CN105787173A - Multi-satellite earth-observation task scheduling and planning method and device - Google Patents

Abstract

The invention provides a multi-satellite earth-observation task scheduling and planning method and device.The method comprises the steps that an observation task and satellite resource data are read; the observation task is partitioned according to the satellite resource data, and an observation meta-task set is obtained; task planning is conducted on the observation meta-task set through an evolutionary algorithm, and a planning result is obtained; the planning result is decoded, and an observation sequence is obtained; in this way, when multiple satellites observe the earth, the characteristics such as intelligence and concurrency of the evolutionary algorithm are fully utilized, and the task scheduling problem of multiple observation satellites which need to work cooperatively is solved; a dynamic task planning method based on the evolutionary algorithm is utilized, a reasonable load scheduling scheme is made for an in-orbit operation satellite group, resource allocation of a satellite system is optimized, and satellite system resources are utilized fully and reasonably.

Description

A kind of scheduling planning method of many stars earth observation task and device

Technical field

The invention belongs to satellite task planning technology field, particularly relate to scheduling planning method and the device of a kind of many stars earth observation task.

Background technology

Earth observation is with the earth for object of study, relies on the space platforms such as satellite, airship space shuttle and Near Space Flying Vehicles, utilizes the multiple detection means such as visible ray infrared high spectrum and microwave, and acquisition information also carries out processing and forming product.

At the beginning of earth observation satellite develops, owing to satellite load is limited in one's ability, user task is also relatively fewer, and observation time and the observation angle of task are all relatively fixed, and Satellite Management and control are fairly simple, and mission planning problem does not also highlight.But along with the increase of the development of earth observation satellite technology and ground image data demand, satellite needs the side-looking angle adjusting remote sensing equipment to be observed on a surface target.In the process arranged, many constraints need to be considered, to ensure that safety satellite runs reliably and is observed task smoothly.Generally speaking, it is impossible in a mission planning time range, carry out all of task requests carry out imaging, the task that satellite performs every time is a subset of task data set.The demand that user proposes can not be met.

For making full use of satellite resource, it is necessary to the mission requirements that user is proposed, and earth observation satellite carries out unified management and scheduling planning.Prior art can not be met satellite daily management and the demand of commander's control by simple reasoning and calculation, it is necessary to could better manage by suitable mathematical model and software tool and distribute satellite resource, meet growing mission requirements to maximize.But at present, great majority concentrate on single satellite scheduling of resource and time series planning aspect about the scheduling research of satellite task.And these researchs all lack the complete analysis from user's request, model, algorithm and final software system, and great majority research is all closely related with concrete satellite system and task, it is impossible to meet very well and be actually needed.

Based on this, how realizing the satellite resource scheduling in many stars multitask situation, optimize the resource distribution of satellite system, abundant Appropriate application satellite system resource, is the problem in the urgent need to address currently faced.

Summary of the invention

For prior art Problems existing, embodiments provide scheduling planning method and the device of a kind of many stars earth observation task, for solving in prior art when the earth observation of many stars, it is impossible to optimize the resource distribution of satellite system, the technical problem of abundant Appropriate application satellite system resource.

The present invention provides a kind of scheduling planning method of many stars earth observation task, and described method includes:

Read observation mission and satellite resource data；

According to described satellite resource data, observation mission is divided, obtain observation Meta task collection；

Utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results；

Described program results is decoded, obtains observation sequence；Wherein,

When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

In such scheme, described according to described satellite resource data, observation mission is divided, obtains observation Meta task collection and include:

In satellite tool kit (STK, SatelliteToolKit), model of place is set up according to described satellite resource data；

Calculate the SEE time window between satellite load and ground target；

Calculate the sub-satellite track of satellite；

Initial time according to sub-satellite track and described time window determines the described satellite substar position at the initial time place of described time window；

The four angular coordinate of described moonscope cover strip is determined according to described substar position；

Described observation Meta task collection is obtained according to described observation covering tape.

In such scheme, utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results and include:

Side-sway number of times, observation working time, sun altitude and side-sway angle that described observation Meta task is concentrated carry out constraint definition；

Side-sway count constraint, the constraint of observation working time, sun altitude constraint and side-sway angle restriction are carried out it is assumed that set up satellite load scheduling model on the basis assumed；

Utilize described evolution algorithmic that described satellite load scheduling model is solved, obtain program results.

In such scheme, described satellite resource data include: satellite information, payload information and land object information.

In such scheme, described satellite load scheduling model includes: $min\underset{v=1}{\overset{n}{\Σ}}(t_v^{begin}-t_v^{early})/s$ And $\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\};$

Wherein, described inRepresent that maximization completes task；Described p_vBeing the priority of v task, described s is total task number；

DescribedFor minimizing resource consumption；Described m is the total number of remote sensor；

DescribedRepresent and minimize task waiting time；DescribedIt is the start time of the time window that the v task is arranged to, described inEarly start for task v observes the moment, described inWaiting time for task v；

DescribedRepresent that the total time of all tasks is minimum；DescribedRepresent the start time of twv time window of v subtask, described inRepresent the tw of v subtask_vThe finish time of individual time window.

The present invention also provides for the scheduling planning device of a kind of many stars earth observation task, and described device includes:

Read module, is used for reading observation mission and satellite resource data；

Divide module, for observation mission being divided according to described satellite resource data, obtain observation Meta task collection；

Planning module, is used for utilizing evolution algorithmic that described observation Meta task collection carries out mission planning, obtains program results；

Decoder module, for described program results is decoded, obtains observation sequence；Wherein,

When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

In such scheme, described division module specifically for:

In satellite tool kit STK, scene is set up according to described satellite resource data；

Calculate the SEE time window between satellite load and ground target；

Calculate the sub-satellite track of satellite；

Initial time according to sub-satellite track and described time window determines the described satellite substar position at the initial time place of described time window；

The four angular coordinate of described moonscope cover strip is determined according to described substar position；

Described observation Meta task collection is obtained according to described observation cover strip.

In such scheme, described planning module specifically for:

Side-sway number of times, observation working time, sun altitude and side-sway angle that described observation Meta task is concentrated carry out constraint definition；

Side-sway count constraint, the constraint of observation working time, sun altitude constraint and side-sway angle restriction are carried out it is assumed that set up satellite load scheduling model on the basis assumed；

Utilize described evolution algorithmic that described satellite load scheduling model is solved, obtain program results.

In such scheme, described satellite resource data include: satellite information, payload information and land object information.

In such scheme, described satellite load scheduling model includes: $min\underset{v=1}{\overset{n}{\Σ}}(t_v^{begin}-t_v^{early})/s$ And $\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\};$

Wherein, described inRepresent that maximization completes task；Described p_vBeing the priority of v task, described s is total task number；

DescribedFor minimizing resource consumption；Described m is the total number of remote sensor；

DescribedRepresent and minimize task waiting time；DescribedIt is the start time of the time window that the v task is arranged to, described inEarly start for task v observes the moment, described inWaiting time for task v；

DescribedRepresent that the total time of all tasks is minimum；DescribedRepresent the tw of v subtask_vThe start time of individual time window, described inRepresent the tw of v subtask_vThe finish time of individual time window.

The invention provides a kind of scheduling planning method of many stars earth observation task and device, described method includes: read satellite resource data；According to described satellite resource data, observation mission is divided, obtain observation Meta task collection；Utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results；Described program results is decoded, obtains observation sequence；So, when many stars are observed over the ground, the features such as intelligent, the concurrency making full use of evolution algorithmic, solve the Mission Scheduling at many observation satellites needing collaborative work: utilize the dynamic task planing method based on evolution algorithmic, rational load scheduling scheme is formulated for satellite group in orbit, optimize the resource distribution of satellite system, abundant Appropriate application satellite system resource.

Accompanying drawing explanation

The scheduling planning method flow schematic diagram of many stars earth observation task that Fig. 1 provides for the embodiment of the present invention one；

The satellite that Fig. 2 embodiment of the present invention one the provides location drawing in terrestrial coordinate system；

The schematic diagram of the covering tape that Fig. 3 embodiment of the present invention one provides；

The scheduling planning apparatus structure schematic diagram of many stars earth observation task that Fig. 4 provides for the embodiment of the present invention two；

Many stars earth observation task system structural representation that Fig. 5 provides for the embodiment of the present invention three；

The overall structure schematic diagram of the system program modules that Fig. 6 provides for the embodiment of the present invention three；

The STK observation program to obtaining through evolutionary optimization that utilizes that Fig. 7 provides for the embodiment of the present invention three carries out simulation demo schematic diagram.

Detailed description of the invention

The resource distribution of satellite system dispatched, optimize in order to realize the satellite resource in many stars multitask situation, abundant Appropriate application satellite system resource, the invention provides a kind of scheduling planning method of many stars earth observation task and device, described method includes: read satellite resource data；According to described satellite resource data, observation mission is divided, obtain observation Meta task collection；Utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results；Described program results is decoded, obtains observation sequence.

Below by drawings and the specific embodiments, technical scheme is described in further detail.

Embodiment one

The present embodiment provides a kind of scheduling planning method of many stars earth observation task, as it is shown in figure 1, said method comprising the steps of:

Step 110, reads observation mission and satellite resource data.

In this step, reading observation mission and satellite resource data from satellite resource data base, described satellite resource data include: satellite information, payload information and land object information etc..Described observation mission is

Wherein, described satellite information includes: the track service condition of satellite, precision and side-sway ability.Described payload information refers to equipment and the load-carrying ability for obtaining view data that satellite carries, such as: camera (EO-1 hyperion, multispectral or panchromatic), synthetic aperture radar and various sensor (frequency range and wave band) etc..Described ground target refers to the ground region needing to be observed, including: the overlay capacity of point target and regional aim and ground target.

Step 111, divides described observation mission according to described satellite resource data, obtains observation Meta task collection.

In this step, after reading satellite resource data, setting up model of place in STK according to described satellite resource data, after model of place establishes, the computing module that accesses calling described STK calculates the SEE time window between satellite load and the ground target of observation mission；After described SEE time window calculation goes out, calculate the sub-satellite track of satellite, determine the described satellite substar position at the initial time place of described time window according to the initial time of sub-satellite track and described time window；Determine the four angular coordinate of described moonscope cover strip further according to described substar position, obtain described observation Meta task collection finally according to described observation covering tape.

Specifically, sub-satellite track requires over satellite ground tracks and determines, and satellite ground tracks need to be determined by reduced latitude and geocentric longitude.Accurate satellite ground tracks generally to calculate according to the integration of equation of satellite motion, in aerospace engineering is applied, owing to remote sensing satellite orbital eccentricity e is general close to zero, when therefore calculating the ground trace of remote sensing satellite, its track can be used as circular orbit when meeting certain required precision and process.Secondly, when consider non-spherical earth perturbation on satellite orbit when affecting, only J in Geopotential coefficient₂It is 10^-3Magnitude, all the other J_n(n=3,4 ...) and the magnitude of the humorous term coefficient in field be generally less than 10^-6, therefore its impact can be ignored.And J₂Coefficient is mainly due to caused by compression of the Earth f, so considering, at this, the first order perturbation that compression of the Earth factor causes.

Here, if Ω is right ascension of ascending node rate of change, ω_eFor the average spin velocity of the earth, then:

ω_e=7.292115 × 10^-5(rad/s)(1)

$\mathrm{\Ω}=-\frac{3}{2}{J}_2\sqrt{\mathrm{\μ}}\frac{a_e^2\cos i}{a^{7/2}{(1-e^2)}^2}(rad/s)---\left(2\right)$

Wherein, in formula (2), described a is the major radius of satellite orbit, and described e is eccentricity of satellite orbit, and described i is the inclination angle of satellite orbit, described a_eFor terrestrial equator footpath, for 6378137m；Described μ is geocentric gravitational constant, 3.986005 × 10¹⁴m³/s²；J₂For second order gravitational potential coefficient, it is 1.083 × 10^-3.So, satellite ascending node is relative to the western back speed rate ω ' of longitude zero point_eFor being represented by formula (3):

ω′_e=ω_e-Ω(3)

Known satellite mean angular velocity of satellite motion n is represented by

$n=\frac{\sqrt{\mathrm{\μ}}}{a^{3/2}}---\left(4\right)$

In fig. 2, X-axis points to the longitude zero point of International Time Bureau's definition.If satellite is through ascending node Ω₀Moment be t₀, longitude is λ₀.Through Δ t=t t₀After time, ascending node west retreats to Ω, and satellite arrives S, wherein,

ΩΩ₀=ω '_eΔt(5)

$\stackrel{\∩}{\mathrm{\Ω}S}=n\mathrm{\Δ}t---\left(6\right)$

In formula (7),Reduced latitude for t satellite.And satellite is relative to Ω₀Can be represented by formula (8) through difference:

$\mathrm{\Δ}\mathrm{\λ}=\stackrel{\∩}{{\mathrm{\Ω}}_{0}D}---\left(8\right)$

In right angle spherical triangle △ Ω SD, ∠ S Ω D=i, i are known orbit inclination angle, so having:

$sin\stackrel{\∩}{SD}=\sin \stackrel{\∩}{\mathrm{\Ω}S}\sin i---\left(9\right)$

Formula (6) (7) is substituted in formula (9) and draws

So reduced latitude can be solved by formula (10)

Further, geodetic latitudeComputing formula be:

Wherein, in formula (12), f=1/298.257.

Order∠SΩ₀D=ψ；So, at spherical triangle Δ Ω S Ω₀In, according to cosine law for sides and five elements formula it follows that

$\begin{array}{c}\cos L={\mathrm{cos\ω}}_e^{\′}\·\cos n\mathrm{\Δ}t+{\mathrm{sin\ω}}_e^{\′}n\mathrm{\Δ}t\·\sin n\mathrm{\Δ}t\·\cos i-\sin L\·\cos \mathrm{\ψ}\\ ={\mathrm{sin\ω}}_e^{\′}\·\cos n\mathrm{\Δ}t-{\mathrm{cos\ω}}_e^{\′}n\mathrm{\Δ}t\·\sin n\mathrm{\Δ}t\·\cos i\end{array}---\left(13\right)$

At right angle spherical triangle △ Ω₀In SD, have:

Therefore, by formula (13)-(15) it follows that

Then, formula (16), (17) draw:

$\tan \mathrm{\Δ}\mathrm{\λ}=\frac{-{\mathrm{sin\ω}}_e^{\′}\mathrm{\Δ}t\·\cos n\mathrm{\Δ}t+{\mathrm{cos\ω}}_e^{\′}\mathrm{\Δ}t\·\sin n\mathrm{\Δ}t\·\cos i}{{\mathrm{cos\ω}}_e^{\′}\mathrm{\Δ}t\·\cos n\mathrm{\Δ}t+{\mathrm{sin\ω}}_e^{\′}\mathrm{\Δ}t\·\sin n\mathrm{\Δ}t\·\cos i}---\left(18\right)$

So, t geocentric longitude λ is

λ=λ₀+Δλ(19)

Here, geocentric longitude is equal to geodetic longitude, need not change.

But in above-mentioned derivation, the non-linear of the closely dynamic rate of heterogeneity and track owing to have ignored satellite motion speed, therefore to ensure computational accuracy, sub-satellite track is started at extrapolation and be must not exceed 1/4 circle from equator, it is to say, in above-mentioned formula, Δ t should meet:

-T/4 < Δ t < T/4 (20)

In formula (20) formula, T is track mean period, it is possible to calculated by formula (21):

T=2 π/n (21)

Wherein, in formula (21), n is the mean angular velocity of satellite motion of satellite.

Like this, when calculating sub-satellite track, need first to start northwards from ascending node every time, return again to after each extrapolation 1/4 circle to the south to start southwards from southbound node, northwards each extrapolation 1/4 circle, then start northwards from the ascending node of next circle, each extrapolation 1/4 to the south is enclosed, and calculates sub-satellite track so round and round again.

For example, as it is known that longitude of ascending node λ₀, take Δ t=1 when northwards extrapolating along satellite flight direction by ascending node, 2 ..., [T/4]；When extrapolating southwards against satellite flight direction by ascending node, take Δ t=-1 ,-2 ...-[T/4], symbol " [] " represents the integer part taking a number, and its algebraic symbol is identical with former number.The computing formula of substar reduced latitude and geocentric longitude is provided by (11) and (19) formula respectively.

Again for example, as it is known that southbound node longitude λ₀, when extrapolating southwards along satellite flight direction by southbound node, take Δ t=1,2 ..., [T/4], when northwards extrapolate in inverse satellite flight direction by southbound node, take Δ t=-1 ,-2 ... ,-[T/4].At this moment substar reduced latitude formula can change formula (22) into, and the computing formula of geocentric longitude is constant:

So, it is possible to calculate satellite substar the earth warp more per second, latitude.Between southbound node and ascending node or between ascending node and previous circle southbound node through difference Δ λ₀Can be drawn by formula (23):

${\mathrm{\&Delta;\&lambda;}}_0=\mathrm{\&pi;}-\frac{1}{2}{\mathrm{\&omega;}}_e^{\&prime;}T---\left(23\right)$

Here, step-length desirable 10s, 30s or 60s of Δ t, step-length is more little, and sub-satellite track is more smooth.

Further, described observation Meta task collection is obtained particularly as follows: using terrestrial equator as X-axis, using zero degree warp as Y-axis, set up rectangular coordinate system according to described observation covering tape, it is thus achieved that four angular coordinates of observed object.Using the minimum longitude in four coordinate points as the left margin of grid division；Using the maximum longitude in four coordinate points as the right margin of grid division；In four coordinate points, minimum latitude is as the lower boundary of grid division；Using the maximum latitude in four coordinate points as the coboundary of grid division.The corner angular zone obtained with this is divided into 100 and takes advantage of the small grid of 100.Adopt area-method to judge that grid is whether in covering tape, obtain described observation Meta task collection.

Wherein, adopt area-method judge grid whether in covering tape particularly as follows: as it is shown on figure 3, set the central point that P point is certain grid, tetragon ABCD is a covering tape, if meeting equation (24), then grid P is covered by band.Otherwise, grid P is not at interior strips；If grid is covered by band, then the degree of covering of grid should be added 1.

S_ΔPAB+S_ΔPAD+S_ΔPDC+S_ΔPBC=S_ABCD(24)

Here, N is used_CoveredRepresenting the degree of covering of mesh point, time initial, the degree of covering of each mesh point is all designated as 0, then according to Meta task band number, and the situation that certain grid of cycle criterion is covered by band.If it is determined that covered by certain band, then the degree of covering of this mesh point adds 1, i.e. N_Covered=N_Covered+1.Calculate the degree of covering terminating to can be obtained by each mesh point.Certain grid described is capped the degree of covering referring to this grid more than 1.

If after certain region is divided, the grid number obtained is N, and after covering analyzing calculates, the grid number obtaining being covered by Meta task band is N₀, then the coverage rate in this region also can be calculated according to formula (25):

$R_{Covered}=\frac{N_0}{N}---\left(25\right)$

Wherein, described coverage rate is one of evaluation index of this planing method.

Step 112, utilizes evolution algorithmic that described observation Meta task collection carries out mission planning, obtains program results.

In this step, after getting observation Meta task collection, first described observation Meta task collection is carried out mission planning definition, set up satellite load scheduling model.

Specifically, described mission planning definition includes two parts: first is constraint definition, and second is load scheduling model；Wherein, in constraint definition, only consider the factor directly related with the problem to study and constraints.For observation Meta task collection, observation working time, sun altitude and side-sway angle, side-sway number of times are carried out constraint definition, additionally instruction template, instruction template interval and maximum operating time is defined.Concrete lexical or textual analysis is as shown in table 1.

Table 1

Further, in plan model, from logical resource list, each observation time window being embodied as an observation Meta task, these observation Meta task have regular time order, and the target of planning is to do for each task choosing or do not do.

First, for above-mentioned bound term, carrying out the definition of reasonable assumption and bound variable, concrete lexical or textual analysis is as follows:

1) assume total m visual time window, be designated asTime window W_vTime started and end time respectively S_vAnd E_v。

2) assume there be s completing of task, be designated as A={a₁,a₂,.....,a_s}；Each required by task time is D={d₁,d₂,.....,d_s, priority is p={p₁,p₂,.....,p_s}；

3) the time started variable of jth task is designated as s_j, end time variable is e_j；

4) definition assignment decisions variable t_jIf task can complete, then t_j=1, otherwise, t_j=0；

5) antenna conversion time r, namely earth station is after accomplishing a task, and performs the antenna attitude needed for next task and adjusts the time；Here, suppose that the antenna conversion time is unified；

6) before the imaging that instruction template requires, template time is T_cs, template time C after imaging_e；Instruction template interval I_t；

7) schedule start time is T_j, scheduling is T by the time_E；

8) the maximum observation duration of individual pen time is T_o, the maximum reception duration of individual pen time is T_r；

Then, based on, on the hypothesis basis of model, setting up following load scheduling model.Described load scheduling model regards the attributes such as satellite imagery importance degree, imaging side-looking angle, resource consumption, SEE time window, task observation time/frequency, loading demands as different resource, and between different resource, reach an optimum balance configuration, final goal is that observation mission is arranged in suitable time window, to reach minimizing of satellite resource consumption and scheduling time.

Specifically, described load scheduling model includes: optimization aim and consideration constraint；Specifically, described optimization aim includes:

$\underset{v=1}{\overset{n}{\&Sigma;}}{\mathrm{\&omega;}}_v{p}_v---\left(26\right)$

$min\underset{v=1}{\overset{m}{\&Sigma;}}{\mathrm{\&omega;}}_v---\left(27\right)$

$min\underset{v=1}{\overset{n}{\&Sigma;}}(t_v^{begin}-t_v^{early})/s---\left(28\right)$

$\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\}---\left(29\right)$

Wherein, formula (26) represents that maximization completes task；p_vBeing the priority of v task, s is total task number, ω_v=1 represents that task v is performed, ω_v=0 represents that task v is not performed.

Formula (27) minimizes resource consumption, and wherein the total number of remote sensor is m, ω_v=1 represents that the v remote sensor is used to certain or certain several imaging tasks, ω_v=0 represents that the v remote sensor is not applied to any imaging task.

Formula (28) represents and minimizes task waiting time, and wherein the start time of the time window that the v task is arranged to isThe early start observation moment of task v isThen the waiting time of task v is $t_v^{begin}-t_v^{early}.$

Formula (29) has represented that the total time of all tasks is minimum, whereinRepresent the tw of v subtask_vThe start time of individual time window,Represent the tw of v subtask_vThe finish time of individual time window.

Further, described consideration constraint includes:

${\mathrm{\&omega;}}_v(s_v,e_v)\&SubsetEqual;{\mathrm{\&omega;}}_v({\mathrm{TS}}_v,{\mathrm{TE}}_v),v\&Element;\&lsqb;1,N_T\&rsqb;---\left(30\right)$

t_jh(e_jh+C_s+C_e+I_t)≤t_jbs_jb, 1≤j≤n, 1≤jh≤jb≤n (31)

Wherein, formula (30) represents that the time started of all tasks and end time must in SEE time window ranges.

Formula (31) represents that the end time of all tasks is all not more than the observation mission time started performed thereafter interval time plus instruction template beginning and ending time and instruction template.Wherein: jh, jb represent former and later two adjacent task number in observation Meta task sequence respectively.

Here, described satellite load scheduling model also includes: satellite resource data base, resource access model.

Described resource access model provides the satellite resource data called in satellite resource data base to carry out the interface of data analysis, calculating for upper strata.And according to satellite, load, ground target physical characteristic, utilize STK to calculate satellite orbit parameter and observed object access relation that may be present and restriction relation, wherein, access relation refers to the access time window between each observation mission and satellite；Here restriction relation mainly has two, one is that the actual time of observation of observation mission must within the addressable time period of task, another be same satellite the adjacent observation mission of any two between can not be overlapping if having time, the content as shown in formula (30) and (31) respectively.

Secondly, after described satellite load scheduling model establishes, utilizing evolution algorithmic that described load scheduling model is solved, the solution obtained is program results.When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

Specifically, it is necessary first to evolution algorithmic is designed, comprise the following steps:

Step a, code Design: the chromosome needed for design evolution algorithmic.Evolution algorithmic is to utilize certain digital coding solved to represent (chromosome), acts on space encoder, solves clear and definite mathematical optimization problem.Each chromosome should represent a mission planning allocative decision, and each gene position identifies the detail of the program.Namely item chromosome represents the load scheduling scheme towards specific tasks between a kind of many stars.

Step b, operator designs: design the evolutional operations such as chromosomal variation, hybridization, selection, restructuring, it is ensured that the correctness of evolution result.Operator design depends on code Design, and therefore this partial design content must with code Design phase mutual feedback: code Design should be succinct as far as possible under the complete premise of guarantee information, simple as far as possible to ensure operator design；After operator design should ensure that chromosome evolution, the new population generated is still in strict conformity with the basic norm of code Design, in accordance with the concordance of chromosome format.

Step c, the evaluation function needed for design evolution algorithmic.Such as the correctness of load distribution, the redundancy of load distribution, the coverage effect (i.e. the observation performance of satellite) of earth observation, the number of satellite etc. employed；It is described as exactly a series of mathematics parameter and formula, as shown in formula (26)-(29), provides clear and definite target for evolutionary optimization.

Step d, constrained designs: the constraint function needed for design evolution algorithmic.Such as: under which kind of condition, target is possessed association when observed relationships, Multiple targets observation between various observed relationships by load, physical restriction and suffered by specific tasks Satellite, load, satellite are to the lift-launch ability of load, the load consumption ability etc. of load on the method for salary distribution of satellite, the track service condition of satellite, various star.It is described as exactly a series of mathematics parameter and formula, as shown in formula (30) (31), provides correct calculated direction for evolutionary optimization.

Step e, EVOLUTIONARY COMPUTATION possesses natural intrinsic parallism: the concrete feature for many stars mission planning problem realizes the many stars mission planning parallel schema based on EVOLUTIONARY COMPUTATION, improves the usefulness of mission planning.Specifically, the Parallel evolutionary algorithm adopted in the present embodiment is master-slave mode model and two kinds of parallel models of Isolate model.

Wherein, master-slave mode model (mast-slavemodel) parallel system is divided into a primary processor and several are from processor.The whole chromosome population of main processor monitors, and selection and the evolutional operation of algorithm is performed based on global statistics；Each individuality accepting to come host processor from processor is evaluated calculating, then result of calculation is returned to primary processor.Such rank parallel owing to adopting overall selection mode, so strict synchronization restriction is carried out in the evolution of colony.Master-slave mode model is applicable to that fitness evaluation is very time-consuming and situation considerably beyond call duration time.

And population is divided into several subgroups and distributes to each self-corresponding processor by Isolate model (islandmodel), each processor not only independently calculates fitness, and independently carry out selecting, recombinating intersection and mutation operation, also mutually to transmit the individuality that fitness is best termly, thus accelerating to meet the requirement of end condition.Isolate model belongs to distributed EVOLUTIONARY COMPUTATION, is current most widely used a kind of Parallel evolutionary algorithm.Isolate model is not high to parallel system Platform Requirements, it is possible to be loose couplings parallel system, the concurrency between main exploitation colony.

Step f, interaction design: based in many stars mission planning algorithm of EVOLUTIONARY COMPUTATION, should be the intelligent interaction strategy that evolution algorithmic provides certain, except the evolution characteristic of algorithm itself, also can merge the evaluation environment of outside, such as: the display decision-making of user, in good time adjustment etc. based on the changing in real time of resources mode of man-machine interaction, optimization aim, real intellectual evolution effect is reached.

After evolution algorithmic designs, utilize described evolution algorithmic that described satellite load scheduling model is optimized and solve, obtain program results.

Step 113, is decoded described program results, obtains observation sequence.

In this step, after getting described program results, it is determined that the search volume of EVOLUTIONARY COMPUTATION, set up evolutionary optimization model according to satellite load scheduling model, utilize evolution algorithmic that observation Meta task is screened, obtain meeting the best observation sequence of resource constraint.After observation sequence is formed, it is also possible to utilize described STK that described observation sequence is emulated.

Further, after program results gets, program results is analyzed, the data in planning process is added up by Utilization assessment algorithm, and for situations such as different basic evolution algorithmics, population scale, evolution algebraically, parallel features, adds up, analyzes the time response of EVOLUTIONARY COMPUTATION.

Specifically, when the data result of many stars mission planning effect is analyzed, by calling external program interface, analyze program results, generate data analysis form according to certain format.Described form can include .txt form.

When data in many stars mission planning process are added up, by calling external interface program, represent the mission planning effect variation tendency in EVOLUTIONARY COMPUTATION process with the form of function curve, curved surface.

When the many stars of planing method of many stars earth observation task that the present embodiment provides are observed over the ground, the features such as intelligent, the concurrency making full use of evolution algorithmic, solve the Mission Scheduling at many observation satellites needing collaborative work: utilize the dynamic task planing method based on evolution algorithmic, rational load scheduling scheme is formulated for satellite group in orbit, optimize the resource distribution of satellite system, abundant Appropriate application satellite system resource.And set up and there is general adaptive satellite task plan model, there is universality and extensibility.

Embodiment two

Corresponding to embodiment one, the present embodiment additionally provides the scheduling planning device of a kind of many stars earth observation task, and as shown in Figure 4, described device includes: read module 41, division module 42, planning module 43 and decoder module 44；Wherein,

Described read module 41 for reading satellite resource data from satellite resource data base, and described satellite resource data include: satellite information, payload information and land object information etc..Wherein, described satellite information includes: the track service condition of satellite, precision and side-sway ability.Described payload information refers to equipment and the load-carrying ability for obtaining view data that satellite carries, such as: camera (EO-1 hyperion, multispectral or panchromatic), synthetic aperture radar and various sensor (frequency range and wave band) etc..Described ground target refers to the ground region needing to be observed, including: the overlay capacity of point target and regional aim and ground target.

When described read module 41 reads satellite resource data, described division module 42 is used for, in STK, model of place is set up according to described satellite resource data, after model of place establishes, the computing module that accesses calling described STK calculates the SEE time window between satellite load and the ground target of observation mission；After described SEE time window calculation goes out, calculate the sub-satellite track of satellite, determine the described satellite substar position at the initial time place of described time window according to the initial time of sub-satellite track and described time window；Determine the four angular coordinate of described moonscope cover strip further according to described substar position, obtain described observation Meta task collection finally according to described observation covering tape.

Specifically, sub-satellite track requires over satellite ground tracks and determines, and satellite ground tracks need to be determined by reduced latitude and geocentric longitude.Accurate satellite ground tracks generally to calculate according to the integration of equation of satellite motion, in aerospace engineering is applied, owing to remote sensing satellite orbital eccentricity e is general close to zero, when therefore calculating the ground trace of remote sensing satellite, its track can be used as circular orbit when meeting certain required precision and process.Secondly, when consider non-spherical earth perturbation on satellite orbit when affecting, only J in Geopotential coefficient₂It is 10^-3Magnitude, all the other J_n(n=3,4 ...) and the magnitude of the humorous term coefficient in field be generally less than 10^-6, therefore its impact can be ignored.And J₂Coefficient is mainly due to caused by compression of the Earth f, so considering, at this, the first order perturbation that compression of the Earth factor causes.

Here, if Ω is right ascension of ascending node rate of change, ω_eFor the average spin velocity of the earth, then:

ω_e=7.292115 × 10^-5(rad/s)(1)

$\mathrm{\&Omega;}=-\frac{3}{2}{J}_2\sqrt{\mathrm{\&mu;}}\frac{a_e^2\cos i}{a^{7/2}{(1-e^2)}^2}(rad/s)---\left(2\right)$

Wherein, in formula (2), described a is the major radius of satellite orbit, and described e is eccentricity of satellite orbit, and described i is the inclination angle of satellite orbit, described a_eFor terrestrial equator footpath, for 6378137m；Described μ is geocentric gravitational constant, 3.986005 × 10¹⁴m³/s²；J₂For second order gravitational potential coefficient, it is 1.083 × 10^-3.So, satellite ascending node is relative to the western back speed rate ω ' of longitude zero point_eFor being represented by formula (3):

ω′_e=ω_e-Ω(3)

Known satellite mean angular velocity of satellite motion n is represented by

$n=\frac{\sqrt{\mathrm{\&mu;}}}{a^{3/2}}---\left(4\right)$

In fig. 2, X-axis points to the longitude zero point of International Time Bureau's definition.If satellite is through ascending node Ω₀Moment be t₀, longitude is λ₀.Through Δ t=t t₀After time, ascending node west retreats to Ω, and satellite arrives S, wherein,

ΩΩ₀=ω '_eΔt(5)

$\stackrel{\&cap;}{\mathrm{\&Omega;}S}=n\mathrm{\&Delta;}t---\left(6\right)$

In formula (7),Reduced latitude for t satellite.And satellite is relative to Ω₀Can be represented by formula (8) through difference:

$\mathrm{\&Delta;}\mathrm{\&lambda;}=\stackrel{\&cap;}{{\mathrm{\&Omega;}}_{0}D}---\left(8\right)$

In right angle spherical triangle △ Ω SD, ∠ S Ω D=i, i are known orbit inclination angle, so having:

$sin\stackrel{\&cap;}{SD}=\sin \stackrel{\&cap;}{\mathrm{\&Omega;}S}\sin i---\left(9\right)$

Formula (6) (7) is substituted in formula (9) and draws

So reduced latitude can be solved by formula (10)

Further, geodetic latitudeComputing formula be:

Wherein, in formula (12), f=1/298.257.

Order∠SΩ₀D=ψ；So, at spherical triangle Δ Ω S Ω₀In, according to cosine law for sides and five elements formula it follows that

$\begin{array}{c}\cos L={\mathrm{cos\&omega;}}_e^{\&prime;}\&CenterDot;\cos n\mathrm{\&Delta;}t+{\mathrm{sin\&omega;}}_e^{\&prime;}n\mathrm{\&Delta;}t\&CenterDot;\sin n\mathrm{\&Delta;}t\&CenterDot;\cos i-\sin L\&CenterDot;\cos \mathrm{\&psi;}\\ ={\mathrm{sin\&omega;}}_e^{\&prime;}\&CenterDot;\cos n\mathrm{\&Delta;}t-{\mathrm{cos\&omega;}}_e^{\&prime;}n\mathrm{\&Delta;}t\&CenterDot;\sin n\mathrm{\&Delta;}t\&CenterDot;\cos i\end{array}---\left(13\right)$

At right angle spherical triangle △ Ω₀In SD, have:

Therefore, by formula (13)-(15) it follows that

Then, formula (16), (17) draw:

$\tan \mathrm{\&Delta;}\mathrm{\&lambda;}=\frac{-{\mathrm{sin\&omega;}}_e^{\&prime;}\mathrm{\&Delta;}t\&CenterDot;\cos n\mathrm{\&Delta;}t+{\mathrm{cos\&omega;}}_e^{\&prime;}\mathrm{\&Delta;}t\&CenterDot;\sin n\mathrm{\&Delta;}t\&CenterDot;\cos i}{{\mathrm{cos\&omega;}}_e^{\&prime;}\mathrm{\&Delta;}t\&CenterDot;\cos n\mathrm{\&Delta;}t+{\mathrm{sin\&omega;}}_e^{\&prime;}\mathrm{\&Delta;}t\&CenterDot;\sin n\mathrm{\&Delta;}t\&CenterDot;\cos i}---\left(18\right)$

So, t geocentric longitude λ is

λ=λ₀+Δλ(19)

Here, geocentric longitude is equal to geodetic longitude, need not change.

But in above-mentioned derivation, the non-linear of the closely dynamic rate of heterogeneity and track owing to have ignored satellite motion speed, therefore to ensure computational accuracy, sub-satellite track is started at extrapolation and be must not exceed 1/4 circle from equator, it is to say, in above-mentioned formula, Δ t should meet:

-T/4 < Δ t < T/4 (20)

In formula (20) formula, T is track mean period, it is possible to calculated by formula (21):

T=2 π/n (21)

Wherein, in formula (21), n is the mean angular velocity of satellite motion of satellite.

Like this, when calculating sub-satellite track, need first to start northwards from ascending node every time, return again to after each extrapolation 1/4 circle to the south to start southwards from southbound node, northwards each extrapolation 1/4 circle, then start northwards from the ascending node of next circle, each extrapolation 1/4 to the south is enclosed, and calculates sub-satellite track so round and round again.

For example, as it is known that longitude of ascending node λ₀, take Δ t=1 when northwards extrapolating along satellite flight direction by ascending node, 2 ..., [T/4]；When extrapolating southwards against satellite flight direction by ascending node, take Δ t=-1 ,-2 ...-[T/4], symbol " [] " represents the integer part taking a number, and its algebraic symbol is identical with former number.The computing formula of substar reduced latitude and geocentric longitude is provided by (11) and (19) formula respectively.

Again for example, as it is known that southbound node longitude λ₀, when extrapolating southwards along satellite flight direction by southbound node, take Δ t=1,2 ..., [T/4], when northwards extrapolate in inverse satellite flight direction by southbound node, take Δ t=-1 ,-2 ... ,-[T/4].At this moment substar reduced latitude formula can change formula (22) into, and the computing formula of geocentric longitude is constant:

So, it is possible to calculate satellite substar the earth warp more per second, latitude.Between southbound node and ascending node or between ascending node and previous circle southbound node through difference Δ λ₀Can be drawn by formula (23):

${\mathrm{\&Delta;\&lambda;}}_0=\mathrm{\&pi;}-\frac{1}{2}{\mathrm{\&omega;}}_e^{\&prime;}T---\left(23\right)$

Here, step-length desirable 10s, 30s or 60s of Δ t, step-length is more little, and sub-satellite track is more smooth.

Further, described observation Meta task collection is obtained particularly as follows: using terrestrial equator as X-axis, using zero degree warp as Y-axis, set up rectangular coordinate system according to described observation covering tape, it is thus achieved that four angular coordinates of observed object.Using the minimum longitude in four coordinate points as the left margin of grid division；Using the maximum longitude in four coordinate points as the right margin of grid division；In four coordinate points, minimum latitude is as the lower boundary of grid division；Using the maximum latitude in four coordinate points as the coboundary of grid division.The corner angular zone obtained with this is divided into 100 and takes advantage of the small grid of 100.Adopt area-method to judge that grid is whether in covering tape, calculate coverage rate, obtain described observation Meta task collection.

Wherein, described division module 22 adopt area-method judge grid whether in covering tape particularly as follows: as it is shown on figure 3, set the central point that P point is certain grid, tetragon ABCD is a covering tape, if meeting equation (24), then grid P is covered by band.Otherwise, grid P is not at interior strips；If grid is covered by band, then the degree of covering of grid should be added 1.

S_ΔPAB+S_ΔPAD+S_ΔPDC+S_ΔPBC=S_ABCD(24)

Here, N is used_CoveredRepresenting the degree of covering of mesh point, time initial, the degree of covering of each mesh point is all designated as 0, then according to Meta task band number, and the situation that certain grid of cycle criterion is covered by band.If it is determined that covered by certain band, then the degree of covering of this mesh point adds 1, i.e. N_Covered=N_Covered+1.Calculate the degree of covering terminating to can be obtained by each mesh point.Certain grid described is capped the degree of covering referring to this grid more than 1.

If after certain region is divided, the grid number obtained is N, and after covering analyzing calculates, the grid number obtaining being covered by Meta task band is N₀, then the coverage rate in this region also can be calculated according to formula (25):

$R_{Covered}=\frac{N_0}{N}---\left(25\right)$

Wherein, described coverage rate is one of evaluation index of this planing method.

After described division module 42 gets described observation Meta task collection, utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results.Specifically, first described observation Meta task collection is carried out mission planning definition by described planning module 43, sets up satellite load scheduling model.

Specifically, described mission planning definition includes two parts: first is constraint definition, and second is load scheduling model；Wherein, in constraint definition, only consider the factor directly related with the problem to study and constraints.For observation Meta task collection, observation working time, sun altitude and side-sway angle, side-sway number of times are carried out constraint definition, additionally instruction template, instruction template interval and maximum operating time is defined.Concrete lexical or textual analysis is as shown in table 1.

Table 1

Further, in plan model, from logical resource list, each observation time window being embodied as an observation Meta task, these observation Meta task have regular time order, and the target of planning is to do for each task choosing or do not do.

First, for above-mentioned bound term, carrying out the definition of reasonable assumption and bound variable, concrete lexical or textual analysis is as follows:

1) assume total m visual time window, be designated asTime window W_vTime started and end time respectively S_vAnd E_v。

2) assume there be s completing of task, be designated as A={a₁,a₂,.....,a_s}；Each required by task time is D={d₁,d₂,.....,d_s, priority is p={p₁,p₂,.....,p_s}；

3) the time started variable of jth task is designated as s_j, end time variable is e_j；

4) definition assignment decisions variable t_jIf task can complete, then t_j=1, otherwise, t_j=0；

5) antenna conversion time r, namely earth station is after accomplishing a task, and performs the antenna attitude needed for next task and adjusts the time；Here, suppose that the antenna conversion time is unified；

6) before the imaging that instruction template requires, template time is T_cs, template time C after imaging_e；Instruction template interval I_t；

7) schedule start time is T_j, scheduling is T by the time_E；

8) the maximum observation duration of individual pen time is T_o, the maximum reception duration of individual pen time is T_r；

Then, based on, on the hypothesis basis of model, setting up following load scheduling model.Described load scheduling model regards the attributes such as satellite imagery importance degree, imaging side-looking angle, resource consumption, SEE time window, task observation time/frequency, loading demands as different resource, and between different resource, reach an optimum balance configuration, final goal is that observation mission is arranged in suitable time window, to reach minimizing of satellite resource consumption and scheduling time.

Specifically, described load scheduling model includes: optimization aim and consideration constraint；Specifically, described optimization aim includes:

$\underset{v=1}{\overset{n}{\&Sigma;}}{\mathrm{\&omega;}}_v{p}_v---\left(26\right)$

$min\underset{v=1}{\overset{m}{\&Sigma;}}{\mathrm{\&omega;}}_v---\left(27\right)$

$min\underset{v=1}{\overset{n}{\&Sigma;}}(t_v^{begin}-t_v^{early})/s---\left(28\right)$

$\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\}---\left(29\right)$

Wherein, formula (26) represents that maximization completes task；p_vBeing the priority of v task, s is total task number, ω_v=1 represents that task v is performed, ω_v=0 represents that task v is not performed.

Formula (27) minimizes resource consumption, and wherein the total number of remote sensor is m, ω_v=1 represents that the v remote sensor is used to certain or certain several imaging tasks, ω_v=0 represents that the v remote sensor is not applied to any imaging task.

Formula (28) represents and minimizes task waiting time, and wherein the start time of the time window that the v task is arranged to isThe early start observation moment of task v isThen the waiting time of task v is $t_v^{begin}-t_v^{early}.$

Formula (29) has represented that the total time of all tasks is minimum, whereinRepresent the tw of v subtask_vThe start time of individual time window,Represent the tw of v subtask_vThe finish time of individual time window.

Further, described consideration constraint includes:

${\mathrm{\&omega;}}_v(s_v,e_v)\&SubsetEqual;{\mathrm{\&omega;}}_v({\mathrm{TS}}_v,{\mathrm{TE}}_v),v\&Element;[1,N_T]---\left(30\right)$

t_jh(e_jh+C_s+C_e+I_t)≤t_jbs_jb, 1≤j≤n, 1≤jh≤jb≤n (31)

Wherein, formula (30) represents that the time started of all tasks and end time must in SEE time window ranges.

Formula (31) represents that the end time of all tasks is all not more than the observation mission time started performed thereafter interval time plus instruction template beginning and ending time and instruction template.Wherein: jh, jb represent former and later two adjacent task number in observation Meta task sequence respectively.

Here, described satellite load scheduling model also includes: satellite resource data base, resource access model.

Described resource access model provides the satellite resource data called in satellite resource data base to carry out the interface of data analysis, calculating for upper strata.And according to satellite, load, ground target physical characteristic, STK is utilized to calculate satellite orbit parameter and observed object access relation that may be present and restriction relation, here restriction relation mainly has two, one is that the actual time of observation of observation mission must within the addressable time period of task, another be same satellite the adjacent observation mission of any two between can not be overlapping if having time, the content as shown in formula (30) and (31) respectively.

Described mission planning model regards the attributes such as satellite imagery importance degree, imaging side-looking angle, resource consumption, SEE time window, task observation time/frequency, loading demands as different resource, and between different resource, reach an optimum balance configuration, final goal is that observation mission is arranged in suitable time window, to reach minimizing of satellite resource consumption and scheduling time.

Secondly, after described satellite load scheduling model establishes, described planning module 43 utilizes evolution algorithmic that load scheduling model is solved, and the solution tried to achieve is program results.When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

Specifically, it is necessary first to evolution algorithmic is designed, comprise the following steps:

Step a, code Design: the chromosome needed for design evolution algorithmic.Evolution algorithmic is to utilize certain digital coding solved to represent (chromosome), acts on space encoder, solves clear and definite mathematical optimization problem.Each chromosome should represent a mission planning allocative decision, and each gene position identifies the detail of the program.Namely item chromosome represents the load scheduling scheme towards specific tasks between a kind of many stars.

Step b, operator designs: design the evolutional operations such as chromosomal variation, hybridization, selection, restructuring, it is ensured that the correctness of evolution result.Operator design depends on code Design, and therefore this partial design content must with code Design phase mutual feedback: code Design should be succinct as far as possible under the complete premise of guarantee information, simple as far as possible to ensure operator design；After operator design should ensure that chromosome evolution, the new population generated is still in strict conformity with the basic norm of code Design, in accordance with the concordance of chromosome format.

Step c, the evaluation function needed for design evolution algorithmic.Such as the correctness of load distribution, the redundancy of load distribution, the coverage effect (i.e. the observation performance of satellite) of earth observation, the number of satellite etc. employed；It is described as exactly a series of mathematics parameter and formula, as shown in formula (26)-(29), provides clear and definite target for evolutionary optimization.

Step d, constrained designs: the constraint function needed for design evolution algorithmic.Such as: under which kind of condition, target is possessed association when observed relationships, Multiple targets observation between various observed relationships by load, physical restriction and suffered by specific tasks Satellite, load, satellite are to the lift-launch ability of load, the load consumption ability etc. of load on the method for salary distribution of satellite, the track service condition of satellite, various star.It is described as exactly a series of mathematics parameter and formula, as shown in formula (30) (31), provides correct calculated direction for evolutionary optimization.

Step e, EVOLUTIONARY COMPUTATION possesses natural intrinsic parallism: the concrete feature for many stars mission planning problem realizes the many stars mission planning parallel schema based on EVOLUTIONARY COMPUTATION, improves the usefulness of mission planning.Specifically, the Parallel evolutionary algorithm adopted in the present embodiment is master-slave mode model and two kinds of parallel models of Isolate model.

Wherein, master-slave mode model (mast-slavemodel) parallel system is divided into a primary processor and several are from processor.The whole chromosome population of main processor monitors, and selection and the evolutional operation of algorithm is performed based on global statistics；Each individuality accepting to come host processor from processor is evaluated calculating, then result of calculation is returned to primary processor.Such rank parallel owing to adopting overall selection mode, so strict synchronization restriction is carried out in the evolution of colony.Master-slave mode model is applicable to that fitness evaluation is very time-consuming and situation considerably beyond call duration time.

And population is divided into several subgroups and distributes to each self-corresponding processor by Isolate model (islandmodel), each processor not only independently calculates fitness, and independently carry out selecting, recombinating intersection and mutation operation, also mutually to transmit the individuality that fitness is best termly, thus accelerating to meet the requirement of end condition.Isolate model belongs to distributed EVOLUTIONARY COMPUTATION, is current most widely used a kind of Parallel evolutionary algorithm.Isolate model is not high to parallel system Platform Requirements, it is possible to be loose couplings parallel system, the concurrency between main exploitation colony.

Step f, interaction design: based in many stars mission planning algorithm of EVOLUTIONARY COMPUTATION, should be the intelligent interaction strategy that evolution algorithmic provides certain, except the evolution characteristic of algorithm itself, also can merge the evaluation environment of outside, such as: the display decision-making of user, in good time adjustment etc. based on the changing in real time of resources mode of man-machine interaction, optimization aim, real intellectual evolution effect is reached.

After evolution algorithmic designs, utilize described evolution algorithmic that described satellite load scheduling model is optimized and solve, obtain program results.

After described planning module 43 gets program results, described decoder module 44 for determining the search volume of EVOLUTIONARY COMPUTATION according to satellite load scheduling model, set up evolutionary optimization model, utilize evolution algorithmic that Meta task is screened, obtain meeting the best Meta task set of resource constraint, form observation sequence.After observation sequence is formed, described decoder module 44 can also utilize described STK that described observation sequence is emulated.

Here, described device also includes: analysis module 45, after described program results obtains, program results is analyzed for Utilization assessment algorithm, the data in planning process is added up by described analysis module 45, and for situations such as different basic evolution algorithmics, population scale, evolution algebraically, parallel features, add up, analyze the time response of EVOLUTIONARY COMPUTATION.

Specifically, when the data result of many stars mission planning effect is analyzed by described analysis module 45, by calling external program interface, analyze program results, generate data analysis form according to certain format.Wherein, described form can include .txt form.

When data in many stars mission planning process are added up, by calling external interface program, represent the mission planning effect variation tendency in EVOLUTIONARY COMPUTATION process with the form of function curve, curved surface.

Read module 41, division module 42, planning module 43, decoder module 44 and analysis module 45 that the present embodiment provides can by the central processing unit (CPU in this device, CentralProcessingUnit), digital signal processor (DSP, DigtalSignalProcessor), programmable logic array (FPGA, FieldProgrammableGateArray), micro-control unit (MCU, MicroControllerUnit) realizes.

When the many stars of planing method of many stars earth observation task that the present embodiment provides are observed over the ground, the features such as intelligent, the concurrency making full use of evolution algorithmic, solve the Mission Scheduling at many observation satellites needing collaborative work: utilize the dynamic task planing method based on evolution algorithmic, rational load scheduling scheme is formulated for satellite group in orbit, optimize the resource distribution of satellite system, abundant Appropriate application satellite system resource.And set up and there is general adaptive satellite task plan model, there is universality and extensibility.

Embodiment three

During practical application, user can utilize evolution algorithmic that satellite resource data are planned by the man machine interface of many stars earth observation task system.

Specifically, as described in Figure 5, described system includes: application layer, algorithm layer, model layer and data Layer；Wherein, data Layer provides based on physical data necessary to many stars mission planning of EVOLUTIONARY COMPUTATION.The mission planning of many stars needs substantial amounts of data input and output, for improving operation efficiency, need to provide the data managing method with certain independence.Build the data base that many stars multitask planning is relevant for this, wherein based on observation mission storehouse, load storehouse, satellite storehouse, also include some complementary data in addition.

Model layer realizes the modeling process needed for the mission planning of many stars.First the structure of physical data Access Model is completed, it is provided that the basic operation computing to lower floor's physical data；Secondly physical data information various in resources bank and relation data thereof are carried out abstract by completion logic Resource Abstract model, and Unify legislation is logic satellite resource data；Finally setting up the many stars mission planning model in the face of logical resource, this model is also the direct acting model of EVOLUTIONARY COMPUTATION.

Algorithm layer realizes the evolution method for solving for the mission planning of many stars.By the plan model that model layer provides, mission planning is converted into the optimization problem suitable in Evolutionary Computation, utilizes evolution algorithmic to solve.Content specifically includes that the choosing of basic algorithm, the design of chromosomal coded system, evolutionary operator, the design of dynamic interaction evolutionary pattern, universality are developed the method etc. of planning.Additionally, also provide for parallel computation pattern at algorithm layer, to improve the usefulness of planning of developing.

Application layer realizes the application computation schema when solving actual task planning problem.Prototype system realizes an Exemplary software in the way of man-machine interaction, for verifying the correctness of the many stars mission planning algorithm based on EVOLUTIONARY COMPUTATION, and its evolutionary process is carried out performance evaluation.

Further, as shown in Figure 6, described system specifically includes that graphical interfaces module 61, data management module 62, scheduling pretreatment module 63, genetic module 64 and algorithm performance evaluation module 65.

Wherein, graphical interfaces module 61 provides operation interface for other functional modules, with visualization, friendly interface access stencil for user, is the bridge connecting user with other functional modules, is also the interface basis realizing other program modules.

The operation interface that data management module 62 accesses for systems with data storehouse, operates including database parameter configuration, data query, renewal, insertion, deletion etc., and provides the function of abnormality processing.It utilizes the bottom of ADO fulfillment database to access, above, it is respectively directed to data base querying, renewal, insertion, deletion, the functions such as encapsulation of data inquiry, data renewal, data insertion, data deletion, and corresponding abnormality processing module is provided, ensure availability and the integrity of data, it is to avoid the situation that data are inconsistent occurs；And data manipulation result or error message can be returned to upper layer module, use for upper layer module or terminal use.This module is realize the data basis of other program modules.

Scheduling pretreatment module 63 is the intermediate module connecting mission planning software with external software STK.Its first effect is that the data such as satellite, load, observation mission are carried out pretreatment, utilizes the access computing function of STK, and physical data converts to logical data observation Meta task；Second effect is to utilize the two dimension of STK, three-dimensional artificial function, the final observation program obtained through evolutionary optimization is carried out simulation demo, the reliability of the result.This module provides Data Source for evolution planning module.

Genetic module 64 is the nucleus module of native system.It utilizes the logical data observation Meta task that scheduling pretreatment module is produced by evolution algorithmic to carry out planning of developing, and generates and meets the best observation program that user requires.Evolutionary optimization module mainly includes three parts: evolution algorithmic storehouse, individual evaluation, parallel evolutionary realize.Evolution algorithmic storehouse provides several bases evolution algorithmic, comprises single object optimization algorithm and multi-objective optimization algorithm, is the algorithm basis of the module of whole planning of developing；Individual evaluation is that evolution planning can obtain providing safeguard of optimal solution；Parallel evolutionary its mainly serial evolution algorithmic is improved so that it is develop in a parallel mode, to improve the execution efficiency of evolution algorithmic.

Algorithm performance evaluation module 65 is the supplementary module of system.Realizing, to the change trend curve emulation of optimal solution in evolutionary process, allowing user can understand implementation effect and the performance indications of various evolution algorithmic intuitively, thus allowing user that suitable algorithm can be selected to carry out mission planning, or adjusting related algorithm parameter in good time.

User is when using this system that satellite resource data are planned, idiographic flow is:

First, a newly-built engineering.Engineering is used for preserving task scheduling data and correlation engineering setting, and described engineering installation includes mission requirements data, satellite load data, pre-processed results, algorithm parameter setting, final program results etc. and all can be saved in engineering.

After new construction, all settings are initially empty, and now connect data base, newly-built task in main interface, open task button and can become upstate, and next prompting needs to click these buttons and carry out subsequent operation setting.

Secondly, it is possible to " the data base's configuration " in click tools hurdle, satellite resource data base is connected with this.Read satellite resource data, begin a task with scheduling.

The interface connecting data base provides two interfaces, can select to connect oracle database or Access data base, connect oracle database and need to input the information such as user name, password, database service name and server ip address, connect Access data base and have only to find in the machine the Access data base of correspondence.

Then, click " scheduling pretreatment " module and be scheduling the pretreatment of task.Specifically, it is scheduling pretreatment by calling external software STK, according to the satellite load information read from data base, and calculative demand mission bit stream, call the access computing module of STK to set up STK scene observation mission is divided, obtain observation Meta task collection.Wherein, scheduling pretreatment module can also utilize the two dimension of STK, three-dimensional artificial function, and the final observation program obtained through evolutionary optimization carries out simulation demo, verifies the reliability of program results, as shown in Figure 7.

After being scheduling pretreatment, obtaining task scheduling pre-processed results, result is the observation Meta task representated by single band.Next it is optimized goal setting exactly.

Specifically, optimization aim has four optimization aim respectively for single goal and multiple target in arranging: total revenue is maximum, and task completes number at most, and satellite utilization rate is the highest, the target that priority is high.

Wherein, total revenue reflects the covering performance index over the ground of moonscope to a certain extent.Optimize observation total revenue and can guarantee that the task that priority is high reaches higher coverage rate as far as possible.

The described number of tasks that completes refers to that at most to ensure that Meta task is as much as possible is done.

Described satellite utilization rate is generally referred to as time availability, is completed by the load on satellite owing to earth observation is actually, needs to refine to imaging load so the time availability of satellite calculates.The time availability of described imaging load is the ratio working time accounting for load time in orbit.

The task that priority is higher needs to arrange it to be done as early as possible.So, it is necessary to the waiting time making task is minimum.Because contingency tasks generally has higher priority, therefore this optimization aim is applicable to emergency scene.

Further, for single goal, a target can only being selected to be optimized, when selecting multiple target, can provide corresponding information and inform and cannot select, multiple target may select 2 and 3 optimization aim.

Here, optimization aim has selected to select next step can enter into algorithm parameter setting afterwards.Specifically, algorithm parameter arranges main offer parameters setting.Described algorithm includes: single goal algorithm and multi-objective Algorithm.Single goal algorithm includes DE algorithm and PSO algorithm, and multi-objective Algorithm includes MODE/D algorithm and NSGAII algorithm.

After algorithm parameter is provided with, enter parallel parameter configuration interface configuration parallel parameter.

If optimization aim is directed to single goal, parallel parameter configuration interface can arrange parallel line number of passes and migrate algebraically；If optimization aim is directed to multiple target, parallel parameter configuration interface may only arrange parallel line number of passes.Parallel line number of passes can arrange the Thread Count of unit parallel running, and the scope of setting is from 1-16.

It addition, migrating algebraically is for exchange information between sub-population, migrate algebraically more big, then the information that exchanges between sub-population is more few, is unfavorable for the evolution of whole population, and the result therefore obtained is by less desirable, but it is too much to migrate algebraically, then efficiency will relatively under.Therefore migration algebraically is set to 5.

The above, be only presently preferred embodiments of the present invention, is not intended to limit protection scope of the present invention, all any amendment, equivalent replacement and improvement etc. made within the spirit and principles in the present invention, should be included within protection scope of the present invention.

Claims (10)

1. the scheduling planning method of the task of star earth observation more than a kind, it is characterised in that described method includes:

Read observation mission and satellite resource data；

According to described satellite resource data, described observation mission is divided, obtain observation Meta task collection；

Utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtain program results；

Described program results is decoded, obtains observation sequence；Wherein,

When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

2. the method for claim 1, it is characterised in that described according to described satellite resource data, observation mission is divided, obtains observation Meta task collection and includes:

In satellite tool kit STK, model of place is set up according to described satellite resource data；

Calculate the SEE time window between satellite load and ground target；

Calculate the sub-satellite track of satellite；

Initial time according to sub-satellite track and described time window determines the described satellite substar position at the initial time place of described time window；

The four angular coordinate of described moonscope cover strip is determined according to described substar position；

Described observation Meta task collection is obtained according to described observation covering tape.

3. the method for claim 1, it is characterised in that utilize evolution algorithmic that described observation Meta task collection carries out mission planning, obtains program results and includes:

Side-sway number of times, observation working time, sun altitude and side-sway angle that described observation Meta task is concentrated carry out constraint definition；

Side-sway count constraint, the constraint of observation working time, sun altitude constraint and side-sway angle restriction are carried out it is assumed that set up satellite load scheduling model on the basis assumed；

Utilize described evolution algorithmic that described satellite load scheduling model is solved, obtain program results.

4. the method for claim 1, it is characterised in that described satellite resource data include: satellite information, payload information and land object information.

5. method as claimed in claim 3, it is characterised in that described satellite load scheduling model includes: $\underset{v=1}{\overset{n}{\&Sigma;}}{\mathrm{\&omega;}}_v{p}_v,min\underset{v=1}{\overset{m}{\&Sigma;}}{\mathrm{\&omega;}}_v,\min \underset{v=1}{\overset{n}{\&Sigma;}}(t_v^{begin}-t_v^{early})/s$ And $\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\};$

Wherein, described inRepresent that maximization completes task；Described p_vBeing the priority of v task, described s is total task number；

DescribedFor minimizing resource consumption；Described m is the total number of remote sensor；

DescribedRepresent and minimize task waiting time；DescribedIt is the start time of the time window that the v task is arranged to, described inEarly start for task v observes the moment, described inWaiting time for task v；

DescribedRepresent that the total time of all tasks is minimum；DescribedRepresent the tw of v subtask_vThe start time of individual time window, described inRepresent the tw of v subtask_vThe finish time of individual time window.

6. the scheduling planning device of the task of star earth observation more than a kind, it is characterised in that described device includes:

Read module, is used for reading observation mission and satellite resource data；

Divide module, for observation mission being divided according to described satellite resource data, obtain observation Meta task collection；

Planning module, is used for utilizing evolution algorithmic that described observation Meta task collection carries out mission planning, obtains program results；

Decoder module, for described program results is decoded, obtains observation sequence；Wherein,

When described program results is described many stars earth observation, by corresponding with described observation mission to satellite orbit service condition, satellite load ability, the satellite load scheduling scheme of formation.

7. device as claimed in claim 6, it is characterised in that described division module specifically for:

In satellite tool kit STK, scene is set up according to described satellite resource data；

Calculate the SEE time window between satellite load and ground target；

Calculate the sub-satellite track of satellite；

Initial time according to sub-satellite track and described time window determines the described satellite substar position at the initial time place of described time window；

The four angular coordinate of described moonscope cover strip is determined according to described substar position；

Described observation Meta task collection is obtained according to described observation cover strip.

8. device as claimed in claim 6, it is characterised in that described planning module specifically for:

Side-sway number of times, observation working time, sun altitude and side-sway angle that described observation Meta task is concentrated carry out constraint definition；

Side-sway count constraint, the constraint of observation working time, sun altitude constraint and side-sway angle restriction are carried out it is assumed that set up satellite load scheduling model on the basis assumed；

Utilize described evolution algorithmic that described satellite load scheduling model is solved, obtain program results.

9. device as claimed in claim 6, it is characterised in that described satellite resource data include: satellite information, payload information and land object information.

10. device as claimed in claim 6, it is characterised in that described satellite load scheduling model includes: $\underset{v=1}{\overset{n}{\&Sigma;}}{\mathrm{\&omega;}}_v{p}_v,min\underset{v=1}{\overset{m}{\&Sigma;}}{\mathrm{\&omega;}}_v,\min \underset{v=1}{\overset{n}{\&Sigma;}}(t_v^{begin}-t_v^{early})/s$ And $\min \{\max {\mathrm{ETW}}_v^{{\mathrm{tw}}_v}-\min {\mathrm{STW}}_v^{{\mathrm{tw}}_v}\};$

Wherein, described inRepresent that maximization completes task；Described p_vBeing the priority of v task, described s is total task number；

DescribedFor minimizing resource consumption；Described m is the total number of remote sensor；

DescribedRepresent and minimize task waiting time；DescribedIt is the start time of the time window that the v task is arranged to, described inEarly start for task v observes the moment, described inWaiting time for task v；

DescribedRepresent that the total time of all tasks is minimum；DescribedRepresent the tw of v subtask_vThe start time of individual time window, described inRepresent the tw of v subtask_vThe finish time of individual time window.