A method and system for calculating national land space planning indicators based on large language model

Abstract

The present invention relates to the field of land space planning technology, and in particular to a land space planning indicator calculation method and system based on a large language model, the method comprising: based on a natural language statement with an indicator calculation task requirement input by a user, the name of a calculation tool and the actual numerical value of the input parameter of the calculation tool are derived and output through a large model, and a corresponding space calculation instruction is generated thereby; the space calculation instruction is executed to obtain a corresponding indicator calculation result; the indicator calculation result and the corresponding first task prompt word are input into the large model, and the corresponding land space planning result is obtained through the large model interpretation. The implementation of the method can improve the accuracy, timeliness and reliability of the calculation of land space planning indicators.

Description

Technical Field

[0001] The present invention relates to the technical field of national land space planning, and in particular to a national land space planning indicator calculation method and system based on a large language model. Background Art

[0002] The calculation of national land space planning indicators specifically refers to the process of quantitative analysis and calculation of various planning indicators during the national land space planning process. These planning indicators cover all aspects of national land space resources, including land resources, water resources, ecological environment, urban development, infrastructure construction and other fields.

[0003] At present, with the development of advanced technologies such as big data, geographic information system (GIS), remote sensing technology, and artificial intelligence and machine learning algorithms that have emerged in recent years, these technologies have provided strong data support, spatial analysis capabilities and decision-making basis for national land space planning, making the calculation of planning indicators more scientific, accurate and efficient. For example, the comprehensive evaluation model based on big data can comprehensively evaluate the carrying capacity, environmental conditions and development suitability of national land space resources. Through data analysis, problems and potential in regional development can be discovered, providing a scientific basis for the formulation of planning indicators; however, although big data provides rich information, the quality and accuracy of the data are often uneven. Text data on the Internet may contain erroneous, false or subjective information, which may lead to deviations in the calculation results of planning indicators. For example, the geographic information system (GIS) can conduct in-depth analysis of the distribution, structure, and changes of national land space resources due to its powerful spatial analysis function; however, the GIS system often relies on the update of basic geographic data. If the data update lags behind the actual development and changes, this may cause the calculation results of planning indicators to fail to reflect the latest situation in a timely manner. Machine learning algorithms have also been widely used in the calculation of national land space planning indicators. These algorithms can automatically learn and discover regularities and patterns in data, providing more accurate predictions and evaluations for the formulation of planning indicators; however, the performance of machine learning algorithms is highly dependent on the quantity and quality of training data. If the training data is biased or insufficient, the generalization ability of the model may decrease, thus affecting the accuracy of planning indicators.

[0004] Therefore, there is an urgent need for a comprehensive solution to improve the accuracy, timeliness and reliability of the calculation of national land space planning indicators. Summary of the invention

[0005] In order to improve the accuracy, timeliness and reliability of the calculation of national land space planning indicators, the purpose of the present invention is to provide a method and system for calculating national land space planning indicators based on a large language model. The technical scheme adopted is as follows:

[0006] The present application discloses a method for calculating national land space planning indicators based on a large language model, the method comprising:

[0007] S1. Based on the natural language statements with indicator calculation task requirements input by the user, the big model derives the name of the calculation tool and the actual value of the input parameters of the calculation tool, and generates the corresponding spatial calculation instructions;

[0008] S2, executing the spatial calculation instruction to obtain the corresponding index calculation result;

[0009] S3. Input the indicator calculation results and the corresponding first task prompt words into the large model, and obtain the corresponding national land space planning results through the large model interpretation.

[0010] Furthermore, in step S1, the natural language statement with the indicator calculation task requirement input by the user is derived and output by the big model for the calculation tool name and the actual value of the calculation tool input parameter, and the corresponding spatial calculation instruction is generated therefrom, including:

[0011] S11, obtaining a natural language sentence with indicator calculation task requirements input by the user;

[0012] S12, taking the natural language sentence, the predefined candidate computing tool list, and the second task prompt word for guiding the large language model to recommend computing tools according to the user input and the candidate computing tool list as input, and passing them to the large language model, and the large model feeds back the corresponding computing tool name;

[0013] S13, creating a corresponding tool input parameter list based on the computing tool name, and passing the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input to the large language model, and the large model feeds back a corresponding actual value sequence;

[0014] S14. Based on the calculation tool name and the actual numerical sequence, corresponding spatial calculation instructions are generated through data integration and template filling.

[0015] Furthermore, in step S14, the generating of corresponding spatial calculation instructions based on the calculation tool name and the actual numerical sequence through data integration and template filling includes:

[0016] S141, loading a corresponding spatial instruction template from a preset indicator library based on the calculation tool name, and loading corresponding target spatial data from a preset range line library based on the actual numerical sequence;

[0017] S142: Load the target spatial data into the spatial instruction template to obtain corresponding spatial calculation instructions.

[0018] Furthermore, in step S2, executing the spatial calculation instruction to obtain corresponding index calculation results includes:

[0019] S21, acquiring range line data, and selecting a target block overlapping with the range line data from pre-stored global space data;

[0020] S22, cutting the target image block to obtain a cut image block overlapping with the range line data;

[0021] S23, obtaining the original attribute value of the cropped image block before cropping, and adjusting the original attribute value by using a linear interpolation method to obtain a corresponding new attribute value;

[0022] S24. Summarize the new attribute values ​​obtained to obtain corresponding index calculation results.

[0023] Further, in step S21, the global spatial data is stored in the PostGIS database in the form of points, lines and surfaces in advance;

[0024] The target image block and the cut image block obtained by calculation are stored in a file system in the form of files;

[0025] The index calculation results obtained by executing the spatial calculation instructions are stored in a PostgreSQL database table in a tabular form.

[0026] Furthermore, in step S23, the original attribute value of the cropped image block before cropping is obtained, and the original attribute value is adjusted by using a linear interpolation method to obtain a corresponding new attribute value, including:

[0027] S231, when it is determined that the cut block belongs to a planar element, the original attribute value is linearly adjusted based on the area ratio of the cut block to the target block to obtain a corresponding new attribute value;

[0028] S232: When it is determined that the cut block belongs to a linear element, the original attribute value is

[0029] A linear adjustment is performed based on the length ratio of the blocks to obtain the corresponding new attribute value.

[0030] In the second aspect, the present application also discloses a national land space planning indicator calculation system based on a large language model, the system comprising a spatial calculation instruction generation module, a spatial calculation instruction execution module, and a national land space planning analysis module, wherein:

[0031] The spatial computing instruction generation module is used to generate corresponding spatial computing instructions based on the natural language statement with index computing task requirements input by the user, and to derive and output the computing tool name and the actual value of the computing tool input parameter through the large model;

[0032] The spatial calculation instruction execution module is used to execute the spatial calculation instruction to obtain the corresponding index calculation result;

[0033] The national land space planning analysis module is used to input the indicator calculation results and the corresponding first task prompt words into the large model, and obtain the corresponding national land space planning results through the large model interpretation.

[0034] Furthermore, the spatial computing instruction generation module is also used to obtain a natural language sentence with indicator computing task requirements input by a user; take the natural language sentence, a predefined candidate computing tool list, and a second task prompt word for guiding the large language model to recommend computing tools based on the user input and the candidate computing tool list as input, pass them to the large language model, and the large model feeds back the corresponding computing tool name; create a corresponding tool input parameter list based on the computing tool name, and take the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input, pass them to the large language model, and the large model feeds back the corresponding actual value sequence; based on the computing tool name and the actual value sequence, generate the corresponding spatial computing instruction through data integration and template filling.

[0035] Furthermore, the spatial calculation instruction generation module is also used to load the corresponding spatial instruction template from the preset indicator library based on the calculation tool name, and to load the corresponding target spatial data from the preset range line library based on the actual numerical sequence; the target spatial data is loaded into the spatial instruction template to obtain the corresponding spatial calculation instruction.

[0036] Furthermore, the spatial calculation instruction execution module is also used to obtain range line data, select a target block that overlaps with the range line data from the pre-stored global spatial data; crop the target block to obtain a cropped block that overlaps with the range line data; obtain the original attribute value of the cropped block before cropping, and use a linear interpolation method to adjust the original attribute value to obtain a corresponding new attribute value; and summarize the obtained new attribute values ​​to obtain a corresponding indicator calculation result.

[0037] The present invention has the following beneficial effects:

[0038] 1. Through the derivation of the names and parameters of the calculation tools by the big model, the automatic generation of spatial calculation instructions is realized. This process reduces human errors, improves calculation accuracy, and ensures the scientificity and accuracy of national land space planning indicators;

[0039] 2. The introduction of automated calculation and planning processes makes the entire decision-making process faster. Compared with traditional manual calculation and analysis, which takes a lot of time, the solution based on the big model can complete complex calculation and analysis tasks in a short time, further ensuring the accuracy, timeliness and reliability of the calculation of national land space planning indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A flow chart of a method for calculating national land space planning indicators based on a large language model provided by an embodiment of the present invention;

[0041] Figure 2 The form of the associated data table for storing indicator data;

[0042] Figure 3 A system structure diagram of a national land space planning indicator calculation system based on a large language model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the following, in combination with the accompanying drawings and preferred embodiments, describes in detail the specific implementation methods, structures, features and effects of a method and system for calculating national land space planning indicators based on a large language model proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0045] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0046] like Figure 1 As shown, it shows a flow chart of a method for calculating national land space planning indicators based on a large language model provided by an embodiment of the present invention, the method comprising:

[0047] Step S1, based on the natural language statement with indicator calculation task requirements input by the user, the name of the calculation tool and the actual numerical value of the input parameter of the calculation tool are derived and output through the big model, and the corresponding spatial calculation instructions are generated.

[0048] Specifically, the natural language statement records the indicator calculation task requirements. Based on this requirement, this application will first use the instruction understanding ability of the large model to analyze the most matching calculation tool in the tool library. Then, extract the relevant parameters of spatial calculation in the natural language statement, such as the calculation range, etc. Finally, these extracted parameters are input into the calculation tool and translated into the corresponding spatial calculation instructions.

[0049] Step S2, execute the spatial calculation instruction to obtain the corresponding index calculation result.

[0050] Specifically, the present application will utilize the computing power of the spatial computing engine and bring in the basic spatial data in the database. After executing the spatial computing instructions, the corresponding indicator calculation results will be returned in the form of a table.

[0051] Step S3, inputting the indicator calculation results and the corresponding first task prompt words into the large model, and obtaining the corresponding national land space planning results through the large model interpretation.

[0052] Specifically, the first task prompt is used to explain the input data format and give corresponding interpretation examples, such as which numerical ranges are sparse, suitable, and crowded for population density. It should be noted that the indicator calculation results are usually in a table format. This application will first convert the indicator calculation results in the table format into a comma-delimited text format (csv format), and then input the text into the large model for interpretation, and obtain and feedback the corresponding national land space planning results to the user.

[0053] From the above, it can be seen that the present application discloses a method for calculating national land space planning indicators based on a large language model, which realizes the automatic generation of spatial calculation instructions by deducing the names and parameters of calculation tools through a large model. This process reduces human errors, improves calculation accuracy, and ensures the scientificity and accuracy of national land space planning indicators; the introduction of automated calculation and planning processes makes the entire decision-making process faster. Compared with traditional manual calculations and analysis that take a lot of time, solutions based on large models can complete complex calculation and analysis tasks in a short time, further ensuring the accuracy, timeliness and reliability of national land space planning indicator calculations.

[0054] In one embodiment, in step S1, the natural language statement with the indicator calculation task requirement input by the user is derived and output by the big model for the name of the calculation tool and the actual value of the input parameter of the calculation tool, and the corresponding spatial calculation instruction is generated therefrom, including:

[0055] Step S11, obtaining a natural language sentence with indicator calculation task requirements input by the user.

[0056] Specifically, the text input by the user may be collected through a user interface (such as a webpage chat window).

[0057] Step S12, taking the natural language sentence, the predefined candidate computing tool list, and the second task prompt word for guiding the large language model to recommend computing tools based on the user input and the candidate computing tool list as input, passing them to the large language model, and the large model feeds back the corresponding computing tool name.

[0058] Specifically, the application will pre-organize the available candidate computing tools and their functional descriptions into a corresponding list, where each item in the list records the key-value pair of "computing tool name" and "tool functional description". After that, a second task prompt is created, which is mainly used to give instructions to the large model to guide the large model to use its own tool selection ability to select a computing tool that adapts to the user input.

[0059] Step S13, creating a corresponding tool input parameter list based on the computing tool name, and taking the natural language sentence, the tool input parameter list, and the third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input, passing them to the large language model, and the large model feeds back the corresponding actual value sequence.

[0060] Specifically, after obtaining the name of the computing tool, the application will obtain the tool input parameters and parameter description text of the computing tool, and form a corresponding tool input parameter list, wherein each item in the tool input parameter list records the key-value pair of "tool input parameter name" and "parameter description text". Afterwards, a third task prompt word is created, which is used to guide the large model to extract the actual value of each computing tool input parameter from the user input through semantic comparison based on the tool input parameter list.

[0061] Step S14, based on the calculation tool name and the actual numerical sequence, generate corresponding spatial calculation instructions through data integration and template filling.

[0062] Specifically, this application will first obtain the corresponding spatial instruction template from the preset indicator library according to the name of the calculation tool. This template defines the basic structure and logic required to perform spatial calculations. Then, according to the actual numerical sequence, the corresponding target spatial data is retrieved and loaded from the relevant spatial data warehouse. Next, data integration is performed, that is, the retrieved target spatial data is matched and integrated with the spatial instruction template to obtain the corresponding spatial calculation instructions.

[0063] In the above embodiment, based on the predefined list of candidate computing tools, it is allowed to flexibly select the most suitable computing tool according to the actual needs of the user, and through the intervention of the large language model, it is possible to automatically extract the necessary computing parameters from the user input and convert them into executable computing instructions, which not only reduces manual intervention, but also reduces the risk of human errors and improves the accuracy and reliability of data processing.

[0064] In one embodiment, in step S14, the generating of the corresponding spatial calculation instruction based on the calculation tool name and the actual numerical sequence through data integration and template filling includes:

[0065] Step S141, loading the corresponding spatial instruction template from the preset indicator library based on the calculation tool name, and loading the corresponding target spatial data from the preset range line library based on the actual numerical sequence.

[0066] Specifically, this application maintains a correspondence library (i.e., an indicator library) between computing tool names and spatial instruction templates. When the computing tool name is known, this application will retrieve the corresponding spatial instruction template from the indicator library based on the computing tool name. It should be noted that the spatial instruction template is a structural framework designed based on SQL statements, which contains the SQL structure required to perform specific spatial calculations, but the specific data input part is replaced by numbered placeholders (such as {param1}, {param2}).

[0067] Furthermore, the above-mentioned preset range line library contains the geographic boundary information of administrative districts, streets, and communities at all levels, and the retrieved target spatial data usually exists in the form of a data table, wherein each data table contains a specific level of spatial data and related attribute information.

[0068] Step S142, loading the target spatial data into the spatial instruction template to obtain corresponding spatial calculation instructions.

[0069] Specifically, the present application will load the retrieved target spatial data into the relevant placeholder position in the spatial instruction template. For example, if there is a placeholder for the data table name representing the administrative district in the spatial instruction template, then after loading, this placeholder will be replaced by the actual data table name. After this step, a complete SQL statement that can be executed can be formed, and the spatial calculation can be performed by executing the statement.

[0070] In the above embodiment, the accuracy and standardization of the instruction are ensured by loading the spatial instruction template that matches the name of the calculation tool from the preset indicator library. At the same time, the corresponding target spatial data is loaded from the preset range line library to ensure the accuracy and consistency of the data. This precise matching and loading method greatly improves the efficiency of instruction generation.

[0071] In one embodiment, in step S2, executing the spatial calculation instruction to obtain the corresponding index calculation result includes:

[0072] Step S21 , obtaining range line data, and selecting a target block that overlaps with the range line data from pre-stored global space data.

[0073] Specifically, before performing complex spatial calculations, in order to reduce the amount of calculations and increase the calculation speed, it is first necessary to select target tiles that overlap with specific range lines (i.e., the range line data) from the global spatial data. This range line data is usually pre-stored in a database and is used to define the area of ​​interest.

[0074] Step S22, cropping the target image block to obtain a cropped image block overlapping with the range line data.

[0075] Specifically, after selecting the target tiles that overlap with the range line, the next step is to cut out the cut tiles that overlap with the range line from these target tiles. It should be noted that in addition to including the geometric shape attributes after cutting, in order to facilitate the calculation of subsequent attribute values, the present application also adds a new field in the cut tile, which is mainly used to record geometric attribute information related to the target tile, such as length information and area information.

[0076] Step S23, obtaining the original attribute value of the cropped image block before cropping, and adjusting the original attribute value using a linear interpolation method to obtain a corresponding new attribute value.

[0077] Specifically, when the target tile is cropped to match the range line data, the cropped tile obtained after cropping only represents a portion of the target tile. Therefore, if the original attribute value before cropping is directly used to represent the cropped tile, the attribute value will not match the actual area size, resulting in misleading results.

[0078] To solve this problem, this application uses a linear interpolation method to adjust the original attribute value before cutting, so as to obtain a new attribute value that is suitable for the cut block. This method specifically recalculates a new attribute value based on the ratio of the area (or length) of the cut block to the area (or length) of the original block (i.e., the target block).

[0079] Step S24, summarizing the obtained new attribute values ​​to obtain corresponding index calculation results.

[0080] Specifically, this step involves summarizing all the new attribute values ​​adjusted by linear interpolation. This usually means summing or averaging the attribute values ​​of the same type or belonging to the same category. The new attribute values ​​after summarization can be used to calculate various indicators, such as total population, average traffic flow, land cover type ratio, etc. These indicators can reflect the specific characteristics or conditions of the cropped graphic area.

[0081] In the above embodiment, by obtaining the range line data and selecting the target block overlapping with the range line data, the area to be analyzed can be accurately located. In addition, by adjusting the original attribute value of the cut block using the linear interpolation method, a new attribute value matching the area (or length) of the cut block can be obtained, which helps to more accurately reflect the characteristics of the cut area.

[0082] In one of the embodiments, in step S21, the global spatial data is pre-stored in a PostGIS database in the form of points, lines, and surfaces; the calculated target blocks and the cropped blocks are stored in a file system in the form of files; and the indicator calculation results obtained by executing the spatial calculation instructions are stored in a PostgreSQL database table in the form of a table.

[0083] On the other hand, the storage of indicator data is in the form of Figure 2 As shown, it is divided into 5 data tables that are related in sequence, from right to left: indicator template table (zbk_zb_template), indicator table (zbk_zhibiao), indicator date table (zbk_zb_epoch), input layer table (zbk_zb_input_layer), and input field table (zbk_zb_input_field), among which:

[0084] 1) Indicator template table (zbk_zb_template): stores the SQL statement template (zb_template) used to calculate the indicator, the name of the statement template (zb_template_name), and the relevant description of the template (zb_template_description).

[0085] 2) Indicator table (zbk_zhibiao): stores the indicator name (zb_name), formula category / formula (zb_formula), the latest date of the indicator (latest_pointer), the referenced indicator template id (zb_template_id), and the display order (zb_order). It is specifically associated with the indicator template table (zbk_zb_template) through the indicator template id (zb_template_id).

[0086] 3) Indicator date table (zbk_zb_epoch): Since the data is constantly updated, the same data often has different versions of multiple years or dates. Indicators need to be calculated for each version of the data. Therefore, the indicator date table corresponds to the data of different dates, which mainly records the date (epoch_name), date order (epoch_order), the layer id (Map_id) corresponding to the date, and the associated indicator id (zb_id). It is specifically associated with the indicator table (zbk_zhibiao) through the indicator id (zb_id).

[0087] 4) Input layer table (zbk_zb_input_layer): Since the calculation of an indicator requires calling one or more spatial data, the input layer table (zbk_zb_input_layer) mainly records the input data id (geo_data_id), data name (geo_data_name), data order (geo_data_order), input direction (input_layer_type), and date id (epoch_id). It is specifically associated with the indicator date table (zbk_zb_epoch) through the date id (epoch_id).

[0088] 5) Input field table (zbk_zb_input_field): Since the field names and order in the layer need to be called for clear indicator calculations, this table mainly records the input field name (input_field_name), input field order (input_field_order), and layer id (input_field_id). It is specifically associated with the input layer table (zbk_zb_input_layer) through the layer id (input_field_id).

[0089] In one embodiment, in step S23, obtaining the original attribute value of the cropped image block before cropping, and adjusting the original attribute value using a linear interpolation method to obtain a corresponding new attribute value includes:

[0090] Step S231 : when it is determined that the cut block belongs to a planar element, the original attribute value is linearly adjusted based on the area ratio of the cut block to the target block to obtain a corresponding new attribute value.

[0091] Step S232: when it is determined that the cut block belongs to a linear element, the original attribute value is between the cut block and the

[0092] A linear adjustment is made based on the length ratio of the target block to obtain the corresponding new attribute value.

[0093] It should be noted based on step S231 to step S232 that when the cut block is a planar element, its attribute value is adjusted according to the area ratio of the cut block to the target block. This usually means that if the cut block is part of the target block, its attribute value will be reduced proportionally to reflect the reduction in its area ratio. For example, if the area of ​​the cut block accounts for 50% of the area of ​​the target block, some attributes (such as density, intensity, etc.) may also need to be halved accordingly. For linear elements, the attribute value is adjusted using the length ratio of the cut block to the target block. Similar to planar elements, if the cut block is part of the target block, its attribute value will be adjusted proportionally. This applies to linear areas such as roads and rivers.

[0094] In one embodiment, when it is determined that the cut block belongs to a planar element, the corresponding new attribute value is calculated as follows: new attribute value = original attribute value * (area of ​​the cut block / area of ​​the target block). When it is determined that the cut block belongs to a linear element, the corresponding new attribute value is calculated as follows: new attribute value = original attribute value * (length of the cut block / length of the target block).

[0095] In the above embodiment, by adjusting the attribute value based on the area or length ratio of the cut block to the target block, it is possible to ensure that the attributes of the cut block remain relatively accurate and reasonable. This helps to avoid inaccurate attribute values ​​during the graphic cutting or segmentation process, thereby improving the overall accuracy of the data.

[0096] Please refer to Figure 3 The present application discloses a national land space planning indicator calculation system based on a large language model, the system comprising a spatial calculation instruction generation module, a spatial calculation instruction execution module, and a national land space planning analysis module, wherein:

[0097] The spatial computing instruction generation module is used to generate corresponding spatial computing instructions based on the natural language statements with indicator computing task requirements input by the user, and to derive and output the computing tool name and the actual numerical values ​​of the computing tool input parameters through a large model.

[0098] The spatial calculation instruction execution module is used to execute the spatial calculation instruction to obtain corresponding indicator calculation results.

[0099] The national land space planning analysis module is used to input the indicator calculation results and the corresponding first task prompt words into the large model, and obtain the corresponding national land space planning results through the large model interpretation.

[0100] From the above, it can be seen that the national land space planning indicator calculation system based on the large language model disclosed in this application realizes the automatic generation of spatial calculation instructions through the derivation of the calculation tool name and parameters by the large model. This process reduces human errors, improves calculation accuracy, and ensures the scientificity and accuracy of national land space planning indicators; the introduction of automated calculation and planning processes makes the entire decision-making process faster. Compared with traditional manual calculation and analysis, which takes a lot of time, the solution based on the large model can complete complex calculation and analysis tasks in a short time, further ensuring the accuracy, timeliness and reliability of the calculation of national land space planning indicators.

[0101] In one of the embodiments, the spatial computing instruction generation module is also used to obtain a natural language sentence with indicator computing task requirements input by a user; the natural language sentence, a predefined candidate computing tool list, and a second task prompt word for guiding the large language model to recommend computing tools based on the user input and the candidate computing tool list are passed as input to the large language model, and the large model feeds back the corresponding computing tool name; a corresponding tool input parameter list is created based on the computing tool name, and the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input are passed as input to the large language model, and the large model feeds back the corresponding actual value sequence; based on the computing tool name and the actual value sequence, the corresponding spatial computing instruction is generated through data integration and template filling.

[0102] In one of the embodiments, the spatial calculation instruction generation module is also used to load the corresponding spatial instruction template from the preset indicator library based on the calculation tool name, and to load the corresponding target spatial data from the preset range line library based on the actual numerical sequence; the target spatial data is loaded into the spatial instruction template to obtain the corresponding spatial calculation instruction.

[0103] In one of the embodiments, the spatial calculation instruction execution module is also used to obtain range line data, select a target block that overlaps with the range line data from pre-stored global spatial data; crop the target block to obtain a cropped block that overlaps with the range line data; obtain the original attribute value of the cropped block before cropping, and adjust the original attribute value using a linear interpolation method to obtain a corresponding new attribute value; and summarize the obtained new attribute values ​​to obtain a corresponding indicator calculation result.

[0104] In one of the embodiments, the global spatial data is pre-stored in a PostGIS database in the form of points, lines, and surfaces; the calculated target blocks and the cropped blocks are stored in a file system in the form of files; and the indicator calculation results obtained by executing the spatial calculation instructions are stored in a PostgreSQL database table in the form of a table.

[0105] In one of the embodiments, the spatial calculation instruction generation module is also used to, when it is determined that the cut block belongs to a planar element, linearly adjust the original attribute value based on the area ratio of the cut block to the target block to obtain a corresponding new attribute value; when it is determined that the cut block belongs to a linear element, linearly adjust the original attribute value based on the length ratio of the cut block to the target block to obtain a corresponding new attribute value.

[0106] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for calculating national land space planning indicators based on a large language model, characterized in that: The method comprises: S1. Based on the natural language statements with indicator calculation task requirements input by the user, the big model derives the name of the calculation tool and the actual value of the input parameters of the calculation tool, and generates the corresponding spatial calculation instructions; S2, executing the spatial calculation instruction to obtain the corresponding index calculation result; S3, inputting the index calculation results and the corresponding first task prompt words into the large model, and obtaining the corresponding national land space planning results through the large model interpretation; The natural language statement with the indicator calculation task requirements input by the user is derived and output by the big model to the calculation tool name and the actual value of the calculation tool input parameter, and the corresponding spatial calculation instruction is generated based on this, including: S11, obtaining a natural language sentence with indicator calculation task requirements input by the user; S12, taking the natural language sentence, the predefined candidate computing tool list, and the second task prompt word for guiding the large language model to recommend computing tools according to the user input and the candidate computing tool list as input, and passing them to the large language model, and the large model feeds back the corresponding computing tool name; S13, creating a corresponding tool input parameter list based on the computing tool name, and passing the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input to the large language model, and the large model feeds back a corresponding actual value sequence; S14. Based on the calculation tool name and the actual numerical sequence, corresponding spatial calculation instructions are generated through data integration and template filling.

2. The method according to claim 1, characterized in that In step S14, the corresponding spatial calculation instructions are generated based on the calculation tool name and the actual numerical sequence through data integration and template filling, including: S141, loading a corresponding spatial instruction template from a preset indicator library based on the calculation tool name, and loading corresponding target spatial data from a preset range line library based on the actual numerical sequence; S142: Load the target spatial data into the spatial instruction template to obtain corresponding spatial calculation instructions.

3. The method according to claim 1, characterized in that In step S2, the spatial calculation instruction is executed to obtain corresponding index calculation results, including: S21, acquiring range line data, and selecting a target block overlapping with the range line data from pre-stored global space data; S22, cutting the target image block to obtain a cut image block overlapping with the range line data; S23, obtaining the original attribute value of the cropped image block before cropping, and adjusting the original attribute value by using a linear interpolation method to obtain a corresponding new attribute value; S24. Summarize the new attribute values ​​obtained to obtain corresponding index calculation results.

4. The method according to claim 3, characterized in that In step S21, the global spatial data is stored in the PostGIS database in advance in the form of points, lines, and surfaces; The target image block and the cut image block obtained by calculation are stored in a file system in the form of files; The index calculation results obtained by executing the spatial calculation instructions are stored in a PostgreSQL database table in a tabular form.

5. The method according to claim 3, characterized in that: In step S23, the original attribute value of the cropped image block before cropping is obtained, and the original attribute value is adjusted by using a linear interpolation method to obtain a corresponding new attribute value, including: S231, when it is determined that the cut block belongs to a planar element, the original attribute value is linearly adjusted based on the area ratio of the cut block to the target block to obtain a corresponding new attribute value; S232: When it is determined that the cut block belongs to a linear element, the original attribute value is A linear adjustment is performed based on the length ratio of the blocks to obtain the corresponding new attribute value.

6. A national land space planning indicator calculation system based on a large language model, characterized in that: The system includes a spatial computing instruction generation module, a spatial computing instruction execution module, and a national land space planning analysis module, wherein: The spatial computing instruction generation module is used to generate corresponding spatial computing instructions based on the natural language statement with index computing task requirements input by the user, and to derive and output the computing tool name and the actual value of the computing tool input parameter through the large model; The spatial calculation instruction execution module is used to execute the spatial calculation instruction to obtain the corresponding index calculation result; The national land space planning analysis module is used to input the index calculation results and the corresponding first task prompt words into the large model, and obtain the corresponding national land space planning results through the large model interpretation; The spatial computing instruction generation module generates corresponding spatial computing instructions based on the natural language statement with the indicator computing task requirements input by the user, and the derivation output of the computing tool name and the actual value of the computing tool input parameter through the big model, and thus generates the corresponding spatial computing instructions, including: Obtaining natural language sentences with indicator calculation task requirements input by users; The natural language sentence, the predefined candidate computing tool list, and the second task prompt word for guiding the large language model to recommend computing tools according to the user input and the candidate computing tool list are passed as input to the large language model, and the large model feeds back the corresponding computing tool name; Creating a corresponding tool input parameter list based on the computing tool name, and taking the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input, and passing them to the large language model, and the large model feeds back a corresponding actual value sequence; Based on the calculation tool name and the actual numerical sequence, corresponding spatial calculation instructions are generated through data integration and template filling.

7. The system according to claim 6, characterized in that The spatial computing instruction generation module is also used to obtain a natural language sentence with an indicator computing task requirement input by a user; take the natural language sentence, a predefined candidate computing tool list, and a second task prompt word for guiding the large language model to recommend computing tools according to the user input and the candidate computing tool list as input, pass them to the large language model, and the large model feeds back the corresponding computing tool name; create a corresponding tool input parameter list based on the computing tool name, and take the natural language sentence, the tool input parameter list, and a third task prompt word for guiding the large language model to extract the actual value of each computing tool input parameter from the user input as input, pass them to the large language model, and the large model feeds back the corresponding actual value sequence; Based on the calculation tool name and the actual numerical sequence, corresponding spatial calculation instructions are generated through data integration and template filling.

8. The system according to claim 7, characterized in that The spatial calculation instruction generation module is further used to load the corresponding spatial instruction template from the preset indicator library based on the calculation tool name, and load the corresponding target spatial data from the preset range line library based on the actual numerical sequence; The target spatial data is loaded into the spatial instruction template to obtain corresponding spatial calculation instructions.

9. The system according to claim 6, characterized in that The spatial calculation instruction execution module is also used to obtain range line data, select a target block that overlaps with the range line data from the pre-stored global spatial data; crop the target block to obtain a cropped block that overlaps with the range line data; obtain the original attribute value of the cropped block before cropping, and use a linear interpolation method to adjust the original attribute value to obtain a corresponding new attribute value; and summarize the obtained new attribute values ​​to obtain a corresponding indicator calculation result.

Images