Optical maneuvering observation station site selection method, system and equipment facing low earth orbit satellite group

Abstract

The invention relates to an optical maneuvering observation station site selection method, system and equipment oriented to a low earth orbit satellite group, which combines the characteristic that an optical maneuvering observation station needs maneuvering, integrates the dimensions of path accessibility, meteorological conditions, land practicability, task guarantee and the like, and further increases the electromagnetic interference factor and task proceeding index of a site, so that the site selection of the optical maneuvering observation station is realized. According to the method, a site selection evaluation system with six dimensions containing 10 core indexes is constructed, automatic multi-source heterogeneous data fusion is realized through QGIS, the problem of difficulty in weight setting is solved through a machine learning method, and finally, a three-stage intelligent observation station site selection scheme for the optical maneuvering observation station site selection problem is constructed by utilizing QGIS + STK screening. Accurate and appropriate site selection of a target low-orbit satellite group for efficiently executing tasks in a specified range is achieved, part of unnecessary cost of repeated site investigation in the site selection process is reduced, and meanwhile reference is provided for automatic site selection analysis of same-type satellite observation infrastructure construction.

Description

Technical Field

[0001] The present invention belongs to the technical field of satellite observation site selection, and relates to an optical mobile survey station site selection method, system and equipment for a low-orbit satellite cluster. Background Art

[0002] Due to their geographical location, some low-orbit satellite target groups have low visibility to established tracking and control centers (either invisible during specific periods or with limited visibility windows). This results in low-quality observations of specific targets by tracking and control stations. Therefore, observations of these specific target groups in central land areas are considered. However, these areas generally lack well-established tracking and control centers. A literature review reveals that existing site selection techniques for ground-based tracking and control stations are primarily based on manual survey statistics, resulting in high site selection costs and difficulty in transferring these experience to central land areas with fragmented terrain and dense populations. Most existing literature only considers climatic conditions, employing a single evaluation dimension. When dealing with multiple targets, it is difficult to manually assign a reasonable numerical ratio to each target. Therefore, accurately and efficiently selecting suitable sites for optical mobile stations to perform observation missions in central land areas remains a major technical challenge. Summary of the Invention

[0003] In response to the problems existing in the above-mentioned traditional technologies, the present invention proposes a method for selecting an optical mobile station site for a low-orbit satellite cluster, a system for selecting an optical mobile station site for a low-orbit satellite cluster, and a station site selection device, which can accurately and efficiently select suitable sites for optical mobile stations to perform observation tasks in central land areas.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions: On the one hand, a method for selecting an optical mobile station site for a low-orbit satellite cluster is provided, comprising the steps of: Grid the target area where the optical mobile measurement station is to be deployed, extract the grid vertices, and remove the grids corresponding to unusable locations based on the terrain of the target area to obtain the initial station site set; Load the trained XGBoost model to perform classification prediction on the initial site set to obtain the predicted probability of each initial site in the initial site set; Based on the comparison between the predicted probability of each initial site and the set threshold, a set of optimal sites is screened out; According to the coverage range and electromagnetic interference score of the target low-orbit satellite cluster corresponding to the current preferred site, the optimal site set of the optical mobile station in the target area is determined from the preferred site set; The XGBoost model is obtained by using the training set for model training. The training set includes positive sample location data and negative sample location data. The positive sample location data includes the existing ground observation stations in the target area, the site location comprehensive data set and the first label column. The negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the site location comprehensive data set and the second label column. The site location comprehensive data set is obtained by using the quantum geographic information system to match and fuse the feature data corresponding to the initial site set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the scope, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

[0005] On the other hand, an optical mobile station site selection system for low-orbit satellite clusters is also provided, including: The indicator construction module is used to grid the target area where the optical mobile measurement station is to be deployed, extract the grid vertices, and eliminate the grids corresponding to unusable locations based on the terrain of the target area to obtain the initial station site set; The model prediction module is used to load the trained XGBoost model to perform classification prediction on the initial site set and obtain the prediction probability of each initial site in the initial site set; The optimal screening module is used to screen out the optimal site set based on the comparison results of the predicted probability of each initial site and the set threshold; The optimal determination module is used to determine the optimal site set of the optical mobile station in the target area from the set of preferred sites based on the coverage range and electromagnetic interference score of the target low-orbit satellite cluster corresponding to the current preferred site; wherein, The XGBoost model is obtained by using the training set for model training. The training set includes positive sample location data and negative sample location data. The positive sample location data includes the existing ground observation stations in the target area, the site location comprehensive data set and the first label column. The negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the site location comprehensive data set and the second label column. The site location comprehensive data set is obtained by using the quantum geographic information system to match and fuse the feature data corresponding to the initial site set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the scope, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

[0006] On the other hand, a station site selection device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the above-mentioned optical mobile station site selection method for low-orbit satellite clusters when executing the computer program.

[0007] One of the above technical solutions has the following advantages and beneficial effects: The above-mentioned optical mobile station site selection method, system and equipment for low-orbit satellite clusters designed a staged site selection theoretical framework and practical method for optical mobile station site selection: combining the characteristics of optical mobile stations requiring mobility, integrating dimensions such as path accessibility, meteorological conditions, land practicality and mission assurance, and further adding electromagnetic interference factors and mission progress indicators of the site, constructing a six-dimensional site selection evaluation system containing 10 core indicators and realizing automatic multi-source heterogeneous data fusion through QGIS, solving the problem of difficult weight setting through machine learning methods, and finally using QGIS+STK screening to complete the construction of a three-stage intelligent station site selection solution for the optical mobile station site selection problem, efficiently executing the accurate and appropriate site selection of the mission target low-orbit satellite cluster within the specified range, reducing some unnecessary costs of repeated site surveys during the site selection process, and also providing a reference for automated site selection analysis of the construction of similar satellite observation infrastructure. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the conventional technology, the following briefly introduces the drawings required for use in the embodiments or the conventional technology descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0009] Figure 1 1 is a flow chart of a method for selecting a site for an optical mobile survey station for a low-orbit satellite cluster in one embodiment; Figure 2 A schematic diagram of the overall design process of a method for selecting an optical mobile station site for a low-orbit satellite cluster in one embodiment; Figure 3 Construct a flow chart for the indicator system in one embodiment; Figure 4 A schematic diagram of a two-level multi-dimensional site selection indicator system in one embodiment; Figure 5 This is a flowchart of the first stage: QGIS initial site selection method in one embodiment; Figure 6 A schematic diagram of the QGIS operation process for gridding the central land area in one embodiment; Figure 7 A schematic diagram of the QGIS operation process for grid point data in one embodiment; Figure 8 A schematic diagram of an example QGIS process for generating a road distance score in one embodiment; Figure 9This is a flow chart of the second stage: XGBoost candidate site screening method in one embodiment; Figure 10 Schematic diagram of the module framework of an optical mobile station site selection system for a low-orbit satellite cluster in one embodiment. DETAILED DESCRIPTION

[0010] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the description of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0011] It should be noted that, when referred to in this document as an "embodiment", it means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The presentation of this phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It will be understood by those skilled in the art that the embodiments described herein may be combined with other embodiments. The term "and / or" used in the specification of the present invention refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0012] The following describes the implementation of the present invention in detail with reference to the accompanying drawings in the embodiments of the present invention.

[0013] In one embodiment, Figure 1 As shown, a method for selecting an optical mobile station site for a low-orbit satellite cluster is provided, which may include the following steps S10 to S16: S10, dividing the target area where the optical mobile measurement station is to be deployed into a grid, extracting grid vertices, and eliminating grids corresponding to unusable locations based on the terrain of the target area to obtain an initial station site set; S12, loading the trained XGBoost model to perform classification prediction on the initial site set, and obtaining the prediction probability of each initial site in the initial site set; S14, based on the comparison result of the predicted probability of each initial site and the set threshold, a set of better sites is screened; S16, determining the optimal site set of the optical mobile station in the target area from the set of preferred sites based on the coverage range and electromagnetic interference score of the target low-orbit satellite constellation corresponding to the current preferred site; wherein, The XGBoost model is obtained by using the training set for model training. The training set includes positive sample location data and negative sample location data. The positive sample location data includes the existing ground observation stations in the target area, the site location comprehensive data set and the first label column. The negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the site location comprehensive data set and the second label column. The site location comprehensive data set is obtained by using the quantum geographic information system to match and fuse the feature data corresponding to the initial site set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the scope, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

[0014] It can be understood that this embodiment systematically designs a phased site selection framework and practical method for the site selection of optical mobile survey stations: combining the characteristics of the need for mobile deployment of optical mobile survey stations, integrating dimensions such as path accessibility, meteorological conditions, land practicality and mission assurance, and further adding electromagnetic interference factors and mission progress indicators of the site, constructing a site selection evaluation system containing no less than 10 core indicators in the above six dimensions, and realizing automatic multi-source heterogeneous data fusion by using QGIS (Quantum Geographic Information System); solving the problem of difficult weight setting by machine learning methods; finally, using QGIS+STK screening to construct an optical mobile survey station site selection method for low-orbit satellite clusters, which is a three-stage intelligent survey station site selection method, and completing a case analysis by taking the target area as the central land area. The three-stage intelligent survey station site selection method proposed in this article can quickly process and fuse multi-source heterogeneous data, and select suitable sites within the specified range for the low-orbit satellite cluster of the mission target, which is conducive to reducing the unnecessary cost of repeated site surveys during the site selection process, and also provides a reference for the automated site selection analysis of the construction of similar satellite observation infrastructure, such as Figure 2 The figure shows the overall design flow chart of the optical mobile station site selection method for low-orbit satellite clusters.

[0015] The above-mentioned optical mobile station site selection method for low-orbit satellite clusters designs a staged site selection theoretical framework and practical method for optical mobile station site selection: combining the characteristics of optical mobile stations requiring mobility, integrating dimensions such as path accessibility, meteorological conditions, land practicality and mission assurance, and further adding electromagnetic interference factors and mission progress indicators of the site, constructing a six-dimensional site selection evaluation system containing 10 core indicators and realizing automated multi-source heterogeneous data fusion through QGIS, solving the problem of difficult weight setting through machine learning methods, and finally using QGIS+STK screening to complete the construction of a three-stage intelligent station site selection solution for the optical mobile station site selection problem, efficiently executing the accurate and appropriate site selection of the mission target low-orbit satellite cluster within the specified range, reducing some unnecessary costs of repeated site surveys during the site selection process, and also providing a reference for automated site selection analysis of similar satellite observation infrastructure construction.

[0016] in, Figure 3 The flowchart constructed for the site selection index system mainly consists of two main steps: site selection problem background analysis and site selection index system design: first, a multi-dimensional analysis of the problem background is conducted, and then the site selection indicators are selected based on the analysis and the site selection index system is designed.

[0017] Since some low-orbit satellite constellations have low visibility to the established tracking and control centers (invisible during specific periods or with a small number of visible time windows), it is considered to dispatch optical mobile stations to the central land area to observe specific target low-orbit satellite constellations. However, the central land area lacks well-built tracking and control centers. Therefore, it is necessary to first select a set of station sites in the central land area that can better carry out specific observation tasks and are suitable for the deployment of optical mobile stations for observation, so as to provide pre-emptive guarantees for the subsequent planning of multi-station rendezvous observation missions.

[0018] When researching the site selection system for mobile optical observation stations, the background research primarily focused on the target area and the mission context. The site selection area for this specification is the central mainland region, which has unique environmental characteristics due to its geographical location, climatic characteristics, and diverse ecological and economic landscape. The central mainland region boasts complex terrain and diverse landforms, but is generally dominated by low-elevation plains and hills, making it particularly suitable for the deployment of mobile optical observation equipment. Furthermore, the central mainland region boasts a well-developed water system, with numerous rivers flowing through it. Therefore, during the site selection process, special attention must be paid to ensuring that the station location intersects with this complex water system.

[0019] The impact of climate is particularly important for observation missions, so minimizing climate impacts is a primary consideration when selecting a site. For optical observation missions, atmospheric cloud thickness and cloud cover are crucial for effective mission execution. Therefore, it is important to consider whether the number of clear days and nights at the selected site meets the requirements for long-term observations. Land attributes in the central land region encompass a variety of land types, including wetlands, forests, ecological reserves, farmland, and residential areas. When considering the site selection system, it is important to consider site availability: whether the selected site location allows for the deployment of an optical mobile station and the execution of continuous observations within the expected mission timeframe. In summary, the environmental characteristics of the central land region exhibit a combination of geographical diversity, monsoonal climate, and land attribute diversity, providing additional considerations for the design of a site selection index system for mobile optical stations.

[0020] Mobile optical stations need to be transported from the measurement and control center to the target site before conducting observation missions and returned to the measurement and control center after the mission. Therefore, transportation accessibility is also an important factor to consider. From the perspective of mission support, to ensure the continuous and effective execution of the mission, issues such as the transportation of bulk supplies, whether the power supply supports the deployment and operation of the station, and the commuting convenience of staff must also be considered. Electromagnetic interference can be a major factor affecting the support of collaborative missions, so it requires further consideration during the site selection process. Finally, from the perspective of mission execution, the site selection must also consider whether the site can ultimately effectively cover the target mission group, that is, whether the coverage rate of the target mission group meets the expected target.

[0021] The site selection indicators selected in this manual are based on the following principles: (1) Systematic principle, that is, the indicators should cover as many dimensions as possible to avoid single-dimensional optimization. (2) Operability principle, that is, the use of accessible and quantifiable spatial data sources is preferred. (3) Regional adaptability principle, that is, the indicators should be customized according to the characteristics of the central land area.

[0022] Then, combining the regional characteristics and mission background of the optical mobile observation station site selection problem in the central land area, a two-level multi-dimensional observation and control station site selection index system (2 levels, 6 dimensions, 10 indicators) was designed. The system framework diagram is shown in Figure 4 The selection and design details of each indicator will be described in detail below. Table 1 is a table of symbols involved: Table 1

[0023] Compared to fixed ground-based stations built at the original site, which are immovable after construction, mobile optical stations offer greater mobility: Each time a mission is assigned, the mobile optical station is dispatched from the measurement and control center to the designated target site, where it is deployed and used as a fixed station for observation. After the mission is complete, the mobile optical station must be returned to the measurement and control center. Between observation missions, if time permits, observation equipment may be replaced. Therefore, when selecting a site, transportation needs, i.e., the site's path accessibility, must be fully considered. This specification uses the nearest neighbor distance between the site and the road (i.e., road proximity) as one of the site selection criteria. The formula for calculating road proximity is as follows:

[0024] in, d road is the straight-line distance (km) from the candidate site to the nearest road, and ALL Roads is the road set consisting of all roads near the candidate site; d j For candidate site locations to j The Euclidean distance (km) of each road. After obtaining the road proximity, the path accessibility is rescored using the road proximity reclassification scoring table in Table 2 below.

[0025] Table 2

[0026] To ensure the mission can be carried out continuously and effectively, it is also necessary to consider issues such as the transportation of bulk supplies, whether the power supply can support the deployment and operation of the observation station, and the commuting convenience of the staff. Therefore, two site selection indicators are designed to represent these issues: railway proximity and the number of buildings within the range. Among them, based on construction costs and subsequent support needs, the railway proximity is designed as follows: d railway =min{ d k | k ∈ALL Roads} in, d railway is the straight-line distance from the candidate site to the nearest railway (km), ALL railways The railway set is composed of all railways near the candidate station location; d k For candidate site locations to k The Euclidean distance (km) between the railways. After obtaining the railway proximity, the railway proximity reclassification scoring table in Table 3 below is used for rescoring.

[0027] Table 3

[0028] To ensure power supply and daily commuting for staff, the site selection must not be a desolate place. A relatively intuitive indicator of population activity is the number of buildings within the range. When selecting a site, the number of buildings within a certain range is designed as a comprehensive indicator that combines the above two types of security effects. The grid distance is selected as 5 kilometers, and the number of buildings within the grid distance is counted:

[0029] in, N building The total number of buildings within a 5-kilometer radius of the candidate site; N total is the total number of buildings in the study area; I (·) is an indicator function, which takes 1 when the condition is met and 0 otherwise; d n For candidate site locations to n Euclidean distance (km) to buildings.

[0030] Since optical observations require clear weather, the main consideration for climate conditions is cloud cover. Since observations also need to be conducted at night, the meteorological condition indicators include the average number of clear nights and the average number of clear days per year. t A clear night is defined as a night with cloud cover less than 20% between 18:00 and 6:00. CN t , No. t A sunny day is one with cloud cover less than 20% between 6:00 and 18:00. CD t The formula for calculating the average number of clear nights (days) per year is as follows:

[0031] in are the average number of clear nights and the average number of clear days per year, N year is the number of years for statistical analysis. Table 4 shows the reclassification scoring rules for the number of sunny days and clear nights: Table 4

[0032] When selecting a site, the primary consideration is whether the land at that location is suitable for equipment installation. Land availability is the basis for all subsequent work. Therefore, the design land suitability indicators are as follows: (1) Terrain analysis: When performing observation missions, the optical mobile station needs to avoid high-altitude peaks and plateaus. The central part of the land is mostly plains or low mountain and hilly areas. Such terrain can be avoided in the large-scale screening. However, due to the developed water system in the central part of the land, it is necessary to consider whether the pre-selected site location overlaps with the water system. Therefore, the water system site selection index is designed: I water =

[0033] in I water represents a binary indicator variable (value 10 represents yes, value 0 represents no).

[0034] (2) Land attributes: The availability of site selection also requires judging the land attributes of the area and whether it is in a restricted area. Restricted areas usually need to be excluded when selecting a site. The land attribute integer encoding vector is designed as follows: L =[ l 1, l 2,…, l p ], l i ∈{0,1},

[0035] in, l i For the i binary identification of land-like attributes (1 = belongs to the category, 0 = does not belong); p The total number of land attributes (such as agricultural, industrial and residential).

[0036] Electromagnetic interference (EMI) indicators are mainly targeted at airports and high-voltage lines. After the interference source generates a buffer zone within a certain distance range, the distance from the station location to the buffer zone of the interference source is calculated. The closer the distance, the more susceptible the station location is to electromagnetic interference. The following is a general formula for calculating the distance of electromagnetic interference factors: d EMI =min{ d m | m ∈ALL EMI Sources} in, d EMI is the distance (km) from the candidate site to the buffer zone (airport, high-voltage line) of the nearest electromagnetic interference source; ALL EMI Sources is the interference source set consisting of all interference sources near the candidate site; d m For candidate site locations to mThe Euclidean distance (km) of the buffer zone around each interference source.

[0037] Table 5

[0038] Table 5 shows a reclassification score table based on the distance from the interference source buffer zone. In some embodiments, susceptibility to electromagnetic interference is primarily determined based on the distance from the site to the interference source's buffer zone. Since electromagnetic interference is not a hard constraint, it is not used as a criterion for eliminating sites. In other words, a score of 0 warrants careful consideration of the candidate site. However, if there are no alternative sites, the candidate site may be considered for use. After calculating and scoring the two electromagnetic interference factors, a linear sum is used to obtain the electromagnetic environment score for a candidate site.

[0039] The selection of the station site is based on the specific observation target group. Therefore, the station site selection needs to consider whether the station location has a high mission performance for the target low-orbit satellite group. The design uses the ratio of the total number of target low-orbit satellites that can be observed at a certain station location to the total number of target low-orbit satellites in the target low-orbit satellite group as the mission performance indicator. The formula is as follows:

[0040] in, N visible , N totaltarget They respectively represent the number of target low-orbit satellites that can be observed by a certain station and the total number of target low-orbit satellites in the target low-orbit satellite constellation; S Target low-orbit satellite cluster S =[ s 1, s 2,…, s Ntotaltarget ]; I (·) is the indicator function, when the satellite s In time T If the elevation angle condition is met at least once, it takes 1, otherwise it takes 0; θ s ( P , t ) indicates satellite s In time Time relative to the station address P elevation angle; θ min is the minimum elevation angle threshold.

[0041] After determining each indicator, the theoretical verification of the site selection indicator system can be further carried out, and the independence of the indicators can be verified by Pearson correlation (i.e., Pearson correlation coefficient, which measures the linear correlation between two continuous variables and has a value range of [-1,1]). Table 6 is a correlation analysis table between indicator characteristics.

[0042] Table 6

[0043] Table 6 shows that, with the exception of the clear night and clear day scores, which exhibit strong correlations, the correlations between the remaining features used for model training are all below 0.5, indicating that the overall selection of indicators for site selection is reasonable. Because the number of clear nights and clear days is strongly correlated, the number of clear nights and clear days was fused into the model for training in the experiment.

[0044] Specifically, the design of the three-stage intelligent station site selection method includes the first stage of initial site data fusion, the second stage of alternative site screening, and the third stage of site set determination.

[0045] Figure 5 This is the flowchart for the first phase of initial site data fusion. The first step is to download vector data for each province. QGIS is used to grid the area covered by the central land area and extract grid vertices as the pre-selected site location set. Based on the terrain (plateaus, high-altitude mountains), a part of the point set is first eliminated to obtain the initial site set. Then, QGIS is used to process the various site selection indicator data of the site selection indicator system. After forming feature data, QGIS is used to match and fuse the initial site set and feature data to obtain a comprehensive site location dataset.

[0046] In order to select the initial site set more evenly in the central land area, the central land area must first be gridded. The satellite data receiving station covers a radius of about 2000km to 3000km. To increase the screening accuracy, the grid scale for the initial screening of potential available site locations in a large area is set to 50km. QGIS is used to grid the central land area at a scale of 50km. The specific QGIS operation process is shown in Figure 6 After gridding, use QGIS to extract the pre-selected site location set. For the specific QGIS operation process, see Figure 7 .

[0047] Regarding the use of QGIS to process the various site selection indicator data of the site selection index system, after forming the characteristic data, QGIS is used to match and fuse the initial site set and the characteristic data. Specifically, Table 7 is a list of data sources used for the site selection indicators. After downloading the obtained data, QGIS is used to process the original data of each site selection indicator and generate the corresponding score (i.e., characteristic data). Figure 8The QGIS processing flow is demonstrated using the example of road data processing to generate road distance scores. QGIS can be used to automate data processing from data input to re-scoring.

[0048] Table 7

[0049] Figure 9 This is a flowchart for the second-stage candidate site screening process. The method employed consists of three main steps: first, data integration and preprocessing of the positive and negative sample labels and features selected from the comprehensive site location dataset are performed to form an intelligent site selection dataset. The intelligent site selection dataset is then used to train and evaluate the XGBoost (eXtreme GradientBoosting) model. Finally, the trained XGBoost model is used to predict and screen the initial site set from the first stage. A classification report is then output. The AUC-ROC ratio is an established and important metric for evaluating classification model performance.

[0050] The intelligent site selection dataset includes a training set and a test set. The training set is used to train and optimize the XGBoost model, and the test set is the data of the initial site set that has undergone preliminary screening in the first stage, which is used for model evaluation.

[0051] The training set mainly consists of two parts: (1) positive sample location data, i.e., existing ground observation stations, combined with the above-mentioned comprehensive data set of station locations and with a label column of 1 (i.e., the first label column). (2) negative sample location data, obtained by randomly sampling points near the positive sample location data, combined with the above-mentioned comprehensive data set of station locations and with a label column of 0 (i.e., the second label column).

[0052] Since the problem data has a small amount of data and an imbalance of positive and negative samples (the number of existing ground observation stations is small, that is, the positive sample location data is significantly less than the negative sample location data), the XGBoost model is selected as the prediction model here, and the training set is used for model training and site suitability prediction of the initial site set. The meaning of the formula symbols of the XGBoost model is shown in Table 8: Table 8

[0053] The core idea of the XGBoost model is to iteratively train multiple weak learners (usually decision trees) to gradually optimize the prediction results, while introducing regularization strategies to prevent overfitting. The following is an analysis of the core mathematical model and key features of the XGBoost model: (1) Objective function design. The objective function of the XGBoost model consists of a loss function and a regularization term, and the formula is:

[0054] in: For samples i Loss function (such as mean square error, cross entropy); is the model prediction value ( f k Indicates the k trees); is the complexity regularization term of the tree ( is the number of leaf nodes, w is the leaf weight, is a hyperparameter).

[0055] (2) Gradient boosting process. t In iterations, the model optimizes the objective function by adding new trees:

[0056] Approximate the objective function through a second-order Taylor expansion:

[0057] in, is the first-order gradient; is a second-order gradient.

[0058] (3) Tree construction and splitting criteria. For leaf nodes j , its weight w j Determined by the following formula:

[0059] in I i Belong to the node j The sample collection.

[0060] Node Split Gains: When choosing a split point, maximize the following gains:

[0061] in, IL , IR are the left and right child node sample sets after splitting.

[0062] (4) Key features and advantages. Regularization: through and Effectively control model complexity to prevent overfitting. Parallel Computing: Pre-sorts and caches data at the feature granularity level to accelerate the search for split points. Handling Missing Values: Automatically learns the default split direction for missing values, improving robustness. Flexibility: Supports custom loss functions and evaluation metrics, adapting to a variety of tasks. Efficiency: Optimizes memory usage using block storage and compression technology.

[0063] The XGBoost model is used to predict the site data in the initial site set. The predicted probability is used to screen and divide the sites according to the screening rules in the predicted probability classification standard Table 9. It can also be directly classified into two categories based on the actual task needs through the threshold (for example, if the predicted probability is greater than the set threshold 0.5, it is retained, and if the predicted probability is less than the set threshold 0.5, it is eliminated. The set threshold can also be set to other values according to the actual task requirements) to obtain a better site set.

[0064] Table 9

[0065] The third stage is to determine the site set: that is, based on the coverage range of the target low-orbit satellite cluster corresponding to the current optimal site, the optimal site set of the optical mobile station in the target area is determined from the optimal site set.

[0066] In one embodiment, the above step S16 may specifically include the following processing: Based on the number of available optical mobile stations for the target mission in the target area, the machine learning site is determined according to the coverage of the target low-orbit satellite constellation corresponding to the current optimal site; Using electromagnetic interference indicators to score electromagnetic interference for site selection in machine learning; The machine-learned site selections after the electromagnetic interference scores are sorted in descending order and then screened to obtain the optimal site set.

[0067] It is understood that the number of available measuring stations is mainly determined by the total number of optical mobile measuring stations that can be actually called ( n a ) and the total number of planned optical mobile stations ( n p ) jointly decided that, in order to further meet the requirements of cost optimization, the final number of available stations is determined by taking the minimum value of the two quantities: min( n a , n p ).

[0068] The total number of optical mobile stations planned to be used ( n p ) is calculated as follows:

[0069] in n t is the number of satellites in the target low-orbit satellite constellation, t t is the total time available to complete all observation tasks, t m The total time required to complete a satellite's observation mission and the time required to convert the target device and adjust the attitude after the observation mission is completed. n g The total number of optical maneuverable stations required to conduct a rendezvous observation of a target.

[0070] After clarifying the number of available optical mobile stations for the target mission, the site selection in the third phase mainly considers the following two main screening criteria: (1) The coverage range of the target low-orbit satellite constellation corresponding to the current optimal site:

[0071] (2) Electromagnetic radiation environment score: Since the existing measurement station sites with public geographical locations do not consider the impact of electromagnetic interference, there is no corresponding data for machine learning model training. Therefore, the electromagnetic interference score is selected in the form of buffer distance based on the machine learning site selection (as shown in Table 5). Finally, the score is sorted in descending order and further site screening is carried out based on the needs (for example, the three best sites with the top three electromagnetic interference scores are selected to form the optimal site set). For the scores of the airport buffer zone and the high-voltage transmission line buffer zone, the final electromagnetic environment score is obtained by linear addition.

[0072] Based on the second stage, the above two main screening conditions in the third stage can be used to jointly screen out the optimal site set that is more applicable to the problem.

[0073] In some embodiments, a simulation experiment is also provided: Use QGIS for preliminary regional gridding and point set extraction: Use QGIS software to generate a 50×50 km standard grid covering the central land area, and use the vector overlay clipping tool to clip the grid that conforms to the boundary of the central land area and extract the vertices of each grid as a backup point set for site selection.

[0074] (1) Analysis of clear day and clear night data in 2024: In this paper, Python is used to filter and organize ERA5 data, and statistics of clear nights and clear days are made on this basis. The experiment is run on a computer. The distribution trend of the total number of clear days and clear nights obtained from the ERA5 data statistics is roughly consistent along latitude and longitude.

[0075] In 2024, the number of sunny days in the central mainland will range from 100 to 250, with a peak of 250 days (in the northeastern and northwest regions), indicating a high proportion of clear daytime weather. The number of clear nights will range from 100 to 300, with a peak of 300 (approximately 82% of nights clear), significantly exceeding the number of sunny days. This suggests that nighttime atmospheric stability may be greater, arguably paving the way for more nighttime observation missions. However, according to ERA5 data, the number of clear days is relatively low in mountainous areas, while the number of clear days and nights is relatively high in plains. Therefore, based on the analysis of the results, given the significant regional variations in the number of clear nights and days, the average annual number of clear nights and days should be considered a key feature in the site selection dataset.

[0076] (2) Terrain Analysis: Since the central region of the land has few high-altitude mountains and plateaus and is dominated by plains and low hills, the data in the initial site set remains almost unchanged after the altitude terrain screening. Check whether the site location is included in or intersects with the hydrological system in the central region of the land, and further obtain the corresponding water system score of the site based on the given water system site selection index. Table 10 shows some sample data for water system determination. The water system score will be used as one of the features of the intelligent site selection dataset and input into the XGBoost model.

[0077] Table 10

[0078] (3) Regional road network analysis: To quantitatively calculate the road network conditions around each station site, the path accessibility index is combined with the nearest neighbor calculation results relative to the road network through QGIS. The station site location and road network distance are recalculated and scored using QGIS combined with the neighborhood grid. The road network distance score data corresponding to each station site can be obtained. Table 11 shows some sample data of the road network neighbor distance score in the region, representing the reclassified score of the distance between the station site location and the road network neighbor corresponding to each longitude and latitude.

[0079] Table 11

[0080] (4) Regional Railway Analysis: Since the basic rules and operating methods for regional railway analysis are consistent with those for regional road network analysis, they will not be elaborated in this section. Similarly, the regional railway proximity score data for each station location can be obtained. Table 12 shows some sample data for railway proximity scores in some regions, representing the reclassified scores of the distance from the station location to the railway neighbor corresponding to each longitude and latitude. The railway distance score data is also one of the features of the intelligent site selection dataset that is subsequently input into the XGBoost model.

[0081] Table 12

[0082] (5) Land attribute analysis: Table 13

[0083] Table 13 shows the land use characteristics in the OSM data. As can be seen from the data in this table, the land attributes in the central region of the mainland can be roughly divided into 17 categories based on analysis of public data. Since land type is also a characteristic feature of the intelligent site selection dataset, to facilitate machine learning model training, the type data is converted from text to integers using integer encoding. This includes land without planning attributes. Land attributes are numbered from 1 to 17 according to their classification.

[0084] Table 14

[0085] After obtaining the integer number of the land attributes of the entire central land area, the land suitability type characteristic data for a specific site in the intelligent site selection dataset can be obtained. An example of the data is shown in Table 14.

[0086] (6) Number of buildings in a region: The number of buildings in a region is one of the characteristic data in the intelligent site selection data. The number of buildings within a certain range in the OSM building data is counted. Table 15 is an example table of the number of buildings in a region obtained by statistical analysis of the building data in the OSM data.

[0087] Table 15

[0088] After processing the data for each site selection indicator, QGIS was used to rescore each site and attribute in the initial site set and fuse the data based on the attributes. This yielded the intelligent site selection dataset used for XGBoost model training below. Table 16 shows a partial data example of the intelligent site selection dataset.

[0089] Table 16

[0090] The XGBoost model was trained by dividing the site selection training set generated above into a training set and a test set with a ratio of 2:8. To improve the classification performance of the model, different values of different hyperparameters were selected for model training and evaluation.

[0091] Since the number of initial site sets obtained in the first part is large (1951 candidate sites), Table 17 partially displays the positive and negative results obtained by using the prediction model.

[0092] Table 17

[0093] Table 17 shows that the samples with positive prediction results have obvious distribution differences on average in each feature compared with the samples with negative prediction results.

[0094] The experiment observed that the site preliminarily selected by the model prediction is highly similar to the real site data in all indicator dimensions, indicating that the results predicted by the prediction model are consistent with the actual situation and are effective and reasonable.

[0095] QGIS+STK screening: Based on the second phase, further consideration is given to electromagnetic interference factors around each point, mainly considering the electromagnetic interference of airports and high-voltage lines. The OSM data of the central land area is obtained through the Geofabrik tool, and then screened and integrated through QGIS to generate buffers and calculate electromagnetic interference factor scores.

[0096] In the electromagnetic interference buffer zone in the central land area, there is a large overlap between the electromagnetic buffer zones of airports and high-voltage lines. After merging the buffer zones using QGIS, the electromagnetic environment of each site was scored according to the designed interference source buffer distance reclassification score table.

[0097] In order to further refine the site set based on the average single-station coverage of the target low-orbit satellite cluster, the experiment sets the target low-orbit satellite cluster for site selection to a low-orbit satellite cluster with 1,282 satellites. The satellite cluster data includes the satellite number and the corresponding six orbital elements. The one-to-one correspondence time window and maximum elevation angle data between the site location coordinates and the orbital elements of the specific target low-orbit satellite cluster are obtained through joint simulation of the site location coordinates and the orbital elements of the specific target low-orbit satellite cluster.

[0098] Table 18 is an example table of the number of observable satellites and electromagnetic interference scores of 5 sites in the randomly selected set of better sites.

[0099] Table 18

[0100] Ultimately, 30 sites were selected as optimal sites based on coverage of the target LEO satellite constellation (with an observation coverage threshold of 60%) and electromagnetic interference scores. The average number of observations from these 30 sites was taken (each site observed an average of 1,217 target LEO satellites).

[0101] It should be understood that although the above process Figure 1 The steps in the flowchart are shown in the order indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Figure 1At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0102] In one embodiment, Figure 10 As shown, a system 100 for selecting an optical mobile station site for a low-orbit satellite cluster is provided, comprising an indicator construction module 11, a model prediction module 13, a preferred screening module 15, and an optimal determination module 17. The indicator construction module 11 is used to grid the target area where the optical mobile station is to be deployed, extract grid vertices, and eliminate grids corresponding to unavailable locations based on the terrain of the target area to obtain an initial site set. The model prediction module 13 is used to load the trained XGBoost model to perform classification prediction on the initial site set to obtain the predicted probability of each initial site in the initial site set. The preferred screening module 15 is used to screen a preferred site set based on the comparison result of the predicted probability of each initial site with a set threshold. The optimal determination module 17 is used to determine the optimal site set for the optical mobile station in the target area from the preferred site set based on the coverage range and electromagnetic interference score of the target low-orbit satellite cluster corresponding to the current preferred site.

[0103] Among them, the XGBoost model is obtained by using the training set for model training. The training set includes positive sample location data and negative sample location data. The positive sample location data includes the existing ground observation stations in the target area, the site location comprehensive data set and the first label column. The negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the site location comprehensive data set and the second label column. The site location comprehensive data set is obtained by using the quantum geographic information system to match and fuse the feature data corresponding to the initial site set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the range, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

[0104] The optical mobile station site selection system 100 for low-orbit satellite clusters has designed a staged site selection theoretical framework and practical method for optical mobile station site selection: taking into account the characteristics of optical mobile stations requiring mobility, comprehensive dimensions such as path accessibility, meteorological conditions, land practicality and mission assurance, and further adding electromagnetic interference factors and mission progress indicators of the site, a six-dimensional site selection evaluation system including 10 core indicators is constructed, and automatic multi-source heterogeneous data fusion is achieved through QGIS. The problem of difficult weight setting is solved through machine learning methods, and QGIS+STK screening is used to complete the construction of a three-stage intelligent station site selection solution for the optical mobile station site selection problem, effectively executing the accurate and appropriate site selection of the mission target low-orbit satellite cluster within the specified range, reducing some unnecessary costs of repeated site surveys during the site selection process, and also providing a reference for automated site selection analysis of similar satellite observation infrastructure construction.

[0105] In one embodiment, the site selection index in the site selection index system also includes electromagnetic interference indexes for airports and high-voltage transmission lines.

[0106] In one embodiment, the optimal determination module 17 determines the site selection for machine learning based on the number of available optical mobile measurement stations in the target mission corresponding to the target area and the coverage range of the target low-orbit satellite group corresponding to the current preferred site. It also uses the electromagnetic interference index to perform electromagnetic interference scoring on the site selection for machine learning, and sorts the machine learning sites after the electromagnetic interference score in descending order to obtain the optimal site set.

[0107] Regarding the specific limitations of the above-mentioned optical mobile station site selection system 100 for low-orbit satellite clusters, please refer to the corresponding limitations of the various embodiments of the optical mobile station site selection method for low-orbit satellite clusters above, which will not be repeated here.

[0108] In one embodiment, a station site selection device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the following processing steps when executing the computer program: gridding the target area where the optical mobile station is to be deployed, extracting grid vertices and eliminating grids corresponding to unavailable locations according to the terrain of the target area to obtain an initial station site set; loading a trained XGBoost model to perform classification prediction on the initial station site set to obtain a predicted probability of each initial station site in the initial station site set; screening a better station site set based on a comparison result of the predicted probability of each initial station site with a set threshold; and determining the optimal station site set of the optical mobile station in the target area from the better station site set based on the coverage range of the target low-orbit satellite group corresponding to the current better station site.

[0109] Among them, the XGBoost model is obtained by using the training set for model training. The training set includes positive sample location data and negative sample location data. The positive sample location data includes the existing ground observation stations in the target area, the site location comprehensive data set and the first label column. The negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the site location comprehensive data set and the second label column. The site location comprehensive data set is obtained by using the quantum geographic information system to match and fuse the feature data corresponding to the initial site set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the range, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

[0110] In one embodiment, when the processor executes the computer program, it can also implement the sub-steps of other embodiments of the above-mentioned optical mobile station site selection method for low-orbit satellite clusters.

[0111] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus DRAM (RDRAM), and DDR DRAM.

[0112] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0113] The above embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that a person skilled in the art would be able to make numerous modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of the present invention. Therefore, the scope of the present invention shall be determined by the appended claims.

Claims

1. A method for selecting an optical mobile station site for a low-orbit satellite cluster, characterized in that: Including steps: Gridding the target area where the optical mobile survey station is to be deployed, extracting grid vertices and eliminating grids corresponding to unusable locations based on the terrain of the target area to obtain an initial station site set; Loading the trained XGBoost model to perform classification prediction on the initial site set, and obtaining the prediction probability of each initial site in the initial site set; Based on the comparison between the predicted probability of each initial site and the set threshold, a set of optimal sites is screened out; According to the coverage range and electromagnetic interference score of the target low-orbit satellite cluster corresponding to the current preferred site, the optimal site set of the optical mobile station in the target area is determined from the preferred site set; wherein, The XGBoost model is obtained by using a training set for model training, wherein the training set includes positive sample location data and negative sample location data, the positive sample location data includes existing ground observation stations in the target area, a comprehensive data set of station locations and a first label column, the negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the comprehensive data set of station locations and a second label column, and the comprehensive data set of station locations is obtained by using a quantum geographic information system to match and fuse the feature data corresponding to the initial station location set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the range, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

2. The method for selecting an optical mobile station site for a low-orbit satellite cluster according to claim 1, wherein: The site selection index system also includes electromagnetic interference indexes for airports and high-voltage transmission lines.

3. The method for selecting an optical mobile station site for a low-orbit satellite cluster according to claim 2, wherein: The step of determining an optimal site set for the optical mobile station in the target area from the set of preferred sites based on the coverage range and electromagnetic interference score of the target low-orbit satellite constellation corresponding to the current preferred site comprises: Based on the number of optical mobile measurement stations available for the target mission corresponding to the target area, the location of machine learning is determined according to the coverage range of the target low-orbit satellite constellation corresponding to the current preferred location; Using the electromagnetic interference index to score the electromagnetic interference of the site selected for machine learning; The machine-learned site selections after the electromagnetic interference scores are sorted in descending order and then screened to obtain the optimal site set.

4. An optical mobile station site selection system for low-orbit satellite clusters, characterized in that: include: An indicator construction module is used to grid the target area where the optical mobile measurement station is to be deployed, extract grid vertices, and eliminate grids corresponding to unusable locations based on the terrain of the target area to obtain an initial station site set; A model prediction module is used to load the trained XGBoost model to perform classification prediction on the initial site set to obtain the prediction probability of each initial site in the initial site set; The optimal screening module is used to screen out the optimal site set based on the comparison results of the predicted probability of each initial site and the set threshold; The optimal determination module is used to determine the optimal site set of the optical mobile station in the target area from the set of preferred sites based on the coverage range and electromagnetic interference score of the target low-orbit satellite cluster corresponding to the current preferred site; wherein, The XGBoost model is obtained by using a training set for model training, wherein the training set includes positive sample location data and negative sample location data, the positive sample location data includes existing ground observation stations in the target area, a comprehensive data set of station locations and a first label column, the negative sample location data includes data obtained by randomly sampling location data other than the positive sample location, the comprehensive data set of station locations and a second label column, and the comprehensive data set of station locations is obtained by using a quantum geographic information system to match and fuse the feature data corresponding to the initial station location set and the constructed site selection index system; the site selection indicators in the site selection index system include road proximity, railway proximity, number of buildings within the range, average number of clear nights per year, average number of sunny days per year, water system site selection indicators, land attributes and task progress indicators.

5. The optical mobile station site selection system for low-orbit satellite clusters according to claim 4, characterized in that: The site selection index system also includes electromagnetic interference indexes for airports and high-voltage transmission lines.

6. The optical mobile station site selection system for low-orbit satellite clusters according to claim 4 or 5, characterized in that: The optimal determination module determines the site selection for machine learning based on the number of the optical mobile measurement stations available in the target mission corresponding to the target area, and the coverage range of the target low-orbit satellite cluster corresponding to the current preferred site. It also uses the electromagnetic interference index to perform electromagnetic interference scoring on the site selection for machine learning, and sorts the machine learning sites after the electromagnetic interference score in descending order and then filters to obtain the optimal site set.

7. A station site selection device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the optical mobile station site selection method for a low-orbit satellite cluster as described in any one of claims 1 to 3 are implemented.

Images