividing the optimization model into an upper layer and a lower layer by using an analytic hierarchy process;

an upper optimization model:

(1) collecting natural landmarks with easy identification characteristics by using Google Earth, and forming an original natural landmark library by the landmarks and corresponding landmark longitude and latitude information;

(2) according to the preset orbit parameters of the spacecraft, the natural landmarks are used as target points, and an over-the-top prediction theory is utilized to obtain a coverage analysis result, wherein the result comprises the number of times, the starting time, the ending time and the duration of covering each natural landmark under the orbit;

(3) using an MATLAB writing program to judge whether the landmark in the original landmark library is in the coverage analysis result, if the landmark is not in the coverage analysis result, the landmark is redundant, otherwise, the landmark is effective, and the original data used as a next layer model must be reserved for further optimization;

the lower optimization model is as follows:

(1) establishing a natural landmark library obtained from an upper model

A type order table;

(2) for the ordered list established in step 1, firstly calculating the number N of bad points_dAnd entering the original data intoPerforming line backup;

(3) attempt to delete each record and test for N compliance at the same time_dInvariance rules, which indicate that a record is meaningful if it is not compliant, restore the data to the state before deletion of the record, and conversely, if N is compliant after deletion of the record_dThe invariance rule indicates that the record has no effect and the record is deleted;

(4) if all records have been tested, the algorithm ends, based on what is finally left

Establishing a final optimized natural landmark library by all natural landmarks in the type ordered table;

wherein: the landmark names are denoted by the symbol P, and P (i) denotes T_s(i) An identifier of the corresponding landmark; λ represents longitude information of a landmark, λ_iRepresenting the longitude information corresponding to the natural landmark P (i);

it is represented latitude information of the landmark,

representing latitude information corresponding to the natural landmark P (i); starting time T_sRepresenting the starting time, T, at which a landmark is observed by the spacecraft_s(i) The meaning of (a) is the ith "start time" in order of "start time" from small to large; the time interval Δ T is defined in the coverage result as the starting time T_sIn order of small to large, Δ T (i) means T_s(i-1) and T_s(i) The interval therebetween is expressed by the formula Δ T (i) ═ T_s(i)-T_s(i-1); dead pixel P_dWhen the time is delta t (i) is more than 0.5h, the corresponding P (i) at the moment is a dead point; number of bad points N_dIs a dead point P_dThe number of (2); the element e is a group containing at least { P, T_s-an element of }; the set S is a set related to the element e, namely e belongs to S;

the type ordered list is based on the set S according to T_sA table is obtained by sorting from small to large.

The invention has the following beneficial effects: the optimization selection strategy of the satellite natural landmark library utilizes an Analytic Hierarchy Process (AHP) to divide a model into an upper layer and a lower layer. The upper layer optimization model is used for analyzing the landmark coverage by combining a satellite over-the-top prediction technology; lower optimization model utilization N_dThe invariant algorithm (DPNIA) enables optimization of the landmark library. The invention solves the problems of low matching efficiency of the natural landmark library and limited on-satellite storage space when the image is used for autonomous navigation in the prior art, thereby greatly improving the real-time performance of the autonomous navigation based on the natural landmarks, reducing the storage space requirement of the on-satellite landmark library and laying a foundation for the application of the autonomous navigation technology based on the natural landmarks.

Description of the drawings:

fig. 1 is a distribution plot of natural landmarks prior to the present invention being unoptimized.

FIG. 2 is an overall flow chart of the present invention.

FIG. 3 is a schematic diagram illustrating the theory of over-top prediction according to the present invention.

Fig. 4 is a distribution diagram of the natural landmarks obtained after the upper layer optimization model.

FIG. 5 is a distribution diagram of natural landmarks obtained by the present invention using the underlying optimization model.

Fig. 6 is a comparison graph of the observed time interval Δ t of the natural landmark of the present invention before and after optimization.

The specific implementation mode is as follows:

the on-board natural object library optimization selection strategy described in the present invention is further described in detail with reference to the accompanying drawings and specific embodiments. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.

As shown in fig. 1, the distribution diagram of the natural landmarks before optimization, the latitude and longitude of these natural landmarks are from 180 ° E to 180 ° W, 50 ° N to 50 ° S. The original natural landmarks are obtained by collecting a large number of natural objects with the attributes of remarkable characteristics, easy observation, difficult transition and the like. The original natural landmark library is formed by combining the landmark identifiers, the landmark image characteristics, the landmark longitude and latitude and other information of the natural landmarks.

As shown in fig. 2, an overall flowchart of the selection strategy for the satellite natural landmark library optimization is shown, and the flowchart is specifically described as follows:

dividing an optimization model into an upper layer and a lower layer by using an Analytic Hierarchy Process (AHP);

upper optimization model

Theoretical basis:

the upper layer optimization model is established on the basis of an over-top prediction theory, and the semimajor axis of the satellite orbit is assumed to be a, the eccentricity is assumed to be e, the orbit inclination angle is assumed to be i, the average angular velocity is assumed to be n, and the earth equator radius is assumed to be R_eThe gravity parameter is mu, the second order band harmonic coefficient is J₂The focal period T of the track_ΩComprises the following steps:

wherein intermediate variables η and c are assumed to avoid formula overlength, an

Fig. 3 is a schematic diagram of the nth circle passing through the top of the satellite after the satellite passes through the target point latitude from the first orbit reduction. Point A is the intersection point of the satellite orbit and the earth equator, point C is the intersection point of the orbit of the satellite in the Nth circle and the latitude of the target point, point D is the middle point of the target point in the observable time period, and point P is the intersection point of the target point in the observable time period_sIs the starting point of the satellite observing the target point, point P_eIs the end point at which the satellite observes the target point.

Is the latitude value of the target point, α is the arc value of arc CA, α 'is the arc value of arc AB, Δ λ is the longitude difference between point C and the target point, Δ λ' is the arc value of the arc formed by point C and the target point with respect to the center of the earth.

In addition, from the geometric relationships in the figure, it can be deduced:

due to the rotation of the earth, when the satellite passes the top, the target point moves relative to the sight, so that the error of the over-top time forecast is increased. In order to eliminate as much as possible the relative apparent movement of the target point when the satellite passes over the top due to the rotation of the earth, the angle theta must be adjusted_cThe angle is appropriately corrected. Suppose the rotational angular velocity of the earth is ω_eThen the corrected value can be obtained as:

wherein V is:

as shown in FIG. 3, a boundary communication area arc value theta formed by the projection of a target point on the celestial sphere when a satellite enters the communication area boundary and the instantaneous position of the satellite relative to the earth center is assumed₀Comprises the following steps:

wherein the target point and the orbital plane form an included angle theta relative to the center of the earth_TDComprises the following steps:

θ_TD＝sin^-1(sinΔλ’sinθ’_c) (7)

then can obtain

θ_CD＝cos^-1(cosΔλ’/cosθ_TD) (8)

Suppose that the moment when the first orbit of the satellite falls and passes through the target point latitude is T₀Finally, when the target point and the predetermined orbit of the satellite are determined, the following conclusion is drawn according to the over-the-top prediction theory:

the time when the observable region is entered from the left/right side is:

the time when the observable region is left/right is:

on the basis of the over-top prediction theory, the natural landmark coverage result under the preset orbit can be obtained. And detecting a natural landmark library, reserving natural landmarks in the coverage result, and rejecting natural landmarks which are not in the coverage result. The specific process is as follows:

(1) the method comprises the steps of collecting natural landmarks with characteristics such as easy identification and the like, such as craters, coastlines, islands and the like by using Google Earth, and forming an original natural landmark library by the landmarks and corresponding landmark longitude and latitude information.

(2) Programming an over-the-top prediction theory by using MATLAB (matrix laboratory) by taking the natural landmarks as target points according to preset orbit parameters of the spacecraft to obtain a coverage analysis result, wherein the result comprises the information of the number of times, the starting time, the ending time and the duration of covering each natural landmark under the orbit;

(3) and (4) utilizing an MATLAB writing program to judge whether the landmarks in the original landmark library are in the coverage analysis result. If not in the coverage analysis results, the landmark is declared redundant. Otherwise, the landmark is valid, and the original data used as the next layer model must be preserved for further optimization.

And screening the natural landmarks before optimization by using the upper-layer optimization model to obtain a distribution map of the natural landmarks optimized by the upper-layer optimization model, as shown in fig. 4.

Lower optimization model

In order to mine the rules hidden in the original data and to facilitate data processing, the present patent defines some variables as follows.

Name of landmark P

The "landmark name" is an identifier for distinguishing different landmarks, and is denoted by a symbol P. P (i) denotes T_s(i) An identifier of the corresponding landmark.

Longitude λ of landmark

The symbol λ represents longitude information of the landmark, λ_iThe representative is longitude information corresponding to the natural landmark p (i).

Latitude of landmark

Symbol

It is represented latitude information of the landmark,

the latitude information corresponding to the natural landmark P (i) is shown in the table.

Starting time T_s

"start time" represents the starting time of a landmark as observed by the spacecraft, using the symbol T_sAnd (4) showing. T is_s(i) Meaning the ith "start time" in order of small to large "start times".

Time interval Δ t

"time interval" is defined as the starting time T in the coverage result_sOrdered from small to large, denoted by the symbol Δ t. Δ T (i) means T_s(i-1) and T_s(i) The interval therebetween is expressed by the formula Δ T (i) ═ T_s(i)-T_s(i-1)。

Dead pixel P_d

The term "dead pixel" is a concept proposed by the present invention, and means that when Δ t (i) > 0.5h, the corresponding p (i) at that time is dead pixel.

Number of bad points N_d

The meaning of the number of bad points is that the bad point P_dThe number of (2).

Element e

e is a group consisting of at least { P, T_sThe elements of (c). It may have other properties but must contain at least P, T_sThese two attributes.

Set S

S is a set of related elements e, i.e., e ∈ S.

Type order table

The specific flow of the lower optimization model is as follows:

(1) establishing a natural landmark library obtained from an upper model

Type ordered list.

(2) For the ordered list established in step 1, firstly calculating the number N of bad points_dAnd the original data is backed up.

(3) Attempt to delete each record and test for N compliance at the same time_dAnd (4) invariance rules. If the rule is not met, meaning of the record is indicated, and the data needs to be restored to the state before the record is deleted. Conversely, if N is met after record deletion_dInvariance rules indicate that the record has no effect and the record can be deleted.

(4) If all records have been tested, the algorithm ends. According to what is finally left

And establishing a final optimized natural landmark library by all the natural landmarks in the type ordered table.

On the basis of the primary optimization result of the upper-layer optimization model, the result is further optimized by using the lower-layer optimization model, and a distribution map of the natural landmarks optimized by the lower-layer optimization model is obtained, as shown in fig. 5.

Fig. 6 shows a comparison of the observed time interval Δ t of the natural landmarks before and after the optimization, and it can be seen that Δ t more than half an hour after the model optimization is not increased or decreased, and a large number of redundant landmarks are deleted. Through practical tests, the satellite landmark library can be compressed to about 11% of the original satellite landmark library.

The upper-layer optimization model in the optimization selection strategy of the satellite natural landmark library performs over-the-top prediction analysis on natural landmarks in all the landmark libraries aiming at the designed satellite orbit, so that the coverage characteristics of the natural landmarks are obtained. Natural landmarks not covered under the track are considered redundant and need to be culled from the landmark library. And the remaining natural landmarks are regarded as valuable natural landmarks and are used as the raw data of the lower model for further analysis and optimization; the lower optimization model is essentially built on N_dThe invention discloses an optimization algorithm based on invariance rules, which is named as DPNIA (dead pointNumber Invariant Algorithm) algorithm. Therefore, the algorithm is used for further optimizing and selecting the natural landmark library on the basis of the upper model.

The foregoing is only a preferred embodiment of this invention and it should be noted that modifications can be made by those skilled in the art without departing from the principle of the invention and these modifications should also be considered as the protection scope of the invention.

Claims

1. a kind of on-star natural landmark library optimization selection strategy, it is characterized in that: process is as follows

The optimization model is divided into upper and lower layers by using the analytic hierarchy process;

Upper layer optimization model:

(1) Use Google Earth to collect natural landmarks with easily identifiable characteristics, and form the original natural landmark database with these landmarks and the corresponding latitude and longitude information of the landmarks;

(2) According to the predetermined orbit parameters of the spacecraft, the natural landmark is used as the target point and the over-the-top prediction theory is used to obtain the coverage analysis result. time and duration;

(3) Use MATLAB to write a program to judge whether the landmark in the original landmark library is in the coverage analysis result. If it is not in the coverage analysis result, it means that the landmark is redundant; otherwise, it means that the landmark is valid and must be kept. It is used as the raw data of the next layer model for further optimization;

The lower-level optimization model:

(1) On the basis of the natural landmark database obtained by the upper model, establish

Type ordered list;

(2) For the ordered table established in step 1, first calculate its number of bad points N _d , and back up the original data;

(3) Try to delete each record and test whether it conforms to the N _d invariance rule at the same time. If it does not conform to the rule, it means that the record is meaningful. At this time, the data is restored to the state before the record was deleted. On the contrary , if the record conforms to the N _d invariance rule after deletion, it means that the record has no effect and the record is deleted;

(4) If all records have been tested, the algorithm ends, according to the final remaining

All natural landmarks in the ordered list of the type are established, and the final optimized natural landmark library is established;

Among them: the landmark name is represented by the symbol P, and P(i) represents the identifier of the landmark corresponding to T _s (i); λ represents the longitude information of the landmark, and λ _i represents the longitude corresponding to the natural landmark P(i). information;

represents the latitude information of the landmark,

Represents the latitude information corresponding to the natural landmark P(i); the start time T _s represents the start time when a landmark is observed by the spacecraft, and the meaning of T _s (i) is sorted according to the "start time" from small to large The i-th "start time"; the time interval Δt is defined as the starting time T _s in the coverage result sorted from small to large, Δt(i) means the difference between T _s (i-1) and T _s (i) The interval between Δt(i)=T _s (i)-T _s (i-1) is expressed by the formula; for the bad point P _d , when Δt(i)>0.5h, the corresponding P(i) at this moment Then it is a bad point; the number of bad points N _d is the number of bad points P _d ; the element e is an element containing at least {P, T _s }; the set S is the set about the element e, that is, e∈S;

The type ordered list is a list obtained by sorting T _s from small to large on the basis of the set S.