Space-based space target polling observation on-orbit task planning method based on time window conflict degree

Abstract

The invention relates to a space-based space target polling observation on-orbit task planning method based on a time window conflict degree, and the method comprises the steps: constructing a space-based space target polling observation task planning problem mathematical model, and determining a space-based space target polling observation task optimization target; slicing the observation time window; according to a priority rule and a time window conflict degree, carrying out a first round of solution on the sliced visible window set to meet a first-level planning target as much as possible: the target polling number in a polling period is maximum; fine adjustment is carried out on the basis of the first round of solving results, and the target can be observed completely; adjustment is carried out on the basis of the result of the previous step, workload balance of all the satellites is achieved, and the second-level planning target is met as far as possible: the variance of the number of polling targets of all the satellites is minimum. According to the method, the space-based space target polling observation on-orbit task planning problem is efficiently and rapidly solved under the condition of low calculation power, and the problems that the calculation amount is large, the convergence speed is low and the requirement for the calculation power is large are solved.

Description

Technical Field

[0001] The invention belongs to the field of satellite observation mission planning, and in particular relates to an on-orbit mission planning method for patrol observation of space-based space targets based on time window conflict degree. Background Art

[0002] The satellite mission planning problem is to reasonably arrange satellite observation resources and give the order in which satellites perform observation tasks under the condition of satisfying various constraints between satellites and the targets to be observed. Its essence is an NP-hard constraint satisfaction problem. The upper bound of the solution space scale increases sharply with the increase in the number of satellites and observation requirements.

[0003] Space-based space target patrol observation refers to the use of space-based observation satellites to conduct patrol observations on space targets such as spacecraft and space debris based on the orbital element information of the target. The mission planning algorithms currently available for space-based space target patrols mainly include local search algorithms, mixed integer programming, genetic algorithms, ant colony algorithms, etc. The results obtained by optimization algorithms such as local search algorithms and mixed integer programming algorithms are global optimal solutions, but the amount of calculation increases exponentially with the increase in the number of satellites and targets. They are only applicable to small-scale mission planning problems and are not applicable to situations with a large number of satellites or targets. Multi-satellite mission planning is a complex problem with multiple task requests, multiple available resources, multiple constraints, and multiple optimization objectives. It is very unrealistic to traverse all solution spaces. For problems with large numbers of tasks and high scheduling complexity, intelligent optimization algorithms such as genetic algorithms and ant colony algorithms take a long time to solve and cannot adapt to dynamic task changes. When satellites plan autonomous missions, the on-board computing power is weak and cannot meet the computing resource requirements of intelligent optimization algorithms such as genetic algorithms and ant colony algorithms. The satellite's single observation time window for the target is relatively long. When solving the problem of on-orbit mission planning for space-based target patrols, it is necessary to optimize the observation window selection and the observation time period in the observation window at the same time, and the optimization coupling is relatively high. Summary of the invention

[0004] In order to solve the above technical problems, the present invention aims at the problem of on-orbit task planning for patrol observation of space-based space targets, takes into account the limitation of on-board computing resources, and proposes a method for on-orbit task planning for patrol observation of space-based space targets based on time window conflict in order to meet the efficient utilization of satellite observation resources and the patrol observation task requirements of space targets based on the greedy algorithm idea. On the basis of constructing a mathematical model of the problem of patrol observation of space-based space targets, by slicing the time window and adopting a task priority determination method based on the time window conflict degree, a three-round planning method is adopted to achieve the optimization goals of observing as many space targets as possible and balancing the workload of each satellite, thereby realizing the efficient solution of the problem of on-orbit task planning for patrol observation of space-based space targets. At the same time, the method is also applicable to the on-orbit task planning for observation of ground targets. The method specifically includes the following steps:

[0005] Step 1: Construct a mathematical model for the space-based space target patrol observation mission planning problem and determine the optimization target of the space-based space target patrol observation mission;

[0006] Step 2: Slice the observation time window;

[0007] The visible time window can be expressed as Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r >, whose set is represented by Wins={Win1,Win2,…,Win i ,…,Win n}, slice the observation time window to form a sliced ​​visible window set

[0008] Step 3: According to the priority rule, the first round of solving is performed on the sliced ​​visible window set according to the time window conflict degree, so as to meet the first-level planning goal as much as possible: the maximum number of target patrols within the patrol cycle;

[0009] Step 4: Make fine adjustments based on the results of the first round of solutions to achieve the goal of observing as much as possible;

[0010] Step 5: Make adjustments based on the results of step 4 to achieve a balanced workload for each satellite and meet the secondary planning goal as much as possible: the variance of the number of patrol targets for each satellite is minimized.

[0011] The above time window conflict degree calculation method is:

[0012] The conflicting window set of each window in list_Win is counted. If the load capacity constraint or observation transition time constraint is not met, it is judged as a conflict. The judgment rule is:

[0013] If Win i |sat=Win j |sat, then perform conflict judgment:

[0014] if winST i ≤winST j

[0015] if(winST j -winET i )<B i,j

[0016] then conflict i,j =1

[0017] if winST i >winST j

[0018] if(winST i -winET j )<B i,j

[0019] then conflict i,j =1

[0020] Among them, conflict i,j =1 indicates window Win i and Win j A conflict occurs, B i,j Represents the window Win i ,Win j The duration of the turntable maneuvering between

[0021] Win i The conflict set can be expressed as

[0022] The conflict degree of a time window is defined as the number of time windows among all time windows that have conflicting observation time and switching duration with this window, that is:

[0023]

[0024] In the formula, Indicates window win i and win j There is a conflict, Indicates that there is no conflict;

[0025] If the time window Win i The conflict is combined into The time window Win i The conflict degree is k.

[0026] The above priority rules are:

[0027] For the same target, the target with fewer windows has higher priority; for windows with the same conflict degree, the window with higher average observed brightness during the window duration has higher priority;

[0028] For windows between different targets, the target with fewer windows has a higher priority; for targets with the same number of windows, the target with a smaller degree of window conflict has a higher priority; for targets with the same number of windows and the same degree of conflict, the target with a higher average observed brightness has a higher priority.

[0029] Step 1 specifically includes:

[0030] Step 1.1: Description of the problem of planning the space-based space target patrol observation mission;

[0031] The problem of space-based space target patrol mission planning is described as N satellites observing r targets in a planning cycle, so that the optimization target of space-based space target patrol mission is optimal. The final output of mission planning is mainly the allocation plan of observation tasks.

[0032] The single-star patrol mission planning problem is expressed as:

[0033] SPMP={Sat,Tar,Wins,RES}

[0034] in:

[0035] Sat——satellite set, there are N satellites in total, Sat={sat1,sat2,…,sat N};

[0036] Tar——the set of observed targets, there are r targets to be observed, Tar={tar1,tar2,...,tar i ,...,tar r};

[0037] Wins——The set of visible windows of the satellite Sat observation target Tar, n represents the total number of time windows to be planned, Wins={Win1,Win2,...,Win i ,...,Win n}; Win means satellite sat m Observation target tar r Visible window, Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r>winST mr 、winET mr Indicates satellite m Observation target tar r The start time and end time of the visible window; winDT mr , satellite m Observation target tar r The duration of the visible window, winDT mr =winET mr -winST mr ;

[0038] RES – Constraint Set;

[0039] Step 1.2: Mathematical description of the space-based space target patrol mission constraints;

[0040] The constraints are: load capacity constraint, observation time window constraint, observation transition time constraint and patrol time interval constraint;

[0041] Payload capacity constraints: Due to the limitations of satellite payload characteristics, a satellite can only observe one target at a time, that is, each onboard remote sensor can only perform one observation task at any time;

[0042] Observation time window constraint: If the target tar r By satellitesat m To make an observation, the duration interval during which the satellite observes it must be a subset of the time window during which the satellite is visible to it;

[0043] For the visible window win i , if the target tar r By satellitesat m To conduct observation, the following conditions must be met:

[0044] ST mr ≥winST mr ,ET mr ≤winET mr

[0045] Observation transition time constraints: For satellite sat m For a task sequence, the time interval between the end of the previous task and the start of the next task should be greater than or equal to a transition time;

[0046] If the target tar i and tar j All of them need to occupy satellite sat m , and for the target tar i The observation task is followed by the target tar jAfter the observation task is executed, it must meet the following requirements:

[0047] ST mj -ET mi ≥B i,j

[0048] Among them B i,j Indicates satellite m Observation target tar i and tar j Required observation transition time;

[0049] Patrol time interval constraint: When performing a single-satellite patrol, it is necessary to ensure that the observation arcs are as dispersed as possible, and the time interval between two observations of the same target must be greater than one time period:

[0050]

[0051] tar i =tar j ,tar k ∈Win k ,k=i,j

[0052]

[0053] Where, MinObserveInterval is the interval between two adjacent time windows for observing the same target;

[0054] Step 1.3: Mathematical description of the optimization target of the space-based space target patrol mission:

[0055] The optimization objectives of the space-based space target patrol mission are divided into primary planning objectives and secondary planning objectives:

[0056] First-level planning goal: the maximum number of target patrols within the patrol cycle, that is, to ensure that all targets are observed once;

[0057] For the target set to be patrolled, Tar = {tar1, tar2, ..., tar i ,...,tar r}, the number of targets is r, and the observed state is V = {v1, v2, …, v i ,...,v r}, where v i = 0 or 1, 0 means not observed, 1 means observed, then the first-level planning target can be expressed as:

[0058]

[0059] Secondary planning goal: Minimize the variance of the number of patrol targets for each star;

[0060] Assume that the target number set of each star patrol is Amount = {a1, a2, ..., a i ,…,a r},a i ≥0, a i represents the target number of the i-th satellite patrol, then the secondary planning target can be expressed as:

[0061]

[0062] Step 2: The method for segmenting the visible window is as follows:

[0063] Set the window split time split_time to the required observation time of the satellite observation target, that is, for the split time window Win i , split_time = |winET i -winST i |; The interval_time difference between the start and end time of the split window is the satellite sat after splitting m tar the target r The difference between the start times of two adjacent time windows of observation, that is: |winST i -winST j |=interval_time×n(n=1,2,…,m);

[0064] Determine the duration of the time window winDT mr Is it greater than the splitting duration split_time? If so, continue splitting; otherwise, split the next time window;

[0065] Calculate the number of time window divisions m:

[0066]

[0067] The time window is divided, and the parameter calculation method for each time window is as follows:

[0068] winST i =winST mr +interval_time×(i-1),(i=1,2,...,m..)

[0069] winET i =winST i +split_time

[0070] winDT i =split_time

[0071] Win i = <winST i ,winET i ,winDT i ,sat m ,tar r >

[0072]

[0073] For the split time window set Sort by the order of the time window start time winST.

[0074] Step 3 specifically includes:

[0075] Step 3.1: Calculate the conflict degree of each time window;

[0076] Step 3.2: Determine the priority of the time window according to the priority rule and select the time window with the highest priority win select ;

[0077] Step 3.3: Determine whether the planning is completed and update the time window sequence list_Win;

[0078] Step 3.3 According to the selected window win select , determine whether the planning is completed and update the time window sequence list_Win. The specific steps are as follows:

[0079] Step 1: Count the number of observations of each target. If all targets have been observed, step 3 ends. Otherwise, execute step 2.

[0080] Step 2: Calculate win in the time window sequence list_Win select The conflict set And delete it in the time window sequence list_Win;

[0081] Step 3: Delete win in the time window sequence list_Win select Window observation target tar select All windows of

[0082] Step 4: If there is no window in the time window sequence list_Win, then step 3 ends, otherwise execute step 3.1.

[0083] The steps to select the time window with the highest priority are as follows:

[0084] Calculate the number of windows for each target, and obtain the minimum number of windows for the target by comparing the number of windows for each target. tar , set the number of windows to Mintar The target window is stored in the window collection list_Win1;

[0085] Compare the conflict degree Con of each window in the window set list_Win1 i , get the minimum conflict degree Min conflict , set the conflict degree to Min conflict The window is stored in the window collection list_Win2;

[0086] Compare the brightness of each window in the window set list_Win2 and select the window with the highest brightness win select , add it to the task planning result sequence list_win_chose.

[0087] Step 4 is as follows:

[0088] Step 4.1: Count the targets that were not observed in the first round of solution;

[0089] Step 4.2: Count the time window Win of each unobserved target i Con with list_win_chose i , and Win i The corresponding conflict window set Conflict i ;

[0090] Step 4.3: Prioritize the unobserved targets according to the observed target priority rules, and first fine-tune the target with the highest priority, Tar prio2 , for win select2 ∈Tar prio2 , adjust the win in list_win_chose select2 Conflicting target windows, make target Tar prio2 can be observed;

[0091] Step 4.4: Update the priorities of all remaining targets and obtain the new highest priority target Tar prio2 , execute step 4.3 again.

[0092] Step 4.3 The specific steps are as follows:

[0093] Traversing Tar prio2 The available time window is counted in the planning result sequence list_win_chose with win select2 The time window with conflicts is added to the conflict sequence list_win_conflict;

[0094] Traverse the conflict sequence list_win_conflict, for the window win i , if there is a time window win to observe the same target j ,win j Not in the result sequence list_win_chose and does not conflict with the time window in the result sequence, and win j With win select2 If there is no time conflict, win j With win select2 Add to the planning result sequence list_win_chose and remove win i .

[0095] Step 5 is to make adjustments based on the results of step 4 to achieve a balanced workload for each satellite. The specific steps are as follows:

[0096] Step 5.1: Determine whether the satellite workload is balanced: Let the number of targets observed by the satellite be x i , calculate the average and standard deviation std respectively, and define: if x i >average+3×std, then satellite i is considered to have too much workload; if x i <average-3×std, then the workload of satellite i is considered to be too small;

[0097] Step 5.2: Count the satellites with too much workload and too little workload respectively, and record the satellite sequence with too much workload as list_sat over , the satellite sequence with too little workload is recorded as list_sat below ;

[0098] Step 5.3: Prioritize list_sat over The satellite with the most workload over Adjust, for win∈sat over , prioritize adjusting the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , try to make the number of satellite observation targets within the interval [average-3×std,average+3×std];

[0099] Step 5.4: Update satellite workload, count satellites with too little workload, and record the sequence of satellites with too little workload as list_sat below , the rest is list_sat other ;

[0100] Step 5.5: List_sat otherSatellite with normal workload other Adjust, for win∈sat other , adjust the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , increase the number of observation windows of satellites with less workload, and try to make the number of satellite observation targets within the interval of [average-3×std,average+3×std].

[0101] Beneficial effects:

[0102] 1. By slicing the time window, the coupled optimization problem of selecting the observation window and the observation time period in the observation window is decoupled and converted into a method of planning the sliced ​​time window;

[0103] 2. A method for determining task priorities based on time window conflict is proposed, which clarifies the optimization order between different goals and the selection method between different windows for the same goal;

[0104] 3. Based on the greedy idea, the present invention achieves efficient and rapid solution to the on-orbit planning problem of space-based space target patrol missions by setting rules under the condition of using lower computing power, solving the problems of large computational complexity of existing optimization algorithms, slow convergence speed of intelligent optimization algorithms and large computing power requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0105] Figure 1 This is a schematic diagram of the time window constraints for the observation task;

[0106] Figure 2 Schematic diagram of the observed transition time constraints;

[0107] Figure 3 This is a schematic diagram of time window segmentation;

[0108] Figure 4 Segment the flow chart for time windows;

[0109] Figure 5 Flowchart for the first round of solution for step 3;

[0110] Figure 6 Fine-tune the flow chart for the window;

[0111] Figure 7 Assign a Gantt chart to the first round of planning tasks for 60 targets in a simulated patrol scenario for 1 hour;

[0112] Figure 8 Assign a Gantt chart to the second round of planning tasks for 60 targets in a simulated patrol scenario for one hour;

[0113] Fig. 9 This is a comparison chart of the cumulative observation time of each satellite in the second round of planning within 1 hour of simulation;

[0114] Fig.10 Assign a Gantt chart to the third round of planning tasks for 60 targets in a simulated patrol scenario for one hour;

[0115] Fig.11 This is a comparison chart of the cumulative observation time of each satellite in the third round of planning within 1 hour of simulation. DETAILED DESCRIPTION

[0116] The technical solution of the present invention is further described below in conjunction with the embodiments, but is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention should be included in the protection scope of the present invention.

[0117] A method for planning an on-orbit mission for patrol observation of space-based space targets based on time window conflict degree specifically comprises the following steps:

[0118] Step 1: Construct a mathematical model for the space-based space target patrol observation mission planning problem and determine the optimization target of the space-based space target patrol observation mission;

[0119] Step 1 specifically includes:

[0120] Step 1.1: Description of the problem of planning the space-based space target patrol observation mission;

[0121] The problem of space-based space target patrol observation mission planning is described as N satellites observing r targets in a planning cycle, so that the optimization target of space-based space target patrol mission observation is optimal. The final output of mission planning is mainly the allocation plan of observation tasks.

[0122] The single-star patrol observation mission planning problem is expressed as:

[0123] SPMP={Sat,Tar,Wins,RES}

[0124] in:

[0125] Sat——satellite set, there are N satellites in total, Sat={sat1,sat2,…,sat N};

[0126] Tar——Observation target set, there are r satellites in total, Tar={tar1,tar2,...,tar i ,…,tar r};

[0127] Wins is the set of visible windows of the satellite Sat observation target Tar, n represents the total number of time windows to be planned, Wins = {Win1, Win2, ..., Win i ,…,Win n}; Win means satellite sat m Observation target tar r Visible window, Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r >winST mr 、winET mr Indicates satellite m Observation target tar r The start time and end time of the visible window; winDT mr , satellite m Observation target tar r The duration of the visible window, winDT mr =winET mr -winST mr ;

[0128] RES – Constraint Set;

[0129] At the same time, the following assumptions are made:

[0130] ② Each observation task is independent of each other;

[0131] ② Satellite equipment failure is not considered;

[0132] ③ Extreme and special working conditions are not considered.

[0133] Based on the above definitions and assumptions, the purpose of space-based space target patrol observation mission planning is to select a subset from the visible window set Wins so that the generated planning scheme can obtain the maximum target observation benefit while satisfying the constraint set RES.

[0134] Step 1.2: Mathematical description of the space-based space target patrol observation mission constraints;

[0135] The constraints are: load capacity constraint, observation time window constraint, observation transition time constraint and patrol time interval constraint;

[0136] Payload capacity constraints: Due to the limitations of satellite payload characteristics, a satellite can only observe one target at a time, that is, each onboard remote sensor can only perform one observation task at any time;

[0137] Observation time window constraints: such as Figure 1 As shown, if the target tar r By satellitesat m To make an observation, the duration interval during which the satellite observes it must be a subset of the time window during which the satellite is visible to it;

[0138] For the visible window win i , if the target tar r By satellitesat m To conduct observation, the following conditions must be met:

[0139] ST mr ≥winST mr ,ET mr ≤winET mr

[0140] Observation transition time constraints: such as Figure 2 As shown in Figure 2, when a satellite performs two adjacent tasks, a certain transition time must be taken into account to ensure that the satellite attitude and imaging instrument working status are adjusted during this period. m For a task sequence, the time interval between the end of the previous task and the start of the next task should be greater than or equal to a transition time;

[0141] If the target tar i and tar j All of them need to occupy satellite sat m , and for the target tar i The observation task is followed by the target tar j After the observation task is executed, it must meet the following requirements:

[0142] ST mj -ET mi ≥B i,j

[0143] Among them B i,j Indicates satellite m Observation target tar i and tar j Required observation transition time;

[0144] Patrol time interval constraint: When performing a single-satellite patrol, it is necessary to ensure that the observation arcs are as dispersed as possible, and the time interval between two observations of the same target must be greater than one time period:

[0145]

[0146] tar i =tar j ,tar k ∈Win k,k=i,j

[0147]

[0148] Where, MinObserveInterval is the interval between two adjacent time windows for observing the same target;

[0149] Step 1.3: Mathematical description of the optimization target of the space-based space target patrol observation mission:

[0150] The optimization objectives of the space-based space target patrol observation mission are divided into primary planning objectives and secondary planning objectives:

[0151] First-level planning goal: the maximum number of target patrols within the patrol cycle, that is, to ensure that all targets are observed once;

[0152] For the set of targets to be patrolled, Tar = {tar1, tar2, ..., tar i ,…,tar r}, the number of targets is r, and the observed state is V = {v1,v2,...,v i ,…,v r}, where v i = 0 or 1, 0 means not observed, 1 means observed, then the first-level planning target can be expressed as:

[0153]

[0154] Secondary planning goal: the variance of the number of patrol targets of each satellite is minimized, that is, the workload of each satellite is as balanced as possible;

[0155] Assume that the target number set of each star patrol is Amount = {a1, a2, ..., a i ,...,a r},a i ≥0, a i represents the target number of the i-th satellite patrol, then the secondary planning target can be expressed as:

[0156]

[0157] Step 2: Slice the observation time window;

[0158] The visible time window can be expressed as Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r >, whose set is represented by Wins={Win1,Win2,...,Win i ,...,Winn}, slice the observation time window to form a sliced ​​visible window set

[0159] The observation window slicing method is adopted to consider the overlap of time windows. After slicing, the time windows are not connected at the end, but the start time of adjacent time windows is overlapped and divided according to the set difference. In this way, more windows can be generated, which is conducive to subsequent planning.

[0160] The schematic diagram and flow chart of segmenting the visible window are shown in Figure 3 and Figure 4 , the segmentation method is as follows:

[0161] Set the window split time split_time to the required observation time of the satellite observation target, that is, for the split time window Win i , split_time = |winET i -winST i |; The interval_time difference between the start and end time of the split window is the satellite sat after splitting m tar the target r The difference between the start times of two adjacent time windows of observation, that is: |winST i -winST j |=interval_time×n(n=1,2,…,m);

[0162] Determine the duration of the time window winDT mr Is it greater than the split time split_time? If so, continue splitting; otherwise, split the next time window.

[0163] Calculate the number of time window divisions m:

[0164]

[0165] The time window is divided, and the parameter calculation method for each time window is as follows:

[0166] winST i =winST mr +interval_time×(i-1),(i=1,2,…,m)

[0167] winET i =winST i +split_time

[0168] winDT i =split_time

[0169] Win i = <winST i ,winET i ,winDT i ,sat m ,tar r >

[0170]

[0171] For the split time window set Sort by the order of the start time of the time window winST.

[0172] Step 3: According to the priority rule, the first round of solving is performed on the sliced ​​visible window set according to the time window conflict degree, so as to meet the first-level planning goal as much as possible: the maximum number of target patrols within the patrol cycle;

[0173] The above time window conflict degree calculation method is:

[0174] The conflicting window set of each window in list_Win is counted. If the load capacity constraint or observation transition time constraint is not met, it is judged as a conflict. The judgment rule is:

[0175] If Win i |sat=Win j |sat, then perform conflict judgment:

[0176] if winST i ≤winST j

[0177] if(winST j -winET i )<B i,j

[0178] then conflict i,j =1

[0179] if winST i >winST j

[0180] if(winST i -winET j )<B i,j

[0181] then conflict i,j =1

[0182] Among them, conflict i,j=1 indicates window Win i and Win j A conflict occurs, B i,j Represents the window Win i ,Win j The duration of the turntable maneuvering between

[0183] Win i The conflict set can be expressed as

[0184] The conflict degree of a time window is defined as the number of time windows among all time windows that have conflicting observation time and switching duration with this window, that is:

[0185]

[0186] In the formula, Indicates window win i and win j There is a conflict, Indicates that there is no conflict;

[0187] If the time window Win i The conflict is combined into The time window Win i The conflict degree is k.

[0188] The above priority rules are:

[0189] For the same target, the target with fewer windows has higher priority; for windows with the same conflict degree, the window with higher average observed brightness during the window duration has higher priority;

[0190] For windows between different targets, the target with fewer windows has a higher priority; for targets with the same number of windows, the target with a smaller degree of window conflict has a higher priority; for targets with the same number of windows and the same degree of conflict, the target with a higher average observed brightness has a higher priority.

[0191] Step 3 is as follows. For details, see Figure 5 :

[0192] Step 3.1: Calculate the conflict degree of each time window;

[0193] Step 3.2: Determine the priority of the time window according to the priority rule and select the time window with the highest priority win select ;

[0194] Step 3.3: Determine whether the planning is completed and update the time window sequence list_Win;

[0195] Step 3.3 According to the selected window winselect , determine whether the planning is completed and update the time window sequence list_Win. The specific steps are as follows:

[0196] Step 1: Count the number of observations of each target. If all targets have been observed, step 3 ends. Otherwise, execute step 2.

[0197] Step 2: Calculate win in the time window sequence list_Win select The conflict set And delete it in the time window sequence list_Win;

[0198] Step 3: Delete win in the time window sequence list_Win select Window observation target tar select All windows of

[0199] Step 4: If there is no window in the time window sequence list_Win, then step 3 ends, otherwise execute step 3.1.

[0200] The steps to select the time window with the highest priority are as follows:

[0201] Calculate the number of windows for each target, and obtain the minimum number of windows for the target by comparing the number of windows for each target. tar , set the number of windows to Min tar The target window is stored in the window collection list_Win1;

[0202] Compare the conflict degree Con of each window in the window set list_Win1 i , get the minimum conflict degree Min conflict , set the conflict degree to Min conflict The window is stored in the window collection list_Win2;

[0203] Compare the brightness of each window in the window set list_Win2 and select the window with the highest brightness win select , add it to the task planning result sequence list_win_chose.

[0204] Step 4: Fine-tune based on the first round of solution results. See the flowchart of fine-tuning. Figure 6 , to achieve the goal, we should observe as much as possible, as follows:

[0205] Step 4.1: Count the targets that were not observed in the first round of solution;

[0206] Step 4.2: Count the time window Win of each unobserved target iCon with list_win_chose i , and Win i The corresponding conflict window set Conflict i ;

[0207] Step 4.3: Prioritize the unobserved targets according to the observed target priority rules, and first fine-tune the target with the highest priority, Tar prio2 , for win select2 ∈Tar prio2 , adjust the win in list_win_chose select2 Conflicting target windows, make target Tar prio2 can be observed;

[0208] Step 4.4: Update the priorities of all remaining targets and obtain the new highest priority target Tar prio2 , execute step 4.3 again.

[0209] Step 4.3 The specific steps are as follows:

[0210] Traversing Tar prio2 The available time window is counted in the planning result sequence list_win_chose with win select2 The time window with conflicts is added to the conflict sequence list_win_conflict;

[0211] Traverse the conflict sequence list_win_conflict, for the window win i , if there is a time window win to observe the same target j ,win j Not in the result sequence list_win_chose and does not conflict with the time window in the result sequence, and win j With win select2 If there is no time conflict, win j With win select2 Add to the planning result sequence list_win_chose and remove win i .

[0212] Step 5: Make adjustments based on the results of step 4 to achieve a balanced workload for each satellite and meet the secondary planning goal as much as possible: the variance of the number of patrol targets for each satellite is minimized. The specific steps are as follows:

[0213] Step 5.1: Determine whether the satellite workload is balanced: Let the number of targets observed by the satellite be x i , calculate the average and standard deviation std respectively, and define: if xi >average+3×std, then satellite i is considered to have too much workload; if x i <average-3×std, then the workload of satellite i is considered to be too small;

[0214] Step 5.2: Count the satellites with too much workload and too little workload respectively, and record the satellite sequence with too much workload as list_sat over , the satellite sequence with too little workload is recorded as list_sat below ;

[0215] Step 5.3: Prioritize list_sat over The satellite with the most workload over Adjust, for win∈sat over , prioritize adjusting the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , try to make the number of satellite observation targets within the interval [average-3×std,average+3×std];

[0216] Step 5.4: Update satellite workload, count satellites with too little workload, and record the sequence of satellites with too little workload as list_sat below , the rest is list_sat other ;

[0217] Step 5.5: List_sat other Satellite with normal workload other Adjust, for win∈sat other , adjust the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , increase the number of observation windows of satellites with less workload, and try to make the number of satellite observation targets within the interval of [average-3×std,average+3×std].

[0218] simulation:

[0219] In order to verify the above method, the present invention uses 6 satellites to simulate patrols of 60 space targets, and the simulation scenario duration is 1 hour. The observation duration of each target, i.e., the window split duration split_time, is set to 200s. m tar the target rThe interval_time difference between the start times of two adjacent time windows of observation is 30s. A total of 250 time windows of the constellation to the target are calculated, and the observation time windows are sliced ​​according to the method in step 2 to obtain 900 sliced ​​time windows.

[0220] After solving the first round of planning for 60 targets in a patrol scenario for 1 hour, the Gantt chart of the constellation patrol task planning results is as follows: Figure 7 As shown in the figure, the horizontal axis is the time axis, the vertical axis is the satellite number, and each rectangular block in the figure represents the observation task time window. The first round of planning completed the observation of 57 targets within 1 hour.

[0221] After solving the second round of planning for 60 targets in a patrol scenario for 1 hour, the Gantt chart of the constellation patrol task planning results is as follows: Figure 8 As shown in the figure, the second round of planning completed the observation of 58 targets within 1 hour. In the second round of solution, the observation window of satellite No. 4 for target No. 47 was advanced, so as to insert the time window for target No. 33, increasing the number of target observations of the constellation.

[0222] The second round of planning results within 1 hour show that the satellite's cumulative working time is as follows: Fig. 9 As shown in the figure, after the second round of planning, the average cumulative observation time of each satellite is 1933.34s, among which the satellites 1 and 5 have the longest cumulative working time, with cumulative observations of 2200s respectively, and the shortest working time is satellite 6, with cumulative observations of 1400s. The number of observation windows for the six satellites are 11, 10, 10, 11, 9, and 7 respectively.

[0223] After fine-tuning and solving the third round of planning for 60 targets in a patrol scenario for 1 hour, the Gantt chart of the constellation patrol mission planning results and the comparison of the accumulated star observation time are shown in the figure below. Fig.10 As shown in the figure, the comparison of the cumulative observation time of each satellite in the third round of planning within 1 hour is shown in Fig.11 ,It can be seen from the figure that after the third round of planning, the number of observation windows of the six satellites are 10, 10, 10, 10, 9, and 9 respectively, and the load of each satellite is significantly balanced.

Claims

1. A method for planning an on-orbit mission for patrol observation of space-based space targets based on time window conflict degree, comprising the following steps: Step 1: Construct a mathematical model for the space-based space target patrol observation mission planning problem and determine the optimization target of the space-based space target patrol observation mission; Step 2: Slice the observation time window; The visible time window can be expressed as Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r >, whose set is represented by Wins={Win1,Win2,…,Wini,…,Win n }, slice the observation time window to form a sliced ​​visible window set Step 3: According to the priority rule, the first round of solving is performed on the sliced ​​visible window set according to the time window conflict degree, so as to meet the first-level planning goal as much as possible: the maximum number of target patrols within the patrol cycle; Step 4: Make fine adjustments based on the results of the first round of solutions to achieve the goal of observing as much as possible; Step 5: Make adjustments based on the results of step 4 to achieve a balanced workload for each satellite and meet the secondary planning goal as much as possible: the variance of the number of patrol targets for each satellite is minimized.

2. According to the method of claim 1, the method is characterized in that: The time window conflict degree calculation method is: The conflicting window set of each window in list_Win is counted. If the load capacity constraint or observation transition time constraint is not met, it is judged as a conflict. The judgment rule is: Win j ∈list_Win, if Win i |sat=Win j |sat, then perform conflict judgment: if winST i ≤winST j if(winST j -winET i )<B i,j then conflict i,j =1 if winST i >winST j if(winST i -winET j )<B i,j then conflict i,j =1 Among them, conflict i,j =1 indicates window Win i and Win j A conflict occurs, B i,j Represents the window Win i ,Win j The duration of the turntable maneuvering between Win i The conflict set can be expressed as The conflict degree of a time window is defined as the number of time windows among all time windows that have conflicting observation time and switching duration with this window, that is: In the formula, Indicates window win i and win j There is a conflict, Indicates that there is no conflict; If the time window Win i The conflict is combined into The time window Win i The conflict degree is k.

3. According to the method of claim 1, the method is characterized in that: The precedence rules are: For the same target, the target with fewer windows has higher priority; for windows with the same conflict degree, the window with higher average observed brightness during the window duration has higher priority; For windows between different targets, the target with fewer windows has a higher priority; for targets with the same number of windows, the target with a smaller degree of window conflict has a higher priority; for targets with the same number of windows and the same degree of conflict, the target with a higher average observed brightness has a higher priority.

4. According to the method of claim 1, the method is characterized in that: Step 1 specifically includes: Step 1.1: Description of the problem of space-based space target patrol mission planning; The problem of space-based space target patrol mission planning is described as N satellites observing r targets in a planning cycle, so that the optimization target of space-based space target patrol mission is optimal. The final output of mission planning is mainly the allocation plan of observation tasks. The single-star patrol mission planning problem is expressed as: SPMP={Sat,Tar,Wins,RES} in: Sat——satellite set, there are N satellites in total, Sat={sat1,sat2,...,sat N }; Tar——the set of observed targets, there are r targets to be observed, Tar={tar1,tar2,...,tar i ,...,tar r }; Wins is the set of visible windows of the satellite Sat observation target Tar, n represents the total number of time windows to be planned, Wins = {Win1, Win2, ..., Win i ,…,Win n }; Win means satellite sat m Observation target tar r Visible window, Win=<winST mr ,winET mr ,winDT mr ,sat m ,tar r >winST mr 、winET mr Indicates satellite m Observation target tar r The start time and end time of the visible window; winDT mr , satellite m Observation target tar r The duration of the visible window, winDT mr =winET mr -winST mr ; RES – Constraint Set; Step 1.2: Mathematical description of the space-based space target patrol observation mission constraints; The constraints are: load capacity constraint, observation time window constraint, observation transition time constraint and patrol time interval constraint; Payload capacity constraints: Due to the limitations of satellite payload characteristics, a satellite can only observe one target at a time, that is, each onboard remote sensor can only perform one observation task at any time; Observation time window constraint: If the target tar r By satellite m To make an observation, the duration interval during which the satellite observes it must be a subset of the time window during which the satellite is visible to it; For the visible window win i , if the target tar r By satellite m To conduct observation, the following conditions must be met: ST mr ≥winST mr ,AND mr ≤winET mr Observation transition time constraints: For satellite sat m For a task sequence, the time interval between the end of the previous task and the start of the next task should be greater than or equal to a transition time; tar j ∈Tar, if the target tar i and tar j All of them need to occupy satellite sat m , and for the target tar i The observation task is followed by the target tar j After the observation task is executed, it must meet the following requirements: ST mj -AND mi ≥B i,j Among them B i,j Indicates satellite m Observation target tar i and tar j Required observation transition time; Patrol time interval constraint: When performing a single-satellite patrol, it is necessary to ensure that the observation arcs are as dispersed as possible, and the time interval between two observations of the same target must be greater than one time period: takes i =takes j ,takes k ∈Win k ,k=i,j Where, MinObserveInterval is the interval between two adjacent time windows for observing the same target; Step 1.3: Mathematical description of the optimization target of the space-based space target patrol observation mission: The optimization objectives of the space-based space target patrol observation mission are divided into primary planning objectives and secondary planning objectives: First-level planning goal: the maximum number of target patrols within the patrol cycle, that is, to ensure that all targets are observed once; For the target set to be patrolled, Tar = {tar1, tar2, ..., tar i ,...,tar r }, the number of targets is r, and the observed state is V = {v1, v2, …, v i ,...,v r }, where v i = 0 or 1, 0 means not observed, 1 means observed, then the first-level planning target can be expressed as: Secondary planning goal: Minimize the variance of the number of patrol targets for each star; Assume that the target number set of each star patrol is Amount = {a1, a2, ..., a i ,...,a r },a i ≥0, a i represents the target number of the i-th satellite patrol, then the secondary planning target can be expressed as:

5. According to the method of claim 1, the method is characterized in that: Step 2: The method for segmenting the visible window is as follows: Set the window split time split_time to the required observation time of the satellite observation target, that is, for the split time window Win i , split_time = |winET i -winST i |; The interval_time difference between the start and end time of the split window is the satellite sat after splitting m tar the target r The difference between the start times of two adjacent time windows of observation, that is: Win j ∈Win, winST i -winST j |=interval_time×n(n=1,2,…,m); Determine the duration of the time window winDT mr Is it greater than the split time split_time? If so, continue splitting; otherwise, split the next time window. Calculate the number of time window divisions m: The time window is divided, and the parameter calculation method for each time window is as follows: gain i =winST mr +interval_time×(i-1),(i=1,2,…,m) winET i =winST i +split_time winDT i =split_time Win i = <winST i ,winET i ,winDT i ,sat m ,tar r > For the split time window set Sort by the order of the start time of the time window winST.

6. The method for planning on-orbit missions for patrol observation of space-based space targets based on time window conflict degree according to claim 1 is characterized in that: Step 3 specifically includes: Step 3.1: Calculate the conflict degree of each time window; Step 3.2: Determine the priority of the time window according to the priority rule and select the time window with the highest priority win select ; Step 3.3: Determine whether the planning is completed and update the time window sequence list_Win.

7. The method for planning on-orbit missions for patrol observation of space-based space targets based on time window conflict degree according to claim 6 is characterized in that: The steps to select the time window with the highest priority are as follows: Calculate the number of windows for each target, and obtain the minimum number of windows for the target by comparing the number of windows for each target. tar , set the number of windows to Min tar The target window is stored in the window collection list_Win1; Compare the conflict degree Con of each window in the window set list_Win1 i , get the minimum conflict degree Min conflict , set the conflict degree to Min conflict The window is stored in the window collection list_Win2; Compare the brightness of each window in the window set list_Win2 and select the window with the highest brightness win select , add it to the task planning result sequence list_win_chose.

8. The method for planning on-orbit missions for patrol observation of space-based space targets based on time window conflict degree according to claim 1 is characterized in that: Step 4 is as follows: Step 4.1: Count the targets that were not observed in the first round of solution; Step 4.2: Count the time window Win of each unobserved target i Con with list_win_chose i , and Win i The corresponding conflict window set Conflict i ; Step 4.3: Prioritize the unobserved targets according to the observed target priority rules, and first fine-tune the target with the highest priority, Tar prio2 , for win select2 ∈Tar prio2 , adjust the win in list_win_chose select2 Conflicting target windows, make target Tar prio2 can be observed; Step 4.4: Update the priorities of all remaining targets and obtain the new highest priority target Tar prio2 , perform step 4.3 again.

9. The method for planning on-orbit missions for patrol observation of space-based space targets based on time window conflict degree according to claim 1, characterized in that: Step 5 is to make adjustments based on the results of step 4 to achieve a balanced workload for each satellite. The specific steps are as follows: Step 5.1: Determine whether the satellite workload is balanced: Let the number of targets observed by the satellite be x i , calculate the average and standard deviation std respectively, and define: if x i >average+3×std, then satellite i is considered to have too much workload; if x i <average-3×std, then the workload of satellite i is considered to be too small; Step 5.2: Count the satellites with too much workload and too little workload respectively, and record the satellite sequence with too much workload as list_sat over , the satellite sequence with too little workload is recorded as list_sat below ; Step 5.3: Prioritize list_sat over The satellite with the most workload over Adjust, for win∈sat over , prioritize adjusting the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , try to make the number of satellite observation targets within the interval [average-3×std,average+3×std]; Step 5.4: Update satellite workload, count satellites with too little workload, and record the sequence of satellites with too little workload as list_sat below , the rest is list_sat other ; Step 5.5: List_sat other Satellite with normal workload other Adjust, for win∈sat other , adjust the observation satellite corresponding to the target of win to list_sat below The satellite with the least workload below , increase the number of observation windows of satellites with less workload, and try to make the number of satellite observation targets within the interval of [average-3×std,average+3×std].

Images