A star selection method and system based on an improved adaptive genetic algorithm

Abstract

The application discloses a star selection method and system based on an improved adaptive genetic algorithm, and relates to the technical field of image positioning and registration. The technical points of the application comprise: preprocessing obtained basic star table data; inputting the preprocessed basic star table data into a selection model based on an improved genetic algorithm for selection; using a binary mode for coding and carrying out population initialization; calculating individual fitness according to a fitness function, and calculating the crossover and mutation probabilities through an improved adaptive adjustment strategy according to the calculated fitness values, and then carrying out selection, crossover and mutation operations according to the calculated crossover and mutation probabilities; decoding the genes and calculating constraint conditions; and outputting a star selection result when the maximum number of iterations is reached and all the pointing field information has been optimally selected. The application can select uniformly-distributed stars in the premise that the number of stars in a given imager field of view is known, so as to provide the imager with observation and correct the pointing of the imager.

Description

Technical Field

[0001] This invention relates to the field of image localization and registration technology, specifically to a star selection method and system based on an improved adaptive genetic algorithm. Background Technology

[0002] With the rapid development of geostationary orbit (GEO) meteorological satellite technology, the requirements for the accuracy of satellite-based Earth observations are becoming increasingly stringent. To address this, meteorological satellites from various countries utilize image positioning and registration techniques to improve observation accuracy. In practical engineering, due to the harsh temperature environment of space, the GEO meteorological satellite imager (hereinafter referred to as "the imager") experiences thermal deformation, leading to pointing deviations and affecting image positioning and registration. Therefore, it is necessary to select stars within the imager's field of view for observation. By simultaneously solving nonlinear equations based on the deviations between the ideal and actual apparent positions of multiple selected stars, the imager's pointing can be corrected. The uniformity of the distribution of selected stars within a single field of view significantly impacts the accuracy of the correction. Currently, most star selection research is based on star sensors, aiming to correct satellite attitude. Compared to imagers, star sensors do not obstruct the view of Earth or planets within the solar system and do not consider the uniformity of star distribution within a single field of view, making them unsuitable for imager pointing correction. Furthermore, star selection methods based on imager pointing correction have not yet been studied. Summary of the Invention

[0003] To this end, this invention proposes a star selection method and system based on an improved adaptive genetic algorithm, which selects uniformly distributed stars given the number of stars in the imager's field of view.

[0004] According to one aspect of the present invention, a star selection method based on an improved adaptive genetic algorithm is proposed, the method comprising the following steps:

[0005] Step 1: Obtain basic star catalog data;

[0006] Step 2: Preprocess the basic star catalog data;

[0007] Step 3: Input the preprocessed basic star catalog data into the selection model based on the improved genetic algorithm to obtain the selection results.

[0008] Furthermore, the basic star catalog data includes imager pointing information, imager instantaneous position and field of view, stellar catalog and planetary ephemeris.

[0009] Furthermore, the preprocessing procedure includes:

[0010] Step 2: Perform domain extraction, segmentation, and deduplication operations on the basic star catalog data;

[0011] Step 22: Read planetary information from the planetary ephemeris, and perform time conversion, coordinate transformation, spatial position calculation, and error correction on the planets to obtain the instantaneous position of the planets relative to the satellites;

[0012] Steps 2 and 3: Read stellar information from the star catalog, and perform time conversion, coordinate transformation, spatial position calculation, and error correction on the stars to obtain the instantaneous position of the stars relative to the satellite;

[0013] Step 24: Based on the imager's optical axis direction, the instantaneous position of the planet relative to the satellite, and the instantaneous position of the star relative to the satellite, preliminary screening is performed to obtain the pre-processed basic star catalog data, i.e., the observation star catalog.

[0014] Furthermore, the specific process of step three includes: extracting stars from the observed star catalog, determining whether the total number of stars exceeds a preset star count threshold, and if so, using an improved genetic algorithm to select stars, including:

[0015] Step 3: Encode the star selection problem using binary code and initialize the population;

[0016] Step 3.2: Calculate the fitness of individuals based on the fitness function, and calculate the crossover and mutation probabilities based on the calculated fitness values ​​using an improved adaptive adjustment strategy. Use the elite preservation strategy to preserve the best individuals in the population evolution process so far, so that they do not participate in the genetic operator operations.

[0017] Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities;

[0018] Steps three and four: Decode the genes and calculate the constraints;

[0019] Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result.

[0020] Furthermore, the process of encoding the star selection problem in binary mode in step 31 includes: dividing the imager's field of view into multiple regions, assigning an observation value of 1 to one star in each region, and 0 to the rest.

[0021] Furthermore, the fitness function mentioned in step 3.2 is a function of the uniformity of star distribution, and the formula for calculating the fitness value is:

[0022]

[0023] In the formula, N th This is a threshold for the number of stars. is the mean of the difference in azimuth angles; diff represents the difference in azimuth angles.

[0024] Furthermore, the improved adaptive adjustment strategy described in step 3.2 uses the activation function Tanh function as the adaptive strategy for crossover and mutation probabilities, and solves for the crossover probability P using the Tanh function. c and the probability of mutation P m The adaptive adjustment formulas are as follows:

[0025]

[0026]

[0027] In the formula, p c max p represents the maximum crossover probability; c min The minimum crossover probability is represented by 'a'; the adaptive control parameter is represented by 'f'; the larger fitness of the two parent individuals is represented by 'favg'; and the average fitness of the population is represented by 'f'. max p represents the maximum fitness of the population. m max p represents the maximum mutation probability; m min denoted by , where represents the minimum mutation probability; f represents the fitness of the mutated individual.

[0028] Further, the specific process of step 33 includes: using the roulette wheel selection method: calculating the fitness value and fitness probability of individuals in the population based on the fitness function, calculating the cumulative probability and assigning it to the corresponding individuals, generating random numbers between 0 and 1, matching them with the cumulative probability of individuals in the current population, determining which individuals can inherit to the next generation, and using the elite preservation strategy to directly copy the best individuals in the population evolution process to the next generation without crossover; the crossover operation is as follows: calculating the average fitness value and the best fitness value of the population, calculating the crossover probability of individuals through the crossover probability adaptive adjustment strategy, and performing the crossover operation using a two-point crossover method according to the crossover probability; the mutation operation is as follows: calculating the mutation probability of individuals using the average fitness value and the best fitness value of the population through the mutation probability adaptive adjustment strategy, and performing the mutation operation using a single-point mutation method.

[0029] According to another aspect of the present invention, a star selection system based on an improved adaptive genetic algorithm is proposed, the system comprising:

[0030] The data acquisition module is configured to acquire basic star catalog data; the basic star catalog data includes imager pointing information, imager instantaneous position and field of view, stellar catalog and planetary ephemeris;

[0031] A preprocessing module configured to preprocess the base star catalog data;

[0032] The star selection module is configured to input preprocessed basic star catalog data into a selection model based on an improved genetic algorithm to select stars and obtain the selection results.

[0033] Furthermore, the star selection module inputs the preprocessed basic star catalog data into a selection model based on an improved genetic algorithm for selection. The process of obtaining the selection results includes: extracting stars from the observed star catalog, determining whether the total number of stars is greater than a preset star number threshold, and if so, selecting stars using the improved genetic algorithm, including:

[0034] Step 3: Encode the star selection problem using binary code and initialize the population;

[0035] Step 3.2: Calculate the fitness of individuals based on the fitness function, and calculate the crossover and mutation probabilities using an improved adaptive adjustment strategy based on the calculated fitness value. Use an elite preservation strategy to retain the best individuals in the population evolution process so far, preventing them from participating in genetic operator operations. The fitness function is a function of the uniformity of stellar distribution, and the formula for calculating the fitness value is:

[0036]

[0037] In the formula, N th This is a threshold for the number of stars. The mean of the azimuth differences is given by denoted as denoted as 'diff'. The improved adaptive adjustment strategy uses the activation function Tanh as the adaptive strategy for crossover and mutation probabilities, and solves for the crossover probability P using the Tanh function. c and the probability of mutation P m The adaptive adjustment formulas are as follows:

[0038]

[0039]

[0040] In the formula, p c max p represents the maximum crossover probability; c min The minimum crossover probability is represented by 'a'; the adaptive control parameter is represented by 'f'; the larger fitness of the two parent individuals is represented by 'favg'; and the average fitness of the population is represented by 'f'. max p represents the maximum fitness of the population. m max p represents the maximum mutation probability; m min denoted by , where represents the minimum mutation probability; f represents the fitness of the mutated individual.

[0041] Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities;

[0042] Steps three and four: Decode the genes and calculate the constraints;

[0043] Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result.

[0044] The beneficial technical effects of this invention are:

[0045] This invention provides a star selection method and system based on an improved adaptive genetic algorithm, which can select uniformly distributed stars for imager observation and correction of imager pointing, given the number of stars in the imager's field of view, thereby reducing the impact on image positioning and registration. Attached Figure Description

[0046] The present invention can be better understood by referring to the description given below in conjunction with the accompanying drawings, which together with the following detailed description are included in and form part of this specification, and are used to further illustrate preferred embodiments of the invention and explain the principles and advantages of the invention.

[0047] Figure 1 This is a flowchart illustrating a star selection method based on an improved adaptive genetic algorithm according to an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of the star catalog preprocessing process in an embodiment of the present invention;

[0049] Figure 3 This is a graph showing the adjustment of the crossover probability in an embodiment of the present invention;

[0050] Figure 4 This is a graph showing the adjustment of the mutation probability in an embodiment of the present invention;

[0051] Figure 5 This is a flowchart illustrating star selection using an improved adaptive genetic algorithm in an embodiment of the present invention;

[0052] Figure 6 This is an improved crossover probability adjustment curve in an embodiment of the present invention;

[0053] Figure 7 This is an improved mutation probability adjustment curve in an embodiment of the present invention;

[0054] Figure 8 This is a star distribution map before star catalog preprocessing in an embodiment of the present invention;

[0055] Figure 9 This is a star distribution map after star catalog preprocessing in an embodiment of the present invention;

[0056] Figure 10 This is a graph showing the optimal fitness curves of the three algorithms in this embodiment of the invention;

[0057] Figure 11 This is a graph showing the average fitness curves of the three algorithms in this embodiment of the invention;

[0058] Figure 12 This is a graph showing the optimal fitness curve of the improved adaptive genetic algorithm in an embodiment of the present invention.

[0059] Figure 13 This is a diagram showing the star selection results in an embodiment of the present invention;

[0060] Figure 14 This is a simulation diagram of the stellar observation software in an embodiment of the present invention. Detailed Implementation

[0061] To enable those skilled in the art to better understand the present invention, exemplary embodiments or examples of the present invention will be described below in conjunction with the accompanying drawings. Obviously, the described embodiments or examples are merely some, not all, of the embodiments or examples of the present invention. All other embodiments or examples obtained by those skilled in the art based on the embodiments or examples of the present invention without inventive effort should fall within the scope of protection of the present invention.

[0062] Star selection is an indispensable part of image localization and registration systems, widely used in star map recognition, satellite attitude correction, and imager pointing correction. Therefore, how to select suitable stars from the star catalog affects both star sensor-based satellite attitude correction and imager pointing correction. Most existing star selection methods divide them into two main parts: basic star catalog preprocessing and model and algorithm construction. Basic star catalog preprocessing can be further subdivided into star catalog data processing, apparent position calculation, and observed star catalog generation. Therefore, this invention addresses the star selection problem for imagers, focusing on optimizing the number of selected stars and the uniformity of star distribution. It investigates three aspects: basic star catalog preprocessing, star selection model establishment, and optimization algorithm solution, designing a star selection method suitable for imagers.

[0063] This invention proposes a star selection method based on an improved adaptive genetic algorithm, which includes the following steps:

[0064] Step 1: Obtain basic star catalog data;

[0065] Step 2: Preprocess the basic star catalog data;

[0066] Step 3: Input the preprocessed basic star catalog data into the selection model based on the improved genetic algorithm to obtain the selection results.

[0067] Figure 1 A flowchart of the star selection method based on the improved adaptive genetic algorithm is shown.

[0068] The method begins with step one. In step one, the basic star catalog data is first acquired, which includes imager pointing information, imager instantaneous position and field of view, stellar catalog and planetary ephemeris.

[0069] Then, proceed to step two to preprocess the basic star catalog data, such as... Figure 2 As shown, the preprocessing process includes:

[0070] Step 2: Perform domain extraction, segmentation, and deduplication operations on the basic star catalog data;

[0071] Step 22: Read planetary information from the planetary ephemeris, and perform time conversion, coordinate transformation, and spatial position calculation on the planets to obtain the instantaneous position of the planets relative to the satellites;

[0072] Steps 2 and 3: Read stellar information from the star catalog, and perform time conversion, coordinate transformation, and spatial position calculation on the stars to obtain the instantaneous position of the stars relative to the satellite;

[0073] Step 24: Based on the imager's optical axis direction, the instantaneous position of the planet relative to the satellite, and the instantaneous position of the star relative to the satellite, preliminary screening is performed to obtain the pre-processed basic star catalog data, i.e., the observation star catalog.

[0074] According to an embodiment of the present invention, the basic star catalog data contains 43 data points, such as the right ascension, declination, proper motion of right ascension and declination, magnitude, parallax, and radial velocity of stars. To expedite the star selection process, the data fields of the star catalog must be extracted, focusing only on those relevant to the research, namely: star number, magnitude, right ascension (hours, minutes, seconds), declination (degrees, arcminutes, arcseconds), proper motion of right ascension and declination, whether it is a binary star, and whether it is a variable star. Magnitude represents the brightness of the star; a higher magnitude indicates a lower brightness.

[0075] According to the imager's star-sensing process, at least 48 star-sensing operations are required per day, meaning 48 star selections are needed. If the entire star catalog is selected each time, efficiency would be severely impacted. Therefore, star catalog segmentation is necessary. Star catalog segmentation divides the processed star catalog into 72 catalogs at 60° intervals based on right ascension and declination, reducing the size of the catalog read by the program each time and thus accelerating the overall efficiency of star selection.

[0076] Then, based on the above star catalog data processing, the apparent positions of celestial bodies in the solar system are calculated according to the planetary ephemeris, including time conversion, coordinate transformation, and spatial position calculation. Then, error correction is performed, including correcting the diurnal parallax effect caused by the Earth's revolution around the Sun, the gravitational bending effect of light rays, nutation correction, and diurnal parallax correction. Finally, the instantaneous position of the planets relative to the satellites is obtained.

[0077] Then, star information is read from the star catalog to calculate the apparent position of the star, including time conversion, coordinate transformation, spatial position calculation, and finally, by correcting the annual parallax effect caused by the Earth's revolution around the Sun, annual parallax correction, light gravitational bending effect, and precession and nutation correction, the instantaneous position of the star relative to the satellite is obtained.

[0078] Finally, based on the imager's optical axis orientation during the mission, as well as the apparent positions of celestial bodies and stars within the solar system, a preliminary screening is conducted to generate an observation catalog. This observation catalog includes scan angles and step angles to provide data preparation for star selection.

[0079] It should be noted that the instantaneous positions of planets relative to satellites, the instantaneous positions of stars relative to satellites, and the calculation and generation of observed star catalogs are all existing technologies, and will not be elaborated on here.

[0080] Then, in step three, the preprocessed basic star catalog data is input into a selection model based on an improved genetic algorithm to select and obtain the selection results.

[0081] According to an embodiment of the present invention, the constraints for star selection based on the imager are as follows: the selected star should not be obscured by the Earth, the Sun, the Moon, or other major planets; the distance of the selected star from the field of view boundary should be greater than the set minimum value; the magnitude of the star needs to meet the star sensitivity requirements of the imager; no other stars are allowed within a specified range around the star; the system stipulates that the number of stars the imager needs to observe should not be less than 4; multiple star observations should be completed according to the time reserved for star sensitivity by the system; and binary stars and variable stars should be avoided as much as possible when using them as observed stars.

[0082] The principles for star selection based on the imager are: (1) Determine the observable stars in the current field of view based on the satellite attitude information, i.e., the work done in preprocessing, and then select them according to the number of stars required by the system; (2) Use the magnitude of the faintest star that the imager star sensor can observe as the magnitude threshold for selecting stars; (3) Observe the spatial position of the stars, i.e., the right ascension and declination of the stars should be separated as much as possible, i.e. the stars are evenly distributed; (4) When the spatial distance between the observed star and the non-observed star is very close, the time for the imager to stare at the star should be increased, so that multiple stars pass through the imager's field of view in sequence, thereby further determining the pointing deviation.

[0083] Based on the above constraints and principles, the optimization objective of the imager-based star selection model is as follows: Given a fixed number of stars to be observed, maximize the uniformity of star distribution, i.e., minimize the standard deviation of star distribution.

[0084]

[0085] In the formula, N th This is a threshold for the number of stars. is the mean of the differences in azimuth angles; diff represents the difference in azimuth angles, and its mathematical expression is:

[0086]

[0087] In the formula, A represents the azimuth angle of the star.

[0088] The present invention mainly considers maximizing the uniformity of star distribution, i.e. minimizing the standard deviation of star distribution, under the premise of determining the number of stars to be observed.

[0089] In simple genetic algorithms, crossover and mutation probabilities are fixed values. When dealing with star selection problems, these need repeated adjustments, and the optimal probabilities cannot be guaranteed. This leads to unstable convergence and a tendency to get stuck in local convergence. In adaptive genetic algorithms (AGA), mutation and crossover probabilities can be adjusted based on the individual's fitness value, falling between the maximum and average fitness of the population. The adjustment curve for the crossover probability is shown below. Figure 3 As shown, the adjustment curve for the mutation probability is as follows: Figure 4 As shown.

[0090] When the fitness of most individuals in a population is concentrated around the population average fitness, these individuals are of comparable quality and constitute the majority of the population. If the population average fitness value and the fitness value of the best individual in the population are not similar, the adjustment curves for crossover and mutation probabilities will become steeper. Consequently, the crossover and mutation probabilities of these individuals will also differ significantly, leading to lower crossover and mutation probabilities for the vast majority of individuals, causing the algorithm to stagnate. Furthermore, in the initial stage of population evolution, the best individuals in the population are usually not the optimal solution. If a simple genetic algorithm is used for adjustment, a large number of local optimum patterns will be preserved, causing the algorithm to stagnate and eventually get stuck in a local optimum. If the adjustment curves for mutation and crossover probabilities are within f... avg A smoother change can avoid the problems that occur in adaptive genetic algorithms. This will increase the mutation and crossover probabilities of individuals whose fitness values ​​are very close to the average. Furthermore, to retain more elite individuals in the later stages of genetic algorithm evolution, the adjustment curves for crossover and inheritance probabilities should be within f...max The area becomes smooth.

[0091] Therefore, this invention designs an improved adaptive adjustment strategy to calculate crossover and mutation probabilities, in order to solve the local convergence problem that occurs in simple genetic algorithms and traditional adaptive genetic algorithms. Figure 5 As shown, the specific process of step three includes: extracting stars from the observed star catalog, determining whether the total number of stars exceeds a preset star count threshold, and if so, using an improved genetic algorithm to select stars, specifically including:

[0092] Step 31: Encode the star selection problem using binary code and initialize the population; that is:

[0093] First, based on the pre-set stellar threshold N for the stars to be observed... th The imager's field of view is divided into N th Each region can select only one star for observation; if the selected star is selected, it is assigned a value of 1; otherwise, it is assigned a value of 0. The encoding format is as follows:

[0094] chromosome={x1,x2,…,x m}

[0095] x m ={x m1 ,x m2 ,…,x mj}

[0096] In the formula, x m This represents the collection of all stars within each region of the imager;

[0097] Next, the population size is set to N, and i represents the number of individuals already generated in the population, initialized to 0. The imager's field of view is divided according to the selected number of star thresholds. For example, if the star threshold is set to 6, the imager's field of view is divided into 6 regions on average according to the azimuth angle. Then, a star is randomly selected from one region of the imager's field of view, and relevant constraint calculations are performed on the selected observed star to determine whether it meets the magnitude threshold and is within the safe field of view. If all constraints are met, the star is selected and the corresponding x is set. mj =1, insert i, and select a star in each segmented region to form an individual according to the above steps; finally, calculate the fitness value of the individual; determine whether the individual is greater than or equal to the population size, then the population initialization is complete, otherwise i = i + 1, continue to initialize the population until the number of individuals meets the initial population size;

[0098] Step 3.2: Calculate the fitness of individuals according to the fitness function, and calculate the crossover and mutation probabilities based on the calculated fitness values ​​using an improved adaptive adjustment strategy. Then, use an elite retention strategy to preserve the best individuals in the population evolution process so far, so that they do not participate in genetic operator operations. The improved adaptive adjustment strategy is: using the Tanh function, a commonly used activation function in neural networks, as the adaptive strategy for crossover and mutation probabilities.

[0099] The original formula for the Tanh function is:

[0100]

[0101] After simplification, we get:

[0102]

[0103] However, the probability ranges from [0,1]. After adjusting the range, we get:

[0104]

[0105] The adaptive adjustment formulas for the crossover and mutation probabilities, designed using the Tanh function to solve the star selection optimization problem, are as follows:

[0106]

[0107]

[0108] In the formula, p c max p represents the maximum crossover probability; c min The minimum crossover probability is represented by 'a'; the adaptive control parameter is represented by 'f'; the larger fitness of the two parent individuals is represented by 'favg'; and the average fitness of the population is represented by 'f'. max p represents the maximum fitness of the population. m max p represents the maximum mutation probability; m min denoted by , where represents the minimum mutation probability; f represents the fitness of the mutated individual.

[0109] The adaptive adjustment curves of crossover and mutation probabilities after Tanh function adjustment are shown below. Figure 6 and Figure 7 As shown in the figure. It can be seen that if most individuals in the current population have similar fitness values, and the fitness of these individuals is similar to the population average fitness f avg When the individuals are relatively close, their crossover and mutation probabilities are significantly higher than the original adaptive adjustment function; if the fitness values ​​of some individuals in the current population are close to the maximum fitness f... maxWhen this happens, the crossover and mutation probabilities of these individuals will be significantly reduced, thus allowing the best-performing individuals to be retained as much as possible.

[0110] Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities;

[0111] Specifically, a roulette wheel selection method is used. First, the fitness value and fitness probability of individuals in the population are calculated based on the fitness function, and the cumulative probability is calculated and assigned to the corresponding individuals. Then, a random number between 0 and 1 is generated and matched with the cumulative probability of individuals in the current population to determine which individuals can inherit to the next generation. Finally, an elite preservation strategy is used to directly copy the best individuals in the population evolution process to the next generation without crossover, to ensure the convergence of the algorithm. The crossover operation is as follows: first, the average fitness value and the best fitness value of the population are calculated, and the crossover probability of individuals is calculated through an adaptive crossover probability adjustment strategy. Crossover is then performed using a two-point crossover method according to the crossover probability. The mutation operation is as follows: based on the average fitness value and the best fitness value of the population, the mutation probability of individuals is calculated using a mutation probability adaptive adjustment strategy, and mutation is performed using a single-point mutation method.

[0112] Steps 3 and 4: Decode the genes and calculate the constraints to generate a new generation of population;

[0113] Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result, obtain the optimal solution, and output the number, right ascension, and declination of the selected star.

[0114] To address the issue of the feasibility of this invention in application, the following software simulation was performed.

[0115] Simulation conditions, such as satellite orbit, imager parameters, and basic star parameters, are set up to conduct a star selection simulation based on the imager. This verifies the practicality and effectiveness of the star selection algorithm based on the improved adaptive genetic algorithm. Here, the planetary ephemeris and star catalog are taken as examples of the dedicated star catalogs of DE421 and Fengyun-4B satellite, respectively.

[0116] Satellite orbit information is shown in Table 1, imager parameter information is shown in Table 2, and basic star parameter information is shown in Table 3.

[0117] Table 1

[0118]

[0119] Table 2

[0120]

[0121] Table 3

[0122]

[0123]

[0124] Figure 8 The star distribution before cataloging preprocessing. Figure 9 The results are shown in Table 4 after preprocessing the star catalog. To verify the effectiveness of the Tanh function-based improvement strategy, 100 evaluation experiments will be conducted to compare the improved adaptive genetic algorithm with the simple genetic algorithm and the traditional adaptive algorithm. The experimental environment is the same for all experiments, and the convergence performance is shown in Table 4.

[0125] Table 4

[0126]

[0127] It can be seen that the improved adaptive genetic algorithm has the most convergence times in 100 experiments, 28 more times than the simple genetic algorithm and 10 more times than the traditional adaptive genetic algorithm. Figure 10 and Figure 11 Performance comparisons of three genetic algorithms are presented. The improved adaptive genetic algorithm proposed in this invention improves the smoothness of the initial stage, increases the crossover and mutation probabilities, and helps the algorithm escape local optima. When the simple genetic algorithm and the adaptive genetic algorithm are in a plateau phase, the improved adaptive genetic algorithm shows an upward trend, escaping local convergence, demonstrating its superior performance. Furthermore, the average fitness of the improved adaptive genetic algorithm is closer to the optimal fitness than that of the simple genetic algorithm and the traditional adaptive genetic algorithm.

[0128] In summary, the star selection method based on the improved adaptive genetic algorithm of this invention has good performance and can improve the stability and speed of star selection.

[0129] Simulation results of the star selection method based on the improved adaptive genetic algorithm are as follows: Figure 12 and Figure 13 As shown, the improved adaptive genetic algorithm achieves global optimal convergence after 75 iterations, with a fitness value of 134.3575 and a running time of 3.39 s. In 100 evaluation experiments, the method converged 4 times at a fitness value of 110.8492, 6 times at a fitness value of 121.7152, and 90 times at a fitness value of 134.3575.

[0130] The single-field star distribution obtained by the star selection method based on the improved adaptive genetic algorithm is shown in Table 5. Figure 9After stellar selection using genetic methods, there were 7 stars in the field of view, a reduction of 695 stars compared to before selection. The azimuth angles of each observed star are shown in Table 5. The uniformity of star distribution was evaluated using the standard deviation of the difference in azimuth angles between adjacent stars. The azimuth angle differences between the 7 satellites were 51.853°, 52.676°, 52.36°, 51.936°, 51.51°, 51.919°, and 47.746°, respectively, with a standard deviation of 0.7842. Figure 14 This is a graph showing the results of running the instruction file generated by the method of this invention in stellar observation testing software. The results demonstrate that the stars selected by the method of this invention meet the requirements for stellar observation. The uniform distribution of the selected stars reduces the pointing error of the imager, proving the effectiveness of the method.

[0131] Table 5

[0132]

[0133]

[0134] Another embodiment of the present invention proposes a star selection system based on an improved adaptive genetic algorithm, the system comprising:

[0135] The data acquisition module is configured to acquire basic star catalog data; the basic star catalog data includes imager pointing information, imager instantaneous position and field of view, stellar catalog and planetary ephemeris;

[0136] A preprocessing module configured to preprocess the base star catalog data;

[0137] The star selection module is configured to input preprocessed basic star catalog data into a selection model based on an improved genetic algorithm to select stars and obtain the selection results.

[0138] In this embodiment, preferably, the process of inputting the preprocessed basic star catalog data into a selection model based on an improved genetic algorithm in the star selection module to obtain the selection results includes: extracting stars from the observed star catalog, determining whether the total number of stars is greater than a preset star number threshold, and if it is greater, selecting stars using an improved genetic algorithm, including:

[0139] Step 3: Encode the star selection problem using binary code and initialize the population;

[0140] Step 3.2: Calculate the fitness of individuals based on the fitness function, and calculate the crossover and mutation probabilities using an improved adaptive adjustment strategy based on the calculated fitness value. Use an elite preservation strategy to retain the best individuals in the population evolution process so far, preventing them from participating in genetic operator operations. The fitness function is a function of the uniformity of stellar distribution, and the formula for calculating the fitness value is:

[0141]

[0142] In the formula, N th This is a threshold for the number of stars. The mean of the azimuth differences is given by denoted as denoted as 'diff'. The improved adaptive adjustment strategy uses the activation function Tanh as the adaptive strategy for crossover and mutation probabilities, and solves for the crossover probability P using the Tanh function. c and the probability of mutation P m The adaptive adjustment formulas are as follows:

[0143]

[0144]

[0145] In the formula, p c max p represents the maximum crossover probability; c min The minimum crossover probability is represented by 'a'; the adaptive control parameter is represented by 'f'; the larger fitness of the two parent individuals is represented by 'favg'; and the average fitness of the population is represented by 'f'. max p represents the maximum fitness of the population. m max p represents the maximum mutation probability; m min denoted by , where represents the minimum mutation probability; f represents the fitness of the mutated individual.

[0146] Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities;

[0147] Steps three and four: Decode the genes and calculate the constraints;

[0148] Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result.

[0149] The function of the star selection system based on the improved adaptive genetic algorithm in this embodiment of the invention can be described by the aforementioned star selection method based on the improved adaptive genetic algorithm. Therefore, for the parts not described in detail in the system embodiment, please refer to the above method embodiment, and they will not be repeated here.

[0150] Although the invention has been described with respect to a limited number of embodiments, those skilled in the art will understand from the foregoing description that other embodiments are conceivable within the scope of the invention described herein. The disclosure of the invention is illustrative and not restrictive, and the scope of the invention is defined by the appended claims.

Claims

1. A star selection method based on an improved adaptive genetic algorithm, characterized in that, Includes the following steps: Step 1: Obtain basic star catalog data; Step 2: Preprocess the basic star catalog data; Step 3: Input the preprocessed basic star catalog data into a selection model based on an improved genetic algorithm to obtain the selection results; this includes: extracting stars from the observed star catalog, determining whether the total number of stars exceeds a preset star count threshold, and if so, using the improved genetic algorithm to select stars, including: Step 3: Encode the star selection problem using binary code and initialize the population; Step 3.2: Calculate the fitness of individuals based on the fitness function, and calculate the crossover and mutation probabilities using an improved adaptive adjustment strategy based on the calculated fitness values. An elite preservation strategy is used to retain the best individuals in the population evolution process so far, preventing them from participating in genetic operator operations. The improved adaptive adjustment strategy uses the activation function - Tanh function as the adaptive strategy for crossover and mutation probabilities, and uses the Tanh function to solve for the crossover probability. and mutation probability The adaptive adjustment formulas are as follows: ； ； In the formula, Indicates the maximum crossover probability; Indicates the minimum crossover probability; 'a' represents the adaptive control parameter. This indicates the higher fitness among the two parent individuals; Indicates the average fitness of the population; Indicates the maximum fitness of the population; Indicates the maximum mutation probability; Indicates the minimum mutation probability; Indicates the fitness of the variant individual; Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities; Steps three and four: Decode the genes and calculate the constraints; Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result.

2. The star selection method based on an improved adaptive genetic algorithm according to claim 1, characterized in that, The basic star catalog data includes imager pointing information, imager instantaneous position and field of view, stellar catalog and planetary ephemeris.

3. The star selection method based on an improved adaptive genetic algorithm according to claim 2, characterized in that, The preprocessing process includes: Step 2: Perform domain extraction, segmentation, and deduplication operations on the basic star catalog data; Step 22: Read planetary information from the planetary ephemeris, and perform time conversion, coordinate transformation, spatial position calculation, and error correction on the planets to obtain the instantaneous position of the planets relative to the satellites; Steps 2 and 3: Read stellar information from the star catalog, and perform time conversion, coordinate transformation, spatial position calculation, and error correction on the stars to obtain the instantaneous position of the stars relative to the satellite; Step 24: Based on the imager's optical axis direction, the instantaneous position of the planet relative to the satellite, and the instantaneous position of the star relative to the satellite, preliminary screening is performed to obtain the pre-processed basic star catalog data, i.e., the observation star catalog.

4. The star selection method based on an improved adaptive genetic algorithm according to claim 1, characterized in that, The process of encoding the star selection problem in step 3.1 using binary method includes: dividing the imager's field of view into multiple regions, assigning an observation value of 1 to one star in each region, and 0 to the rest.

5. The star selection method based on an improved adaptive genetic algorithm according to claim 1, characterized in that, The fitness function mentioned in step 3.2 is a function of the uniformity of star distribution, and the formula for calculating the fitness value is: ； In the formula, This is a threshold for the number of stars. This is the mean of the differences in azimuth angles; This indicates the difference in azimuth angle.

6. The star selection method based on an improved adaptive genetic algorithm according to claim 1, characterized in that, Step 3.3 involves the following steps: Selection is performed using a roulette wheel method: The fitness value and fitness probability of individuals in the population are calculated based on the fitness function, and the cumulative probability is assigned to the corresponding individuals. A random number between 0 and 1 is generated and matched with the cumulative probability of individuals in the current population to determine which individuals can inherit the traits. An elite preservation strategy is used to directly copy the best individuals from the current evolutionary process to the next generation without crossover. Crossover is performed by calculating the average fitness value and the optimal fitness value of the population, calculating the crossover probability of individuals using an adaptive crossover probability adjustment strategy, and performing crossover using a two-point crossover method according to the crossover probability. Mutation is performed by calculating the mutation probability of individuals using the average fitness value and the optimal fitness value of the population, using an adaptive mutation probability adjustment strategy, and performing mutation using a single-point mutation method.

7. A star selection system based on an improved adaptive genetic algorithm, characterized in that, include: The data acquisition module is configured to acquire basic star catalog data. The basic star catalog data includes imager pointing information, imager instantaneous position and field of view, star catalog and planetary ephemeris; A preprocessing module configured to preprocess the base star catalog data; The star selection module is configured to input preprocessed basic star catalog data into a selection model based on an improved genetic algorithm for selection, and obtain the selection results. This includes: extracting stars from the observed star catalog, determining whether the total number of stars exceeds a preset star count threshold, and if so, using the improved genetic algorithm to select stars, including: Step 3: Encode the star selection problem using binary code and initialize the population; Step 3.2: Calculate the fitness of individuals based on the fitness function, and calculate the crossover and mutation probabilities using an improved adaptive adjustment strategy based on the calculated fitness value. Use an elite preservation strategy to retain the best individuals in the population evolution process so far, preventing them from participating in genetic operator operations. The fitness function is a function of the uniformity of stellar distribution, and the formula for calculating the fitness value is: ； In the formula, This is a threshold for the number of stars. This is the mean of the differences in azimuth angles; This represents the difference in azimuth angles; the improved adaptive adjustment strategy uses the activation function Tanh function as the adaptive strategy for crossover and mutation probabilities, and uses the Tanh function to solve for the crossover probability. and mutation probability The adaptive adjustment formulas are as follows: ； ； In the formula, Indicates the maximum crossover probability; Indicates the minimum crossover probability; 'a' represents the adaptive control parameter. This indicates the higher fitness among the two parent individuals; Indicates the average fitness of the population; Indicates the maximum fitness of the population; Indicates the maximum mutation probability; Indicates the minimum mutation probability; Indicates the fitness of the variant individual; Step 3: Perform selection, crossover, and mutation operations based on the calculated crossover and mutation probabilities; Steps three and four: Decode the genes and calculate the constraints; Step 35: Determine if the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, proceed to Step 32. If the maximum number of iterations has been reached, determine if all pointing field information has been optimized. If not, continue to determine if the total number of stars is greater than the preset star number threshold and repeat the above steps. If completed, output the selection result.

Images