Land surface remote sensing virtual constellation selection method and device, medium and equipment

Abstract

The invention discloses a land surface remote sensing virtual constellation selection method and device, a medium and equipment. The land surface remote sensing virtual constellation selection method comprises the following steps: retrieving a candidate satellite list meeting requirements from a pre-constructed virtual constellation domain knowledge graph according to an input query demand; generating a plurality of groups of candidate virtual constellations according to the candidate satellite list, and calculating an observation capability index of each candidate virtual constellation; constructing a fuzzy membership function according to a preset fuzzy comprehensive evaluation requirement, and calculating a single-factor evaluation matrix according to the fuzzy membership function; calculating a total score of each candidate virtual constellation according to the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation; and determining a final target virtual constellation according to the total score of each candidate virtual constellation. According to the method, the adaptation degree of the satellite combination and land surface task parameters can be improved, and the satellite resource utilization rate and the observation task planning efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and in particular to a method, apparatus, medium, and equipment for selecting a virtual constellation for land surface remote sensing. Background Technology

[0002] To improve the efficiency of satellite resource utilization and meet the demand for multi-dimensional remote sensing observation capabilities for global land surface dynamic monitoring, a large number of existing studies are guided by typical land surface observation tasks, coordinate the collaborative observation capabilities of multi-source satellites, and develop virtual constellation intelligent optimization technology, thereby improving the efficiency of remote sensing data utilization and providing technical support for natural resource monitoring, ecological environment assessment and other operations.

[0003] However, the existing schemes adopt a general design that lacks targeted optimization for land surface remote sensing observation scenarios and fails to integrate the unique characteristics of land surface observation. This results in a reduced compatibility between satellite combinations and land surface mission parameters, thereby affecting the utilization rate of satellite resources and the planning efficiency of observation missions. Summary of the Invention

[0004] This invention provides a method, apparatus, medium, and equipment for selecting a virtual constellation for land surface remote sensing, which can improve the compatibility between satellite combinations and land surface mission parameters, and enhance satellite resource utilization and observation mission planning efficiency.

[0005] This invention provides a method for selecting a virtual constellation for land surface remote sensing, the method comprising:

[0006] Based on the input query requirements, retrieve a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph;

[0007] Several candidate virtual constellations are generated based on the candidate satellite list, and the observation capability indicators of each candidate virtual constellation are calculated.

[0008] A fuzzy membership function is constructed based on the preset fuzzy comprehensive evaluation requirements, and a single-factor evaluation matrix is ​​calculated based on the fuzzy membership function.

[0009] The total score of each candidate virtual constellation is calculated based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation.

[0010] The final target virtual constellation is determined based on the total score of each candidate virtual constellation.

[0011] Preferably, the construction process of the virtual constellation domain knowledge graph includes:

[0012] Collect data from multiple sources;

[0013] A domain ontology for a virtual constellation is constructed based on the multi-source data;

[0014] The structured data in the multi-source data is converted into triples through field mapping, which serve as entity relations;

[0015] Furthermore, a large language model is used to extract application scenario entities from the unstructured data in the multi-source data, and entity relationships are obtained by prompting engineering design annotation rules.

[0016] The domain ontology and entity relationships obtained by storing them using a distributed storage method are used to obtain the virtual constellation domain knowledge graph.

[0017] Furthermore, the multi-source data includes satellite metadata, sensor data, band data, application scenario data, and data product metadata;

[0018] The domain ontology includes satellite ontology, sensor ontology, band ontology, application scenario ontology, and data product ontology.

[0019] Preferably, the step of retrieving a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph based on the input query specifically includes:

[0020] Transform the query requirements into a query statement;

[0021] Based on the filtering rules corresponding to the query statement, candidate entities that meet the requirements are retrieved from the virtual constellation domain knowledge graph to obtain the candidate satellite list;

[0022] The selection criteria include at least one of the following: spatial resolution range, temporal resolution range, and spectral range.

[0023] Preferably, the step of generating several candidate virtual constellations based on the candidate satellite list specifically includes:

[0024] The candidate satellites in the candidate satellite list are grouped by sensor type to generate several candidate virtual constellations;

[0025] The observation capability indicators include spatial resolution, temporal resolution, spectral resolution, and spectral coverage.

[0026] As a preferred embodiment, the step of constructing a fuzzy membership function based on preset fuzzy comprehensive evaluation requirements and calculating a single-factor evaluation matrix based on the membership function specifically includes:

[0027] Based on the threshold, breakthrough value, and target value in the fuzzy comprehensive evaluation requirements, determine the fuzzy membership function of different observation capability indicators;

[0028] Based on the fuzzy membership function, a fuzzy synthesis operation is performed to determine the fuzzy relationship matrix of the membership degree of different observation capability indicators to each evaluation level, which is the single-factor evaluation matrix.

[0029] Preferably, the total score is a score. constellation =WRV T ;

[0030] Where W is the index weight vector of the observation capability indicators of the candidate virtual constellation, R is the single-factor evaluation matrix of the candidate virtual constellation, and V T This represents the score vector corresponding to each level of the observation capability index of the candidate virtual constellation.

[0031] This invention also provides a land surface remote sensing virtual constellation selection device, the device comprising:

[0032] The retrieval module is used to retrieve a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph based on the input query.

[0033] The index module is used to generate several groups of candidate virtual constellations based on the candidate satellite list, and to calculate the observation capability index of each candidate virtual constellation.

[0034] The fuzzy module is used to construct a fuzzy membership function based on the preset fuzzy comprehensive evaluation requirements, and to calculate a single-factor evaluation matrix based on the fuzzy membership function.

[0035] The scoring module is used to calculate the total score of each candidate virtual constellation based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation.

[0036] The determination module is used to determine the final target virtual constellation based on the total score of each candidate virtual constellation.

[0037] This invention also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the land surface remote sensing virtual constellation selection method described in any of the above embodiments.

[0038] This invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the land surface remote sensing virtual constellation selection method described in any of the above embodiments.

[0039] This invention provides a method, apparatus, medium, and device for selecting a virtual constellation for land surface remote sensing. The method involves retrieving a list of candidate satellites that meet the requirements from a pre-constructed virtual constellation domain knowledge graph based on input query needs; generating several candidate virtual constellations from the candidate satellite list and calculating the observation capability index of each candidate virtual constellation; constructing a fuzzy membership function based on preset fuzzy comprehensive evaluation requirements; calculating a single-factor evaluation matrix based on the fuzzy membership function; calculating the total score of each candidate virtual constellation based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation; and determining the final target virtual constellation based on the total score of each candidate virtual constellation. This application can improve the adaptability of satellite combinations to land surface mission parameters, enhance satellite resource utilization, and improve the efficiency of observation mission planning. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating a method for selecting a virtual constellation for land surface remote sensing provided in an embodiment of the present invention;

[0041] Figure 2 This is a flowchart illustrating the unstructured data extraction process provided in an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of the structure of the domain ontology provided in an embodiment of the present invention;

[0043] Figure 4 This is a schematic diagram of the hierarchical model provided in an embodiment of the present invention;

[0044] Figure 5 This is a system architecture diagram of the virtual constellation question-and-answer system provided in an embodiment of the present invention;

[0045] Figure 6 A schematic diagram of the structure of a land surface remote sensing virtual constellation selection device provided in this embodiment of the invention;

[0046] Figure 7 This is a schematic diagram of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] This invention provides a method for selecting a virtual constellation for land surface remote sensing, see [link to relevant documentation]. Figure 1The diagram shown is a flowchart illustrating a method for selecting a virtual constellation for land surface remote sensing according to an embodiment of the present invention, including steps S1 to S5:

[0049] Step S1: Retrieve a list of candidate satellites that meet the requirements from the pre-built virtual constellation domain knowledge graph according to the input query requirements;

[0050] Step S2: Generate several candidate virtual constellations based on the candidate satellite list, and calculate the observation capability index of each candidate virtual constellation;

[0051] Step S3: Construct a fuzzy membership function based on the preset fuzzy comprehensive evaluation requirements, and calculate the single-factor evaluation matrix based on the fuzzy membership function;

[0052] Step S4: Calculate the total score of each candidate virtual constellation based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation;

[0053] Step S5: Determine the final target virtual constellation based on the total score of each candidate virtual constellation.

[0054] In the specific implementation of this embodiment, a candidate satellite list retrieval is performed first.

[0055] A knowledge graph is pre-constructed and maintained, collecting a large amount of satellite data, including basic satellite information, orbital parameters (orbital altitude, inclination, eccentricity, etc.), payload parameters (type, resolution, swath width, etc.), and constellation affiliation. Knowledge extraction techniques are used to extract relevant information from structured, semi-structured, and unstructured data, transforming it into triples of the knowledge graph and storing it in a graph database. Data is regularly updated to ensure the timeliness and accuracy of the knowledge graph. The pre-constructed virtual constellation domain knowledge graph is a structured form of knowledge representation, organizing information in a graph format. It includes various satellite-related entities (such as the satellite itself, its constellation affiliation, orbital parameters, etc.) and relationships between satellites (such as a satellite belonging to a certain constellation, a satellite operating in a specific orbit, etc.).

[0056] Receive the query requirements input by the user, such as conditions on satellite orbit types (low Earth orbit, medium Earth orbit, high Earth orbit), resolution requirements (greater than a certain value), coverage area preferences, etc. Use natural language processing technology to parse the requirements and transform them into a query statement structure that the graph database can understand, clarifying the entity types and relationship conditions to be matched. Apply the parsed query statement to the knowledge graph and perform a query operation in the graph database. The database traverses the nodes and edges in the knowledge graph, filters out satellite entities that meet the query conditions, and returns a list of candidate satellites. Exemplarily, if the requirement is "an optical satellite in low Earth orbit with a resolution greater than 1 meter", then find satellite nodes with a low Earth orbit type in the knowledge graph, and then filter out satellites equipped with optical payloads and with a resolution greater than 1 meter to form a list of candidate satellites.

[0057] According to the list of candidate satellites, use a combination algorithm to combine the list of candidate satellites. For example, if there are n candidate satellites and each virtual constellation is set to contain k satellites (k < n), then calculate all possible combinations containing k satellites through permutations and combinations to form a set of candidate virtual constellations. The combination method can also be optimized according to actual requirements, such as considering constraints like the uniformity of satellite orbit distribution, to reduce unreasonable combinations.

[0058] Establish corresponding calculation models for different observation ability indicators. Exemplarily:

[0059] Perform coverage area calculation. According to satellite orbit parameters (height, inclination) and the Earth model, use geometric methods to calculate the visible area of a single satellite on the Earth's surface at a certain moment, and then comprehensively consider the visible areas of all satellites in the constellation to calculate the total coverage area of the constellation on the Earth. Considering the satellite's operating cycle and orbital phase, calculate the coverage situation at different time points to obtain the average coverage area and the coverage change law.

[0060] Substitute the parameters of each satellite in each candidate virtual constellation into the corresponding calculation model to calculate the values of various observation ability indicators for each candidate virtual constellation. For example, for a candidate virtual constellation containing 3 satellites, calculate the coverage area of each satellite respectively, and then combine them to calculate the total coverage area of the constellation; calculate the revisit period according to the satellite orbit parameters; calculate the detection probability according to the payload and target parameters, etc. Organize the calculation results in tabular form for convenient subsequent analysis and comparison.

[0061] Construction of fuzzy membership functions: Define fuzzy membership functions according to the characteristics of observation ability indicators and actual requirements. Common function forms include triangular functions, trapezoidal functions, Gaussian functions, etc.

[0062] For the coverage area indicator, if the excellent coverage area threshold is set as A1, good as A2 (A2 < A1), and medium as A3 (A3 < A2), a triangular membership function can be defined:

[0063] For each observation capability index of each candidate virtual constellation, the membership degree of that index at different evaluation levels (excellent, good, average, poor) is calculated based on its calculated value and the corresponding fuzzy membership function. The membership degree results of all indicators are arranged into a matrix to obtain the single-factor evaluation matrix. For example, assuming there are m observation capability indicators and n evaluation levels, the single-factor evaluation matrix R is m rows and n columns, r ij This represents the membership degree of the i-th indicator to the j-th evaluation level.

[0064] The weights of each observation capability indicator were determined using methods such as the Analytic Hierarchy Process (AHP). A judgment matrix was constructed, and experts were invited to score the relative importance of different indicators pairwise. For example, for the three indicators of coverage area, revisit period, and detection probability, experts judged that coverage area was slightly more important than revisit period, giving it a score of 3; coverage area was significantly more important than detection probability, giving it a score of 5, and so on. The eigenvectors and the largest eigenvalue of the judgment matrix were calculated to perform a consistency check, ensuring the rationality of the expert judgments. If the check passed, the resulting eigenvectors were the weight vectors of each indicator.

[0065] The single-factor evaluation matrix R and the weight vector A are subjected to fuzzy operations, generally using a weighted average fuzzy operator. Finally, based on the principle of maximum membership, the evaluation level with the highest comprehensive membership is selected as the evaluation level of the candidate virtual constellation; alternatively, the total score can be calculated and used to rank the candidate virtual constellations.

[0066] Fuzzy comprehensive evaluation, based on the membership theory of fuzzy mathematics, transforms qualitative evaluation into quantitative evaluation. For the evaluation of virtual constellations, fuzzy membership functions are constructed to describe the degree of membership of each observation capability indicator to different evaluation levels (e.g., excellent, good, average, poor). Since observation capability indicators are often fuzzy, such as the difficulty in precisely quantifying descriptions like "good coverage performance," fuzzy membership functions can reasonably quantify them. A single-factor evaluation matrix is ​​calculated based on these functions, reflecting the membership distribution of each observation capability indicator at different evaluation levels. Then, combined with the weights of each indicator (reflecting the relative importance of the indicator in the overall evaluation), fuzzy computation is used to obtain the total score for each candidate virtual constellation, thereby achieving a quantitative evaluation of the comprehensive performance of the virtual constellation.

[0067] Based on the total score of each candidate virtual constellation, they are sorted from highest to lowest. The candidate virtual constellation with the highest total score is selected as the final target virtual constellation. If multiple candidate virtual constellations have the same total score, their performance on key indicators can be further compared, or other practical factors can be combined for screening to determine the target virtual constellation that best meets the needs.

[0068] By rapidly retrieving candidate satellites using a knowledge graph, the system avoids blindly sifting through massive amounts of satellite data, significantly reducing data processing volume and screening time. It can flexibly generate different lists of candidate satellites and candidate virtual constellations based on diverse user query requirements. Whether it's a requirement for a specific orbit, specific payload performance, or a specific observation area and mission, it can quickly respond and provide suitable constellation solutions. This improves the compatibility of satellite combinations with land surface mission parameters, enhancing satellite resource utilization and the efficiency of observation mission planning.

[0069] In another embodiment provided by the present invention, the construction process of the virtual constellation domain knowledge graph includes:

[0070] Collect data from multiple sources;

[0071] A domain ontology for a virtual constellation is constructed based on the multi-source data;

[0072] The structured data in the multi-source data is converted into triples through field mapping, which serve as entity relations;

[0073] Furthermore, a large language model is used to extract application scenario entities from the unstructured data in the multi-source data, and entity relationships are obtained by prompting engineering design annotation rules.

[0074] The domain ontology and entity relationships obtained by storing them using a distributed storage method are used to obtain the virtual constellation domain knowledge graph.

[0075] In the specific implementation of this embodiment, multi-source data collection is performed first.

[0076] Clearly define data sources: Establish diverse data acquisition channels for the virtual constellation domain, obtaining structured data such as satellite parameter data (e.g., orbital altitude, inclination, orbital period, payload type, and performance indicators); and unstructured data including satellite application cases and mission scenario descriptions. Furthermore, dynamic data such as real-time satellite operational status can be acquired through sensor networks.

[0077] Data acquisition execution: For structured data, data is extracted and downloaded in batches from official databases using API interfaces, database connections, and other methods; for unstructured data, text information is crawled from designated websites, literature databases, etc., according to set rules using web crawling technology.

[0078] Construct a domain ontology to clarify the hierarchical and relational relationships between different entities. Use ontology modeling tools (such as Protege) to construct these concepts and their relationships into a domain ontology model in graphical or textual form, providing a framework for subsequent data transformation and knowledge graph construction.

[0079] The collected structured data is analyzed to determine the correspondence between data fields and concepts and relationships in the domain ontology. For example, the "Satellite ID" field in the data corresponds to the unique identifier attribute of the "Satellite" concept in the domain ontology; the "Orbital Altitude" field corresponds to the orbital attribute of the "Satellite" concept; and the "Constellation Name" field establishes a "belongs to" relationship between "Satellite" and "Constellation".

[0080] During structured data extraction, the structured data is converted into triples based on the field mapping results. By writing data processing scripts, the structured data is converted into triples (entity-attribute-value) in batches, serving as entity relationships in the knowledge graph.

[0081] For unstructured data extraction, see [link / reference]. Figure 2 This is a flowchart illustrating the unstructured data extraction process provided in an embodiment of the present invention.

[0082] Entity extraction is performed using large-scale language models. Appropriate large-scale language models (such as GPT series, LLaMA, etc.) are selected, and the collected unstructured text data is input into the model. Targeted prompts are designed to guide the model in identifying and extracting application scenario entities related to virtual constellations from the text. The prompt content and format are optimized based on the characteristics of the virtual constellation domain to improve the accuracy and consistency of model extraction. Simultaneously, detailed annotation rules are formulated, clearly defining the annotation standards for entity types and relationship types. Based on these annotation rules, the model extraction results are manually reviewed and corrected, supplementing missing relationship information to ultimately obtain complete entity relationships.

[0083] Considering the large volume and rapid growth of the virtual constellation domain knowledge graph, a distributed storage system (such as Apache Cassandra or HBase) is adopted. This highly scalable, highly available, and fault-tolerant storage system can meet the storage and management needs of large-scale data. The constructed domain ontology and extracted entity relationship data are organized and stored according to the format requirements of the distributed storage system. Through data sharding and replication technologies, the data is distributed across multiple nodes to ensure data security and retrieval efficiency. Simultaneously, an indexing mechanism is established to facilitate quick querying and retrieval of information within the knowledge graph, ultimately forming a complete virtual constellation domain knowledge graph.

[0084] For example, Neo4j graph database is used to store entities and relationships, and remote sensing image metadata is stored using PostgreSQL+MinIO distributed storage.

[0085] The construction of the domain ontology and the extraction of entity relationships clearly demonstrate the intrinsic connections between various concepts within the virtual constellation domain. Based on the knowledge graph, complex semantic queries and reasoning can be performed to uncover hidden knowledge and patterns. The use of distributed storage effectively solves the storage challenges of large-scale knowledge graph data.

[0086] In another embodiment provided by the present invention, the multi-source data includes satellite metadata, sensor data, band data, application scenario data, and data product metadata;

[0087] The domain ontology includes satellite ontology, sensor ontology, band ontology, application scenario ontology, and data product ontology.

[0088] In this specific implementation, information on satellites, sensors, bands, application scenarios, and data products required for virtual constellation construction is collected, specifically including:

[0089] Satellite metadata: covers orbit type, operational status, affiliated agency, launch site, etc., and sources include public databases such as GEOSatDB, WMO, and CEOS;

[0090] Sensor information: Records sensor technical parameters (observation technique, field of view geometry, imaging method, etc.);

[0091] Band data includes band type (optical / microwave), observation mode (active / passive), spectral range, etc.

[0092] Application scenario data: Natural language descriptions are obtained from organizations such as CEOS, such as "ice and snow monitoring" and "forest fire monitoring";

[0093] Data product metadata: including product name, format, band information and storage path.

[0094] When constructing a domain ontology, see [link / reference]. Figure 3 This is a schematic diagram of the domain ontology provided in an embodiment of the present invention.

[0095] The virtual constellation domain ontology is constructed using a top-down approach, and includes five core entities.

[0096] Satellite body (OsatOsat): Reuses the GEOSatDB satellite body to define satellite attributes (name, orbital parameters, operational status, etc.);

[0097] Sensor body (OsenOsen): Describes the sensor's technical parameters and its "onboard" relationship with the satellite;

[0098] Band body (OwavOwav): Records the band type, observation mode, and its "inclusion" relationship with the sensor;

[0099] Application scenario ontology (OappOapp): Based on WMO standards, it defines land surface parameters (such as NDVI), observation requirements (spatial / temporal resolution thresholds), and scenario classification;

[0100] Data Product Body (OdatOdat): Associates sensors and products, recording product format, band coverage, and storage path.

[0101] In another embodiment provided by the present invention, step S1 specifically includes:

[0102] Transform the query requirements into a query statement;

[0103] Based on the filtering rules corresponding to the query statement, candidate entities that meet the requirements are retrieved from the virtual constellation domain knowledge graph to obtain the candidate satellite list;

[0104] The selection criteria include at least one of the following: spatial resolution range, temporal resolution range, and spectral range.

[0105] In this specific implementation, the natural language requirement, such as "satellites suitable for sea surface temperature monitoring", is converted into a Cypher query statement through a question-and-answer system;

[0106] Based on the filtering rules corresponding to the query statement, candidate entities that meet the requirements are retrieved from the virtual constellation domain knowledge graph, and satellites that meet the requirements are retrieved to obtain the candidate satellite list;

[0107] It should be noted that the screening rules include at least one of the following: spatial resolution range, temporal resolution range, and spectral range.

[0108] In another embodiment of the present invention, the step of generating several groups of candidate virtual constellations based on the candidate satellite list specifically includes:

[0109] The candidate satellites in the candidate satellite list are grouped by sensor type to generate several candidate virtual constellations;

[0110] The observation capability indicators include spatial resolution, temporal resolution, spectral resolution, and spectral coverage.

[0111] In this specific implementation, candidate satellites are grouped according to optical / microwave sensor type to avoid mixing of heterogeneous sensors and generate multiple candidate virtual constellations;

[0112] Calculate the observation capabilities of candidate virtual constellations and determine the observation capability indicators.

[0113] The indicators of observation capabilities include spatial resolution, temporal resolution, spectral resolution, and spectral coverage.

[0114] The spatial resolution and spectral resolution are taken as the minimum values ​​of the satellites in the combination;

[0115] The time resolution can be calculated using the CEOS COVE tool;

[0116] Spectral coverage can be calculated to determine the proportion of combined bands that cover the required bands.

[0117] Spectral coverage

[0118] Among them, B constellation B indicates the wavelength range included in this optical virtual constellation. rrquirement This indicates the band range required for observation in the current application scenario.

[0119] It should be noted that, in addition to spatial resolution, temporal resolution, spectral resolution, and spectral coverage, the observation capability indicators also support user-defined indicators.

[0120] In another embodiment provided by the present invention, step S3 specifically includes:

[0121] Based on the threshold, breakthrough value, and target value in the fuzzy comprehensive evaluation requirements, determine the fuzzy membership function of different observation capability indicators;

[0122] Based on the fuzzy membership function, a fuzzy synthesis operation is performed to determine the fuzzy relationship matrix of the membership degree of different observation capability indicators to each evaluation level, which is the single-factor evaluation matrix.

[0123] In this specific implementation, the Analytic Hierarchy Process (AHP) is used to determine the weights:

[0124] To construct a hierarchical model, see [link / reference]. Figure 4 This is a schematic diagram of the hierarchical model provided in the embodiments of the present invention.

[0125] The target layer is the "optimal virtual constellation", the criterion layer is the evaluation index, and the scheme layer is the candidate combination;

[0126] Construct a pairwise comparison matrix (see Table 1), calculate the weights using the geometric mean method, and obtain the evaluation factor weight vector W. Pass the consistency test (CR < 0.1).

[0127] Table 1. Pairwise Comparison Matrix of Emergency Observation Scenarios for Snow Cover Monitoring

[0128]

[0129] Perform fuzzy comprehensive evaluation;

[0130] Let the evaluation set V = {v1, v2, v3, v4} = {Excellent, Good, Average, Poor}, and the fuzzy comprehensive evaluation requirement X = (t, b, g), where t represents the threshold, b represents the breakthrough value, and g represents the target value. The "threshold" is the minimum requirement that must be met to ensure the data is useful; the "target value" is an ideal requirement, beyond which further improvement is unnecessary; the "breakthrough value" is the intermediate level between the "threshold" and the "target value," and its achievement will lead to a significant improvement in the application of the target.

[0131] Based on the threshold, breakthrough value, and target value in the fuzzy comprehensive evaluation requirements, the fuzzy membership functions of different observation capability indicators are determined, and the single-factor evaluation matrix R is calculated. The specific method is as follows:

[0132] Let μ1(x), μ2(x), μ3(x), and μ4(x) represent four single-factor fuzzy comprehensive evaluation subsets: “Excellent,” “Good,” “Average,” and “Poor,” respectively. The membership function is chosen as a trapezoidal distribution function. Then, the four fuzzy membership functions are as follows:

[0133]

[0134]

[0135] Here, x represents the specific value of a certain indicator of the current virtual constellation. For example, if the spatial resolution of a constellation is 30m, then the value of x is 30.

[0136] Through fuzzy synthesis operations, the fuzzy relation matrix of the degree of membership of each satellite observation capability factor to each evaluation level is obtained, i.e., the single-factor evaluation matrix R.

[0137] Where r in matrix R ij This indicates that in a certain virtual constellation combination, factor u i For the level v in the fuzzy subset j The degree of membership.

[0138] In another embodiment provided by the present invention, the total score is a score. constellation =WRV T ;

[0139] Where W is the index weight vector of the observation capability indicators of the candidate virtual constellation, R is the single-factor evaluation matrix of the candidate virtual constellation, and V T This represents the score vector corresponding to each level of the observation capability index of the candidate virtual constellation.

[0140] Specifically, the total score is calculated based on weighted averages, and the best results are output in order of score ranking. This represents the score vector corresponding to each level. constellationThe overall score for this virtual constellation combination will be between 4 and 10, with 4 to 6 being average, 6 to 8 being good, and 8 to 10 being excellent.

[0141] It should be noted that this embodiment of the invention also provides a virtual constellation question-and-answer system, see [link to documentation]. Figure 5 This is a system architecture diagram of the virtual constellation question-and-answer system provided in an embodiment of the present invention.

[0142] The system includes:

[0143] Knowledge Q&A module: Integrates Neo4j knowledge graph and GPT-3.5, and implements natural language to Cypher query through the Langchain framework;

[0144] Combination optimization module: calls the AHP-FCE algorithm library to output virtual constellation rankings;

[0145] Data retrieval module: Accesses multi-source remote sensing data and supports semantic queries.

[0146] In practical applications:

[0147] It supports intelligent question-and-answer functionality, where users can input "I want to conduct water body monitoring in the ** area, please provide me with some satellites and display the basic information of these satellites in a table," and the system will return a list of satellites and data access links.

[0148] It supports a combined recommendation function, which outputs virtual constellation scores and optimal solutions based on user scenarios (such as "snow accumulation monitoring");

[0149] It supports data retrieval, allowing users to filter images by time, location, and cloud cover, and display the results visually.

[0150] This invention proposes a method for optimizing virtual constellation combinations for typical land surface remote sensing observation tasks based on a knowledge graph-enhanced large-scale language model. By designing a virtual constellation domain ontology, it fully extracts and integrates virtual constellation domain knowledge from satellites, sensors, bands, application scenarios, and data products.

[0151] The embodiments provided in this invention construct a domain knowledge graph. Compared with existing methods that rely on keywords or tags for retrieval, this invention can understand user query intent at a deeper level, providing rich and professional domain knowledge support for intelligent retrieval of satellite data. A virtual constellation combination optimization method combining the Analytic Hierarchy Process (AHP) and the Fuzzy Comprehensive Evaluation (FCE) method is also presented. A large-scale language model assists in the initial satellite screening, and the constellation observation capability evaluation method of FCE and the evaluation index weight calculation method of AHP are combined to evaluate the performance of observation indicators such as spatiotemporal resolution of the virtual constellation, ensuring that the optimized virtual constellation better meets user needs. This method overcomes the limitation of existing technologies that lack virtual constellation construction methods for different land surface remote sensing observation tasks, providing a scientific basis for the adaptive use and promotion of virtual constellations.

[0152] The embodiments provided by this invention construct a question-answering system based on a virtual constellation domain knowledge graph. By enhancing a large-scale language model, it achieves accurate understanding of the user's query intent and finds suitable remote sensing satellites or data products from the knowledge base. Simultaneously, by combining a multi-source data access and retrieval system, it realizes intelligent retrieval of satellite data. Compared to traditional image data retrieval services, this invention can provide query services based on user semantics, significantly improving the efficiency and accuracy of satellite data retrieval.

[0153] This invention covers the diverse needs of typical land surface observation missions and has a wide range of applications. By defining a virtual constellation domain ontology that encompasses core entities such as satellites, sensors, and application scenarios, and designing customized evaluation indicators for scenarios such as snow and ice monitoring and disaster response, it can be extended to multiple fields such as agriculture, ecology, and urban monitoring, which is superior to the limitations of single-scenario technologies in existing technologies.

[0154] In another embodiment of the present invention, a land surface remote sensing virtual constellation selection device is provided, see [link to previous document]. Figure 6 This is a schematic diagram of a land surface remote sensing virtual constellation selection device provided in an embodiment of the present invention. The device includes:

[0155] The retrieval module is used to retrieve a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph based on the input query.

[0156] The index module is used to generate several groups of candidate virtual constellations based on the candidate satellite list, and to calculate the observation capability index of each candidate virtual constellation.

[0157] The fuzzy module is used to construct a fuzzy membership function based on the preset fuzzy comprehensive evaluation requirements, and to calculate a single-factor evaluation matrix based on the fuzzy membership function.

[0158] The scoring module is used to calculate the total score of each candidate virtual constellation based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation.

[0159] The determination module is used to determine the final target virtual constellation based on the total score of each candidate virtual constellation.

[0160] It should be noted that the specific functions of each module are described in the above embodiments of the land surface remote sensing virtual constellation selection method, and will not be repeated in this embodiment.

[0161] See Figure 7 This is a schematic diagram of a terminal device provided in an embodiment of the present invention. The terminal device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a land surface remote sensing virtual constellation selection program. When the processor executes the computer program, it implements the steps in the various embodiments of the land surface remote sensing virtual constellation selection method described above, for example... Figure 1 The steps S1 to S5 are shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments.

[0162] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device. For example, the computer program can be divided into a land parcel division module, a component generation module, a model generation module, a model adjustment module, a texture pasting module, and a road network generation module; the specific functions are not elaborated here.

[0163] The terminal device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0164] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0165] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0166] Wherein, if the modules / units integrated in the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0167] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0168] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for selecting a virtual constellation for land surface remote sensing, characterized in that, include: Based on the input query requirements, retrieve a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph; Several candidate virtual constellations are generated based on the candidate satellite list, and the observation capability indicators of each candidate virtual constellation are calculated. A fuzzy membership function is constructed based on the preset fuzzy comprehensive evaluation requirements, and a single-factor evaluation matrix is ​​calculated based on the fuzzy membership function. The total score of each candidate virtual constellation is calculated based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation. The final target virtual constellation is determined based on the total score of each candidate virtual constellation.

2. The method for selecting a virtual constellation for land surface remote sensing as described in claim 1, characterized in that, The construction process of the virtual constellation domain knowledge graph includes: Collect data from multiple sources; A domain ontology for a virtual constellation is constructed based on the multi-source data; The structured data in the multi-source data is converted into triples through field mapping, which serve as entity relations; Furthermore, a large language model is used to extract application scenario entities from the unstructured data in the multi-source data, and entity relationships are obtained by prompting engineering design annotation rules. The domain ontology and entity relationships obtained by storing them using a distributed storage method are used to obtain the virtual constellation domain knowledge graph.

3. The method for selecting a virtual constellation for land surface remote sensing as described in claim 2, characterized in that, The multi-source data includes satellite metadata, sensor data, band data, application scenario data, and data product metadata; The domain ontology includes satellite ontology, sensor ontology, band ontology, application scenario ontology, and data product ontology.

4. The method for selecting a virtual constellation for land surface remote sensing as described in claim 1, characterized in that, The step of retrieving a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph based on the input query needs specifically includes: Transform the query requirements into a query statement; Based on the filtering rules corresponding to the query statement, candidate entities that meet the requirements are retrieved from the virtual constellation domain knowledge graph to obtain the candidate satellite list; The selection criteria include at least one of the following: spatial resolution range, temporal resolution range, and spectral range.

5. The method for selecting a virtual constellation for land surface remote sensing as described in claim 1, characterized in that, The step of generating several candidate virtual constellations based on the candidate satellite list specifically includes: The candidate satellites in the candidate satellite list are grouped by sensor type to generate several candidate virtual constellations; The observation capability indicators include spatial resolution, temporal resolution, spectral resolution, and spectral coverage.

6. The method for selecting a virtual constellation for land surface remote sensing as described in claim 1, characterized in that, The step of constructing a fuzzy membership function based on preset fuzzy comprehensive evaluation requirements, and calculating a single-factor evaluation matrix based on the membership function, specifically includes: Based on the threshold, breakthrough value, and target value in the fuzzy comprehensive evaluation requirements, determine the fuzzy membership function of different observation capability indicators; Based on the fuzzy membership function, a fuzzy synthesis operation is performed to determine the fuzzy relationship matrix of the membership degree of different observation capability indicators to each evaluation level, which is the single-factor evaluation matrix.

7. The method for selecting a virtual constellation for land surface remote sensing as described in claim 1, characterized in that, The total score is a score. constellation =WRV T ; Where W is the index weight vector of the observation capability indicators of the candidate virtual constellation, R is the single-factor evaluation matrix of the candidate virtual constellation, and V T This represents the score vector corresponding to each level of the observation capability index of the candidate virtual constellation.

8. A land surface remote sensing virtual constellation selection device, characterized in that, The device includes: The retrieval module is used to retrieve a list of candidate satellites that meet the requirements from a pre-built virtual constellation domain knowledge graph based on the input query. The index module is used to generate several groups of candidate virtual constellations based on the candidate satellite list, and to calculate the observation capability index of each candidate virtual constellation. The fuzzy module is used to construct a fuzzy membership function based on the preset fuzzy comprehensive evaluation requirements, and to calculate a single-factor evaluation matrix based on the fuzzy membership function. The scoring module is used to calculate the total score of each candidate virtual constellation based on the single-factor evaluation matrix and the observation capability index of each candidate virtual constellation. The determination module is used to determine the final target virtual constellation based on the total score of each candidate virtual constellation.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the land surface remote sensing virtual constellation selection method as described in any one of claims 1 to 7.

10. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the land surface remote sensing virtual constellation selection method as described in any one of claims 1 to 7.

Images