Space target perception scheduling method based on genetic algorithm

Abstract

The invention discloses a space target perception scheduling method based on a genetic algorithm, and the method comprises the following steps: S1, creating a space target perception scheduling model which comprises a perception system model, a target model and a task model, then defining a task window model, and carrying out the modeling of a target; and S2, a genetic algorithm of space target perception scheduling is adopted, and optimal perception system startup time planning and task scheduling are searched by adopting the genetic algorithm. According to the method, the startup time of each set of system is optimized, the working time of equipment can be shortened under the condition of ensuring that tasks are completed, the working time of each system is balanced, meanwhile, the number of the target tracking arc segments can be set, the number of the tasks is reduced, and generation of excessive redundant data is avoided.

Description

Technical Field

[0001] The present invention relates to a perception scheduling method, and in particular to a space target perception scheduling method based on a genetic algorithm. Background Art

[0002] There are tens of thousands of space objects such as satellites, debris, and space debris distributed in the outer space of the earth. According to the mission requirements, the orbits of all space objects are monitored in real time, so as to predict the release of space object collision warnings, and evaluate the feasibility of orbital maneuvers for satellites and other objects.

[0003] This requires obtaining the orbital data of these space targets, but the orbital data of space targets has errors and can only be predicted for 3-7 days. In order to be able to monitor space targets uninterruptedly, several sets of space target perception systems are required to track and measure the positions of all space targets and obtain the observation data of space targets. For the same target, at least three sections of high-quality observation data per day are sufficient for orbit calculation. However, the current number of space targets exceeds 30,000, and the amount of observation data required is very large. This requires the space target perception system to be turned on for a long time every day to ensure that enough observation data is obtained to complete the orbit determination task of all space targets. The ground-based space target perception system is expensive, and long-term uninterrupted shutdown will accelerate equipment aging and consume a lot of energy and manpower.

[0004] There are currently several space target perception systems, and they are distributed far from each other. In order to achieve the observation of all space targets, the measure taken is to increase the working hours of all equipment to no less than 15 hours a day.

[0005] Therefore, there are the following disadvantages:

[0006] (1) The equipment startup time is too long, which leads to high energy consumption, high load, frequent equipment failures, and consumption of more human resources.

[0007] (2) The startup time of some equipment is not balanced enough, and some equipment is started for more than 18 hours, which is not conducive to equipment maintenance and personnel rotation.

[0008] (3) Some space targets have too much observation data, resulting in excessive redundancy. Under normal circumstances, orbit determination can be achieved with three arc segment observation data for a space target. Excessive data not only increases the amount of calculation, but also makes storage inconvenient. Summary of the invention

[0009] In order to solve the problems in the above-mentioned background technology, the present invention provides a space target perception scheduling method based on genetic algorithm. The task demander issues a perception task according to the demand, and the perception system completes the space target perception task within the implementation range specified by the task according to the task demand.

[0010] The present invention provides the following technical solutions:

[0011] The present invention focuses on planning and scheduling the working time of the space target perception system and the space target perception tasks, obtaining the observation data of the space target and realizing the orbit determination of the target. It mainly achieves the following three goals.

[0012] (1) Maximize the mission completion rate of space targets.

[0013] (2) Minimize the operating time of the space target perception system to increase the service life of the equipment and reduce energy consumption.

[0014] (3) Balance the working hours of each system.

[0015] The space target sensing scheduling method based on genetic algorithm includes the following steps:

[0016] S1. Create a spatial target perception scheduling model, including the perception system model, target model and task model, then define the task window model, and then model the target;

[0017] S2. Use the genetic algorithm of space target perception scheduling to find the optimal perception system startup time planning and task scheduling, which specifically includes the following steps: encoding and decoding; solving the objective function; GA algorithm usage steps and parameter settings.

[0018] S11. Perception system model

[0019] Define a single perception system model as

[0020] resource={rID,rMaxNum,rPos,rAngel,rPara,rWorkTime} (1)

[0021] The elements of the perception system are device ID, maximum number of perceived targets, location, direction, range, and working time period. The total number of perception systems is defined as N1.

[0022] S12. Perception target model and task model

[0023] Define the spatial target model as

[0024] target={tID,tName,tTLE,tPr,tSize} (2)

[0025] The elements are number, name, track information, priority, and size.

[0026] According to the target situation, the demand side can generate perception tasks according to actual needs, and define the perception task model as

[0027] task={taID,tID,taPr,tLastTime,tMin,taQuan,taInterval} (3)

[0028] The elements are task number, target ID, task priority, minimum duration of a single task, time limit for completing the task, number of tasks, and shortest time interval for perceiving tasks. The number of tasks is N2.

[0029] S13. Define the task window model

[0030] According to the knowledge of orbital mechanics, the window information of space targets relative to a certain perception system can be obtained, and the task window model is defined as follows

[0031] win={wID,taID,dID,wSt,wEt,wLast,vStartAER,vEndAER} (4)

[0032] The elements of each column are windID, task ID, perception system ID, perception start time, perception end time, and duration, and the number of wins is N3.

[0033] S14. Target Modeling

[0034] By building a model, the spatial target scheduling problem is converted into allocating task windows to the perception system and completing the requirements in the task model. Its objective function is:

[0035]

[0036] Among them, taPr i is the priority of the ith task, s i The effective number of actual scheduling;

[0037] For the supplier, the cost is reduced and the cost is balanced. The objective function is

[0038]

[0039] J3=min(max(T i )-min(T i )) (7)

[0040] Where T i is the total working time of the i-th perception system. Considering the needs of both parties, the overall objective function is

[0041]

[0042] When all perception systems have no restrictions on the number of perceived targets and working time, the theoretical total task requirement is

[0043]

[0044] Among them, taQuan i is the number of perceptions required for the corresponding task. For the objective function, it is necessary to meet a certain number of task requirements before ensuring that J1 / J2 is optimal. Therefore, the final objective function is

[0045]

[0046] in is a parameter α and the independent variable is The piecewise function of

[0047]

[0048] Consider the number of perceptions N of each perception system in any time period win The tracking capability limit cannot be exceeded, and the constraint equation is:

[0049] N win ≤rmaxNum j (12)

[0051] S21, encoding and decoding

[0052] Taking the first 8-hour period

[0480] minutes of the first shift as an example, the startup time is l i You can freely choose the time from 1 to 480 minutes, and the start time is start i Only 0 to 480-l i A simple method is to divide a shift into 512 equal parts, and use a 9-bit binary code to represent the start-up time of this section. At the same time, [0480-1 i ] is divided into 512 equal parts, and a 9-bit binary code is used to represent the start-up time of the shift. In this way, an 18-bit binary code can be used to represent the start-up duration and start-up time of a shift. If there are N1 sets of sensing systems, the length of the constructed chromosome is 54N1.

[0053] According to the encoding rules, the decoding formula is as follows

[0054]

[0055] Among them l i is the startup time of the i-th segment, start i is the startup time of the i-th segment, l ci The duration code corresponding to the time period is converted into a decimal value, p ci The decimal value corresponding to the corresponding position code.

[0056] S22. Solution of objective function

[0057] The specific steps are as follows:

[0058] Step 1: First, take the minimum task duration as the time slot, discretize the task time period, and calculate the window occupancy time slot;

[0059] Step 2: Sort the time slots occupied by windows from small to large;

[0060] Step 3: First process the time slots without conflicts, count the task completion status, and clear the windows of completed tasks in time;

[0061] Step 4: When conflicts occur, start with small conflicts and give priority to windows with a larger priority multiplied by the number of remaining tasks;

[0062] Step 5: When the number of remaining windows of all conflicting time slots drops to 0, the processing ends.

[0063] S23, GA algorithm usage steps and parameter settings

[0064] The specific process is as follows:

[0065] Step 1: Set the population size and the upper limit of the number of generations, and randomly generate the initial population. The length of the population is determined according to the encoding process;

[0066] Step 2: Decode the chromosome to obtain the working time period information of the perception system;

[0067] Step 3: Calculate the objective function according to the steps in Section 2.2 and calculate the fitness of each chromosome;

[0068] Step 4: Select the operator using random selection with a ratio of 0.9;

[0069] Step 5: Reorganization operator, the reorganization ratio is 0.7;

[0070] step6: mutation operator, the ratio is 0.3;

[0071] Step 7: After selection, recombination and mutation, the offspring are subjected to step 2 and step 3 operations to calculate the offspring objective function and fitness;

[0072] Step 8: Judge the objective function and fitness. If they meet the design requirements, go to step 11. If not, go to step 9.

[0073] Step 9: Update the population, generate a new population from the parent population and the child population according to the objective function values ​​and fitness of the two generations;

[0074] Step 10: Determine whether the number of offspring in the population exceeds the upper limit of the number of generations. If it exceeds the set value, go to step 11. If it does not exceed the set value, increase the number of offspring by 1 and go to step 2.

[0075] Step 11: Analyze the optimal value and average value of each generation of the population, and select the appropriate solution. If the conditions are not met, adjust the population size or the upper limit of the number of generations and recalculate.

[0076] Compared with the prior art, the present invention has the following beneficial effects:

[0077] 1. Creation of a space target perception and scheduling model, including perception system model, task model and target model.

[0078] 2. In terms of the selection of task optimization methods, a genetic algorithm is used to find the optimal perception system startup time planning and task scheduling.

[0079] 3. The innovation of space target perception task coding has greatly shortened the coding length and has weaker task conflict compared with previous studies, so it takes less time and is more efficient in planning.

[0080] The present invention optimizes the startup time of each system, shortens the working time of the equipment while ensuring the completion of the task, and balances the working time of each system. At the same time, the number of target tracking arcs can be set to reduce the number of tasks and avoid the generation of excessive redundant data. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 This is a corresponding diagram between the working time length and the binary code of the present invention.

[0082] Figure 2 This is a corresponding diagram between the working start-up time and the binary code of the present invention.

[0083] Figure 3 This is the flow chart of the GA algorithm of the present invention.

[0084] Figure 4 This is the objective function value diagram of the present invention.

[0085] Figure 5 This is a task completion diagram of the present invention.

[0086] Figure 6 This is a graph of the total time consumed by the perception resources of the present invention.

[0087] Figure 7 This is a graph showing the difference in working time of the sensing resources of the present invention.

[0088] Figure 8 This is a comparison chart of normalized results of each generation of the present invention. DETAILED DESCRIPTION

[0090] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0091] See also Figure 1-3 The present invention provides a space target sensing scheduling method based on genetic algorithm, comprising the following steps:

[0092] S1. Create a spatial target perception scheduling model

[0093] S11. Perception system model

[0094] The perception system model needs to know the maximum perception quantity, position, direction, azimuth width, pitch width, range, working time and other information of the perception system. Define a single perception system model as

[0095] resource={rID,rMaxNum,rPos,rAngel,rPara,rWorkTime} (1)

[0096] The elements of the perception system are device ID, maximum number of perceived targets, position, direction, range (horizontal width, pitch width, range), and working time period. The total number of perception systems is defined as N1.

[0097] S12. Perception target model and task model

[0098] For space targets, we need to know the target number, orbit information, priority, size, orbit information, etc. The orbit information needs to be continuously updated and maintained. The space target model is defined as

[0099] target={tID,tName,tTLE,tPr,tSize} (2)

[0100] The elements are number, name, track information, priority, and size.

[0101] According to the target situation, the demand side can generate perception tasks according to actual needs. The perception task model is defined as

[0102] task={taID,tID,taPr,tLastTime,tMin,taQuan,taInterval} (3)

[0103] The elements are task number, target ID, task priority, shortest duration of a single task, time limit for completing the task, number of tasks, and shortest time interval for sensing tasks. The number of tasks is N2.

[0104] S13. Define the task window model

[0105] According to the knowledge of orbital mechanics, we can obtain the window information of space targets relative to a certain perception system. The mission window model is defined as follows:

[0106] win={wID,taID,dID,wSt,wEt,wLast,vStartAER,vEndAER} (4)

[0107] The elements of each column are windID, task ID, perception system ID, perception start time, perception end time, and duration. The number of wins is N3.

[0108] S14. Target Modeling

[0109] By building a model, the spatial target scheduling problem is converted into allocating task windows to the perception system and completing the requirements in the task model. For the demand side, it is required to complete as many perception tasks as possible, and its objective function is

[0110]

[0111] Among them, taRr i is the priority of the ith task, s i This is the effective number of actual scheduling.

[0112] For the supplier, the cost should be reduced as much as possible and the cost should be balanced. The objective function is:

[0113]

[0114] J3=min(max(T i )-min(T i )) (7)

[0115] Where T i is the total working time of the i-th perception system. Considering the needs of both parties, the overall objective function is

[0116]

[0117] Considering J1 / J2 as the optimization target, the scheduling plan will fall into the time period corresponding to the largest group, and there will be no perception tasks in other time periods. Obviously, this is not the best solution. In order to avoid this situation, a constraint condition needs to be added, that is, the scheduling plan must reach a certain task completion rate α to be qualified. When all perception systems have no restrictions on the number of perception targets and working hours, the theoretical total task demand is

[0118]

[0119] Among them, taQuan i is the number of perceptions required for the corresponding task. For the objective function, it is necessary to meet a certain number of task requirements before ensuring the optimal J1 / J2. Therefore, the final objective function is

[0120]

[0121] in is a parameter α and the independent variable is The piecewise function of .

[0122]

[0123] Consider the number of perceptions N of each perception system in any time period win The tracking capability limit cannot be exceeded, and the constraint equation is:

[0124] N win ≤rmaxNum j (12)

[0125] S2. Genetic Algorithm for Space Target Perception Scheduling

[0126] S21, encoding and decoding

[0127] Considering that a sensing cycle is generally one day, the number of times a device is turned on and off should not be set too much. According to actual conditions, people work 8 hours a day, there are three shifts a day, and the work is carried out in a three-shift mode. Therefore, a day can be divided into three shifts, and each shift has only one on / off time. Taking the first 8-hour time period of the first shift [0 480] minutes as an example, the on-time l i You can freely choose the time from 1 to 480 minutes, and the start time is start i Only 0 to 480-l i A simple method is to divide a shift into 512 equal parts and use a 9-bit binary code to represent the duration of the power-on period. i] is divided into 512 equal parts, and a 9-bit binary code is used to represent the start-up time of the shift. In this way, an 18-bit binary code can be used to represent the start-up duration and start-up time of a shift. If there are N1 sets of sensing systems, the length of the constructed chromosome is 54N1. Figure 1 and Figure 2 shown.

[0128] Figure 1 It is the corresponding diagram of working time length and binary code, and we can get:

[0129] Figure 2 This is the corresponding diagram between the working startup time and the binary code, and we can get:

[0130]

[0131] In the case of three shifts with fixed time, since the startup duration and startup time are randomly determined by random individuals of the chromosome, the GA method can always find the optimal scheduling solution with enough individuals and enough generations.

[0132] According to the encoding rules, the decoding formula is as follows

[0133]

[0134]

[0135] Among them l i is the startup time of the i-th segment, start i is the startup time of the i-th segment. ci The duration code corresponding to the time period is converted into a decimal value. ci The decimal value corresponding to the corresponding position code.

[0136] S22. Solution of objective function

[0137] According to the boot time and boot time obtained from the binary code, the objective function is calculated by considering the constraints. The constraints of formula (12) need to be considered when calculating the objective function. Considering that the upper limit of the number of perceived targets of the current perception system is relatively high, the number of perceived spatial ranges will not exceed the design index in most periods of time. Therefore, this constraint is a weak conflict constraint, and the conflict can be handled according to the heuristic method. The specific steps are as follows:

[0138] Step 1: First, take the minimum task duration as the time slot, discretize the task time period, and calculate the window occupancy time slot;

[0139] Step 2: Sort the time slots occupied by windows from small to large;

[0140] Step 3: First process the time slots without conflicts, count the task completion status, and clear the windows of completed tasks in time;

[0141] Step 4: When conflicts occur, start with small conflicts and give priority to windows with a larger priority multiplied by the number of remaining tasks;

[0142] Step 5: When the number of remaining windows of all conflicting time slots drops to 0, the processing ends.

[0143] S23, GA algorithm usage steps and parameter settings

[0144] The spatial scheduling optimization problem is transformed into a scheduling solution that optimizes the objective function. Each perception system needs to be encoded, and then the GA algorithm is used to generate a population, and then the GA cycle is entered to perform decoding, selection, recombination, mutation, and insertion operations to find the optimal solution. The specific process is as follows Figure 3 shown.

[0145] The specific process is as follows:

[0146] Step 1: Set the population size and the upper limit of the number of generations, and randomly generate the initial population. The length of the population is determined according to the encoding process;

[0147] Step 2: Decode the chromosome to obtain the working time period information of the perception system;

[0148] Step 3: Calculate the objective function according to the steps in Section 2.2 and calculate the fitness of each chromosome;

[0149] Step 4: Select the operator using random selection with a ratio of 0.9;

[0150] Step 5: Reorganization operator, the reorganization ratio is 0.7;

[0151] step6: mutation operator, the ratio is 0.3;

[0152] Step 7: After selection, recombination and mutation, the offspring are subjected to step 2 and step 3 operations to calculate the offspring objective function and fitness;

[0153] Step 8: Judge the objective function and fitness. If they meet the design requirements, go to step 11. If not, go to step 9.

[0154] Step 9: Update the population, generate a new population from the parent population and the child population according to the objective function values ​​and fitness of the two generations;

[0155] Step 10: Determine whether the number of offspring in the population exceeds the upper limit of the number of generations. If it exceeds the set value, go to step 11. If it does not exceed the set value, increase the number of offspring by 1 and go to step 2.

[0156] Step 11: Analyze the optimal value and average value of each generation of the population, and select the appropriate solution. If the conditions are not met, adjust the population size or the upper limit of the number of generations and recalculate.

[0157] Simulation and analysis:

[0158] 1. Simulation environment and data generation

[0159] Python is used for simulation, and 5 points are selected from the map as the sites of the perception system. The specific parameters are shown in Table 1.

[0160] Table 1. Perception system simulation parameters

[0161]

[0162]

[0163] Download the tle file from a website, first remove the targets with altitudes above 2000 km and below 100 km, then select the first 8000 low-orbit targets for simulation. Use a uniform random method to generate the target priority, which is an integer from 1 to 10.

[0164] According to the target set, the simulation generates a task set. The task priority is the same as the target priority. The minimum tracking time of all target single tasks is 0.5 minutes, and the task completion time is one day, that is,

[01440] . The number of tasks for each target is 3. The details are shown in Table 2.

[0165] Table 2 Task simulation parameter settings

[0166]

[0167] According to the situation of the perception system and the task, the SGP4 algorithm is used to consider the constraints of the perception system and calculate all the windows that meet Table 1 and Table 2. After calculation, the number of windows that can be detected by the perception system and meet the task requirements is 171945.

[0168] 2.GA simulation:

[0169] Set the population size to 480, the number of generations to 100, randomly select individuals, the ratio is 0.9, the recombination ratio is 0.7, and the mutation ratio is 0.3. Simulate. Since the algorithm is not optimized during simulation, the simulation time is about 11 hours. Select the individual with the highest objective function value from each generation for plotting, such as Figure 4 , 5 , 6, and 7.

[0170] Figure 4 , 5 The horizontal coordinates of the four sub-graphs in , 6, and 7 are all the generations of the population, and the vertical coordinates are the objective function value, task completion, the total working time of the five devices, and the difference in the working time of the perception system. From the figure, it can be seen that the task completion has stabilized at 0.99 in the fourth generation. The total time consumed by the perception system fluctuates and decreases between 3000 and 2600. At the same time, the indicator reflecting the balance of the working time of the perception system continues to decrease with the increase of generations. The objective function value continues to increase with the increase of generations, and there is a sudden increase when it approaches 100 generations. The indicator of the perception system balance contributes the most to the sudden increase in the objective function value in the last few generations. Because this indicator is close to 0 at the end, the objective function value suddenly rises and increases.

[0171] Normalize the vertical coordinates of the four sub-graphs to the

[01] interval, and we get Figure 8 .from Figure 8 As can be seen from the figure, except for the slightly lower values ​​(0.96) of the 47th and 49th generations, the task completion degree is relatively high in other positions. Therefore, the scheme selection can be carried out according to the total time curve, and several scheduling schemes in the valley can be selected for decision-making by the decision-making agency. Table 3 shows the three optional schemes provided for this scheduling.

[0172] Table 3 Optional scheduling schemes

[0173]

[0174] in conclusion:

[0175] In the use of genetic algorithms to solve the problem of space target perception scheduling, modeling analysis can be carried out according to the actual situation of personnel and equipment, specific and feasible optimization goals can be proposed, and scheduling can be carried out for massive tasks. Simulation experiments show that the use of improved genetic algorithms can minimize the perception system overhead and balancing overhead while ensuring the requirements of the demand side, and can alleviate the problems of personnel shortage and equipment life reduction in engineering applications. This paper encodes by finding a time period in a specific time period when processing encoding, which is more in line with actual work. However, the optimal scheduling plan may also appear in the case of cross-time periods, and the time periods are not limited to 3. Since this situation is not quite the same as the actual work, and it is difficult to achieve the one-to-one correspondence between encoding and decoding, it is not considered to be implemented. The subsequent work of the paper studies the encoding of any number of time periods to achieve a better scheduling plan.

[0176] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A space target sensing scheduling method based on genetic algorithm, characterized by: The following steps are involved: S1. Create a spatial target perception scheduling model, including the perception system model, target model and task model, then define the task window model, and then model the target; S2, using a genetic algorithm for spatial target perception scheduling, using a genetic algorithm to find the optimal perception system startup time planning and task scheduling, specifically including the steps of: encoding and decoding; solving the objective function; GA algorithm usage steps and parameter settings.

2. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S11. Perception system model Define a single perception system model as resource={rID,rMaxNum,rPos,rAngel,rPara,rWorkTime} (1) The elements of the perception system are device ID, maximum number of perceived targets, location, direction, range, and working time period. The total number of perception systems is defined as N1.

3. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S12. Perception target model and task model Define the spatial target model as target={tID,tName,tTLE,tPr,tSize} (2) The elements are number, name, track information, priority, and size. According to the target situation, the demand side can generate perception tasks according to actual needs, and define the perception task model as task={taID,tID,taPr,tLastTime,tMin,taQuan,taInterval} (3) The elements are task number, target ID, task priority, minimum duration of a single task, time limit for completing the task, number of tasks, and shortest time interval for perceiving tasks. The number of tasks is N2.

4. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S13. Define the task window model According to the knowledge of orbital mechanics, the window information of space targets relative to a certain perception system can be obtained, and the task window model is defined as follows win={wID,taID,dID,wSt,wEt,wLast,vStartAER,vEndAER} (4) The elements of each column are windID, task ID, perception system ID, perception start time, perception end time, and duration, and the number of wins is N3.

5. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S14. Target Modeling By building a model, the spatial target scheduling problem is converted into allocating task windows to the perception system and completing the requirements in the task model. Its objective function is: Among them, taPr i is the priority of the ith task, s i The effective number of actual scheduling; For the supplier, the cost is reduced and the cost is balanced. The objective function is J3=min(max(T i )-min(T i )) (7) Where T i is the total working time of the i-th perception system. Considering the needs of both parties, the overall objective function is When all perception systems have no restrictions on the number of perceived targets and working time, the theoretical total task requirement is Among them, taQuan i is the number of perceptions required for the corresponding task. For the objective function, it is necessary to meet a certain number of task requirements before ensuring that J1 / J2 is optimal. Therefore, the final objective function is in is a parameter α and the independent variable is The piecewise function of Consider the number of perceptions N of each perception system in any time period win The tracking capability limit cannot be exceeded, and the constraint equation is: N win ≤rmaxNum j (12)。 6. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S21, encoding and decoding Taking the first 8-hour period [0 480] minutes of the first shift as an example, the startup time is l i You can freely choose the time from 1 to 480 minutes, and the start time is start i Only 0 to 480-l i A simple method is to divide a shift into 512 equal parts, and use a 9-bit binary code to represent the start-up time of this section. At the same time, [0 480-1 i ] is divided into 512 equal parts, and a 9-bit binary code is used to represent the start-up time of the shift. Thus, an 18-bit binary code can be used to represent the start-up duration and start-up time of a shift. If there are N1 sets of sensing systems, the length of the constructed chromosome is 54N1. According to the encoding rules, the decoding formula is as follows Among them l i is the startup time of the i-th segment, start i is the startup time of the i-th segment, l ci The duration code corresponding to the time period is converted into a decimal value, p ci The decimal value corresponding to the corresponding position code.

7. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S22. Solution of objective function The specific steps are as follows: Step 1: First, take the minimum task duration as the time slot, discretize the task time period, and calculate the window occupancy time slot; Step 2: Sort the time slots occupied by windows from small to large; Step 3: First process the time slots without conflicts, count the task completion status, and clear the windows of completed tasks in time; Step 4: When conflicts occur, start with small conflicts and give priority to windows with a larger priority multiplied by the number of remaining tasks; Step 5: When the number of remaining windows of all conflicting time slots drops to 0, the processing ends.

8. The method for spatial target perception and scheduling based on genetic algorithm according to claim 1, characterized in that: S23, GA algorithm usage steps and parameter settings The specific process is as follows: Step 1: Set the population size and the upper limit of the number of generations, and randomly generate the initial population. The length of the population is determined according to the encoding process; Step 2: Decode the chromosome to obtain the working time period information of the perception system; Step 3: Calculate the objective function according to the steps in Section 2.2 and calculate the fitness of each chromosome; Step 4: Select the operator using random selection with a ratio of 0.9; Step 5: Reorganization operator, the reorganization ratio is 0.7; step6: mutation operator, the ratio is 0.3; Step 7: After selection, recombination and mutation, the offspring are subjected to step 2 and step 3 operations to calculate the offspring objective function and fitness; Step 8: Judge the objective function and fitness. If they meet the design requirements, go to step 11. If not, go to step 9. Step 9: Update the population, generate a new population from the parent population and the child population according to the objective function values ​​and fitness of the two generations; Step 10: Determine whether the number of offspring in the population exceeds the upper limit of the number of generations. If it exceeds the set value, go to step 11. If it does not exceed the set value, increase the number of offspring by 1 and go to step 2. Step 11: Analyze the optimal value and average value of each generation of the population, and select the appropriate solution. If the conditions are not met, adjust the population size or the upper limit of the number of generations and recalculate.

Images