Cross-scale planning key point generation method and system based on NLP and knowledge graph reasoning

Abstract

The invention relates to the technical field of urban planning management, and particularly provides a cross-scale planning key point generation method and system based on NLP and knowledge graph reasoning, and the method comprises the steps: receiving and analyzing a natural language demand of a user, extracting entity information and an instruction intention, and converting the entity information and the instruction intention into a structured instruction; constructing a planning domain knowledge graph containing a planning index constraint relation, a cross-scale planning association rule and a cross-professional parameter mapping relation; extracting a target planning index from the structured instruction, and calling knowledge graph association reasoning to generate an index association set containing association indexes and an adjustment range; performing collaboration verification on the index association set, including cross-scale linkage verification based on upper planning data and GIS spatial analysis and adaptation verification based on knowledge graph cross-professional parameter mapping; and generating a structured planning key point report according to the index association set passing the verification. The method improves the planning key point generation efficiency and compliance, and is suitable for the scenes of urban planning, land development and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban planning management, in particular to a cross-scale planning point generation method and system based on NLP and knowledge graph reasoning. BACKGROUND

[0002] In the field of urban planning management, planning points, as the core basis for guiding plot development and construction and issuing "construction land planning permits", directly affect the quality and progress of planning management. With the development of artificial intelligence and geographic information technology, planning point generation technology has gradually evolved towards intelligentization from the traditional "manual query of plot information, review of regulations and files, and manual report writing" mode. Currently, existing technologies have achieved some basic intelligent functions: for example, through natural language processing (NLP) technology, key information such as plot number and land use nature is automatically extracted from user input plot requirements; relying on planning databases and basis libraries, planning point reports containing indicators such as plot ratio and building height are generated according to fixed rules; some solutions also introduce GIS spatial analysis capabilities to support plot range selection, drawing and basic information extraction, which to some extent reduces manual operation costs and shortens the generation cycle of single plot planning points. At the same time, a few technical solutions attempt to introduce simple association logic, such as presetting index ranges according to land use nature, or supporting online annotation collaboration among multiple professionals, further improving the convenience of planning point generation.

[0003] Although existing technologies have achieved preliminary intelligentization of planning point generation, they still have significant defects in dealing with complex planning scenarios and meeting cross-dimensional collaboration needs, making it difficult to support efficient and refined planning management: first, index generation relies on static rule matching, which cannot dynamically infer the association between indexes (such as the interlocking constraints of plot ratio adjustment and parking spaces, green space ratio), making it easy to cause index isolation or conflict; second, there is a lack of cross-scale collaboration (no linkage with district / city upper-level planning) and cross-professional parameter verification (such as no automatic association between building and municipal parameters), relying on manual coordination; third, the knowledge system is fragmented, and rules and basis are not structurally associated, making it difficult to support complex decision-making. These defects directly cause low efficiency of planning point generation, high risk of compliance omissions caused by manual intervention, and high cost of cross-scale / cross-professional communication.

[0004] Therefore, there is an urgent need for a technical solution with dynamic reasoning and deep collaboration capabilities to overcome existing bottlenecks. SUMMARY

[0005] In order to overcome the deficiencies of the prior art, the present application provides a cross-scale planning point generation method and system based on NLP and knowledge graph reasoning to solve the problems in the prior art.

[0006] One of the embodiments of the present application provides a cross-scale planning key point generation method based on NLP and knowledge graph reasoning, comprising the following steps: S10, receiving and analyzing the natural language requirements input by the user, extracting the core requirement information of the entity information and instruction intention in the natural language requirements, and converting the core requirement information into structured instructions executable by the system; S20, constructing a planning domain knowledge graph, wherein the planning domain knowledge graph comprises constraint relationship information between planning indicators, correlation rules of cross-scale planning, and mapping relationship of cross-professional parameters; S30, extracting the target planning indicator from the structured instruction, and based on the structured instruction, calling the planning domain knowledge graph to perform associated reasoning on the target planning indicator, and generating an indicator association set comprising associated indicators and adjustment ranges; S40, performing a synergistic check on the indicator association set, wherein the synergistic check comprises a cross-scale linkage check and a cross-professional parameter adaptation check, wherein the cross-scale linkage check is based on upper-level planning data and performs a check on the indicator association set through GIS spatial analysis, and the cross-professional parameter adaptation check performs an adaptive check on the indicator association set based on the mapping relationship of cross-professional parameters in the planning domain knowledge graph; S50, generating a structured planning key point report according to the indicator association set that passes the synergistic check.

[0007] In one of the embodiments, step S10 specifically comprises the following steps: S101, receiving the natural language requirements input by the user, performing word segmentation and part-of-speech tagging on the text of the natural language requirements using a pre-trained BERT model, and preliminarily identifying candidate entities contained in the text, wherein the candidate entities at least include plot number, land use property, and planning indicators; S102, calling a customized entity vocabulary in the field of urban planning to preprocess the candidate entities, wherein the customized entity vocabulary in the field of urban planning at least includes plot number format rules, land use property standard terminology library, and planning indicator name library; wherein the preprocessing comprises: eliminating non-planning domain entities in the candidate entities; correcting non-standard terms associated with land use properties in the candidate entities to corresponding standard terms in the land use property standard terminology library; checking the plot numbers in the candidate entities according to the plot number format rules, and retaining plot numbers that conform to the rules; S103, calling a pre-trained intent determination model to analyze the pre-processed candidate entity association relationship, determining that the user intent is one of a newly generated planning point, an existing planning point adjustment, or a planning indicator compliance query, matching the user intent with a preset standard instruction template, and generating a structured instruction executable by the system.

[0008] In one of the embodiments, after step S30, further comprising the following steps: S301, calling a preset historical planning case library, the historical planning case library storing planning cases in a preset time range, same region, and same land use property, the planning case information at least including plot area, core planning indicators, and approval results; S302, taking the land use property and plot area of the target plot in the index association set as the screening conditions, screening a plurality of similar cases with a similarity to the target plot within a preset similarity range from the historical planning case library; S303, calculating the average value and fluctuation range of the core planning indicators in the similar cases, the core planning indicators at least including the volume rate and the building height, and being consistent with the core planning indicator types of the target plot in the index association set; S304, comparing the core planning indicators of the target plot in the index association set with the fluctuation range of the core planning indicators in the similar cases; S305, based on the comparison result, if the core planning indicators of the target plot exceed the fluctuation range, re-calling the planning domain knowledge graph to check whether the regional development restriction clauses in the correlation rules of cross-scale planning are missed, and adjusting the core planning indicators of the target plot to a reasonable range based on the checking result; if the fluctuation range is not exceeded, the original index association set is retained.

[0009] In one of the embodiments, in the cross-scale linkage verification of step S40, further comprising a special verification of the index association set for the rail transit safety protection zone, specifically comprising the following steps: S411, calling a preset rail transit control area database, extracting the boundary coordinates of the target plot in the index association set, judging whether the target plot is located within the rail transit operation safety protection zone range based on the rail line and station location data in the rail transit control area database; S412, if the target plot is located within the rail transit operation safety protection zone range, extracting the rail safety protection parameters from the correlation rules of cross-scale planning in the planning domain knowledge graph, and verifying whether the plot vertical control indicators in the index association set meet the requirements based on the rail safety protection parameters; S413, if the plot vertical control indicators in the index association set do not meet the requirements of the track safety protection parameters, generate targeted optimization suggestions and update to the index association set; if they meet the requirements, mark the track safety compliance in the collaborative verification result.

[0010] In one of the embodiments, the cross-scale linkage verification in step S40 also includes special verification of ecological sensitive areas of the index association set, specifically including the following steps: S421, call a preset ecological data resource package, the ecological data resource package including a structured ecological sensitive area GIS database and an unstructured regional ecological management and control text; use a pre-trained BERT model to extract the implicit ecological constraint rules of the unstructured regional ecological management and control text and convert them into structured implicit ecological constraint rules; S422, fuse the structured implicit ecological constraint rules, the boundary data in the ecological sensitive area GIS database, and the ecological rule sub-module contained in the cross-scale planning associated rules in the planning domain knowledge graph to infer and generate a unique ecological verification parameter set for the target plot in the index association set, wherein the ecological rule sub-module stores ecological management and control rules of different levels and preset priorities, and if there is a rule conflict, it is automatically resolved according to the preset priorities; S423, based on the unique ecological verification parameter set, perform double verification on the ecological sensitive type indicators in the index association set, specifically including: Verify whether the target plot boundary exceeds the management and control range in the ecological sensitive area GIS database; Verify whether the ecological sensitive type indicators violate the structured implicit ecological constraint rules; S424, if any of the verification items does not meet the requirements, call the planning domain knowledge graph again, generate targeted optimization suggestions based on the cross-scale planning associated rules and update to the index association set; if all meet the requirements, mark the ecological sensitive area compliance in the collaborative verification result.

[0011] In one of the embodiments, in the step of collaborative verification of the index association set, the specific steps of the cross-scale linkage verification based on the upper planning data and through GIS spatial analysis to verify the index association set include: S431, call a preset upper planning database, extract the upper planning data corresponding to the planning jurisdiction area to which the target plot in the index association set belongs, the upper planning data at least including upper planning spatial management and control range and upper planning index constraint; S432, perform spatial overlay analysis on the boundary of the target plot in the index association set and the spatial control range of the upper planning by a GIS spatial analysis tool, to determine whether the target plot is completely located within the spatial range allowed for development by the upper planning, or whether it is partially overlapped with the forbidden construction area / limited construction area; S433, if the target plot is completely located within the allowed development range, check whether the core planning index in the index association set meets the threshold requirement of the upper planning index based on the upper planning index constraint; if the target plot is partially overlapped with the forbidden construction area / limited construction area, directly determine that the development type index in the index association set related to the overlapped area does not meet the requirement; S434, if the core planning index meets the threshold requirement of the upper planning index, mark the upper planning compliance in the coordination verification result; if the core planning index does not meet the threshold requirement of the upper planning index, generate a spatial range adjustment suggestion or an index reduction suggestion based on the upper planning data, and update it to the index association set.

[0012] In one of the embodiments, in the step of performing coordination verification on the index association set, the specific steps of the cross-professional parameter adaptation verification based on the mapping relationship of the cross-professional parameters in the planning domain knowledge graph include: S441, call the planning domain knowledge graph and extract the mapping relationship of the cross-professional parameters, the mapping relationship of the cross-professional parameters at least including parameter mapping of building specialty and municipal specialty, parameter mapping of building specialty and landscape specialty; S442, filter out the cross-professional associated index group from the index association set, the cross-professional associated index group at least containing planning indexes of two different specialties; S443, based on the mapping relationship of the cross-professional parameters, perform adaptive verification on each cross-professional associated index group, including: checking whether the pre-sequence index triggers the constraint threshold of the post-sequence index; checking whether there is an implicit conflict in the parameters in the index group; S444, if the verification meets the mapping relationship requirement of the cross-professional parameters, mark the cross-professional parameter adaptation in the coordination verification result; if there is a mismatch, call the knowledge graph reasoning to generate a targeted optimization suggestion and update it to the index association set.

[0013] The application also relates to a cross-scale planning point generation system based on NLP and knowledge graph reasoning, comprising: a demand extraction module, configured to receive and parse the natural language demand input by a user, extract core demand information containing entity information and instruction intention in the natural language demand, and convert the core demand information into a structured instruction executable by the system; The knowledge graph construction module is configured to construct a planning domain knowledge graph, which contains constraint relationships between planning indexes, correlation rule information of cross-scale planning, and mapping relationships of cross-professional parameters. The correlation reasoning module is configured to extract a target planning index from the structured instruction, perform correlation reasoning on the target planning index based on the planning domain knowledge graph, and generate an index correlation set containing correlated indexes and adjustment ranges. The collaborative verification module is configured to perform collaborative verification on the index correlation set, which includes cross-scale linkage verification and cross-professional parameter adaptation verification. The cross-scale linkage verification is based on upper-level planning data and performs verification on the index correlation set through GIS spatial analysis. The cross-professional parameter adaptation verification is based on the mapping relationships of cross-professional parameters in the planning domain knowledge graph to perform adaptability verification on the index correlation set. The report generation module is configured to generate a structured planning point report according to the index correlation set that passes the collaborative verification.

[0014] The present application also relates to a computer device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for generating cross-scale planning points based on NLP and knowledge graph reasoning are implemented.

[0015] The present application also relates to a computer-readable storage medium, which stores a computer program. When the processor executes the computer program, the steps of the above-mentioned method for generating cross-scale planning points based on NLP and knowledge graph reasoning are implemented.

[0016] The above-mentioned method and system for generating cross-scale planning points based on NLP and knowledge graph reasoning have the following advantages: By adopting the scheme, the NLP technology can be used to automatically analyze user natural language requirements and convert them into structured instructions, avoiding the cumbersome operation of traditional manual manual sorting requirements, and significantly improving the requirement processing efficiency; meanwhile, relying on the planning domain knowledge graph containing planning index constraint relationship, cross-scale planning association rule and cross-professional parameter mapping relationship, the index association set containing associated indexes and adjustment range can be automatically associated and inferred, solving the problem of isolated indexes and lack of linkage in traditional planning; through the cross-scale linkage verification of "upper planning data + GIS spatial analysis" and the adaptation verification of "knowledge graph cross-professional parameter mapping", a collaborative mechanism is formed, which effectively avoids cross-level planning conflicts and cross-professional parameter contradictions, and guarantees the compliance and accuracy of planning indexes; finally, the generated structured planning point report can also provide clear and traceable planning basis for urban planning, land development and other scenes, and the intelligentization, high efficiency and collaboration of planning point generation are realized, and the labor cost and communication cost are greatly reduced. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of a cross-scale planning point generation method based on NLP and knowledge graph reasoning provided by an embodiment of the present application is provided. Figure 2 A principle block diagram of a computer device provided by an embodiment of the present application is provided. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings.

[0019] With reference to Figure 1 An embodiment of the present application provides a cross-scale planning point generation method based on NLP and knowledge graph reasoning, which comprises the following steps: S10, receiving and analyzing user input natural language requirements, extracting core requirement information containing entity information and instruction intention in the natural language requirements, and converting the core requirement information into structured instructions executable by the system; In the embodiment, step S10 is a key starting link of the method for receiving user requirements and starting the subsequent process of the system, and the core function is to complete the conversion from "user unstructured natural language requirements" to "system executable structured instructions", which specifically includes three progressive actions: The first is to receive and analyze the natural language requirements input by the user. The natural language requirements refer to the planning-related demands put forward by the user in daily communication language, without following a fixed format, for example: the user inputs "help me generate a planning point report for 09-05-04 land in a certain area, focusing on plot ratio, building height and public supporting facility requirements", "adjust the plot ratio of 07-02-01 land from 2.5 to 2.0 and check whether it meets the upper planning"; analysis refers to the preliminary disassembly of such requirements at the semantic level using natural language analysis algorithms (such as semantic coding, sequence labeling), for example: for the input "help me generate a planning point report for 09-05-04 land in a certain area, focusing on plot ratio, building height and public supporting facility requirements", the algorithm disassembles the key semantic units such as certain area 09-05-04 land, planning point report, plot ratio, building height, and public supporting facility, laying the foundation for subsequent extraction of key information; The second is to extract core demand information, which clearly contains two key contents of entity information and instruction intention. Entity information refers to specific elements of the planning field involved in the demand, such as certain area 09-05-04 land, plot ratio, building height, and 07-02-01 land, 2.5, 2.0, etc. in the above examples; instruction intention refers to the core demand direction of the user, such as extracting the intention of "generating a planning point report" from "generating a planning point report for 09-05-04 land in a certain area", and extracting the intention of "adjusting planning indicators and checking compliance" from "adjusting the plot ratio of 07-02-01 land from 2.5 to 2.0 and checking whether it meets the upper planning"; The third is to convert the extracted core demand information into structured instructions. Structured instructions refer to instructions that conform to the system's preset data format and can be directly called by the subsequent knowledge graph reasoning module, for example, converting "generate a planning point report for 09-05-04 land in a certain area, focusing on plot ratio, building height and public supporting facility requirements" into a standardized format of "target land: certain area 09-05-04; instruction type: generate a planning point report; core indicators: plot ratio, building height, public supporting facility", ensuring that the system can accurately identify user requirements and start the corresponding process.

[0020] S20, construct a planning domain knowledge graph, which contains constraint relationship between planning indicators, correlation rules of cross-scale planning and mapping relationship of cross-professional parameters. In this embodiment, step S20 is the core data and rule support link for realizing intelligent reasoning and collaborative verification of the method, and the core function is to build a domain knowledge graph (i.e., a structured knowledge base containing "constraint relationships between planning indicators, correlation rules for cross-scale planning, and mapping relationships for cross-professional parameters") that adapts to the planning scene, to provide accurate rule support for subsequent correlation reasoning of target planning indicators, and to achieve the whole logic as follows through the graph construction process and collaborative implementation of the core correlation relationship: The construction of the planning domain knowledge graph needs to be gradually implemented from data collection to structured storage, and finally form a semantic network that can carry three types of core correlation relationships. The specific process is as follows: First, data source integration is the basis for graph construction, which needs to collect multi-dimensional planning field data that can extract three types of correlation relationships, including but not limited to planning regulations (such as regulations for regulating urban land layout, local urban planning technical standards, for extracting constraint relationships between planning indicators), planning texts at various levels (such as urban master planning, district control planning, for extracting correlation rules for cross-scale planning), historical planning project approval archives (such as approved plot planning point reports, for supplementing indicator constraints and cross-professional mapping cases), professional technical specifications (such as building fire protection specifications, municipal pipeline design specifications, for extracting mapping relationships for cross-professional parameters), etc. These data can be obtained by technical personnel in the field through online open data platforms, official websites of relevant departments in charge of the industry, professional data service platforms, etc.

[0021] Second, knowledge extraction converts the collected data into the "node-edge-attribute" structure required by the graph, and different types of data use corresponding technologies: for text data (such as regulations and planning texts), natural language processing techniques (such as named entity recognition and relationship extraction models) are used to extract core elements such as "planning indicators (such as plot ratio)", "constraint conditions (such as ≥2.0)", "associated professions (such as architecture and municipal engineering)", and "planning levels (such as plot level and district level)", and then to extract "constraint relationships between planning indicators", "correlation rules for cross-scale planning", and "mapping relationships for cross-professional parameters"; for table data (such as historical project indicator tables and professional parameter correspondence tables), rule engines are used to directly match the relationship between "indicator name-value range-associated basis-corresponding profession / planning level", and convert it into a structured unit in the graph (such as the "plot ratio" node and the "green ratio" node connected by a "constraint relationship" edge, with the edge attribute marked as "plot ratio ≥3.0, green ratio ≥25%").

[0022] Third, knowledge fusion is used to eliminate data conflicts and ensure the consistency of the three types of relationships: if there are differences in the same type of relationship in different data (e.g., "commercial land green space rate lower limit 25%" in A regulations, "commercial land green space rate lower limit 20%" in B regulations), the rules are unified according to the principle of "superior regulations taking precedence over subordinate regulations, and the latest published regulations taking precedence over old regulations"; if there are different expressions of the same indicator or relationship (e.g., "building coverage rate" and "building density" pointing to the same concept, "hierarchical connection rules" and "cross-scale correlation rules" expressing different concepts), they are standardized into unified names (e.g., "building density", "cross-scale planning correlation rules") through a planning domain vocabulary to avoid semantic ambiguity in the graph.

[0023] Fourth, knowledge storage is achieved by using a graph database (e.g., a professional database suitable for storing node-relationship structures) to structure the storage of the fused "nodes (planning indicators, professional types, planning levels, etc.)", "edges (constraint relationships, correlation rules, mapping relationships, etc.)", and "attributes (numerical ranges, basis clauses, applicable scenarios, etc.)", for example, the "volume rate" node and the "parking space number" node are connected by the "cross-professional mapping relationship" edge, and the edge attributes are marked as "commercial land volume rate 2.0 corresponds to parking space ≥ 1.2 per hundred square meters", "according to the relevant clauses of the planning technical specification", ultimately forming a complete planning domain knowledge graph, which supports the subsequent step of "calling corresponding three types of correlation relationships according to land attributes + target indicators".

[0024] The knowledge graph constructed through the above four-step process core carries the following three types of correlation relationships, which are also the key rule sources for subsequent reasoning and verification: Constraint relationships between planning indicators: refer to the logical rules that restrict each other between different planning indicators, such as "when the volume rate of a commercial land is ≥ 3.0, the green space rate must be ≥ 25%", "when the building height exceeds 100 meters, a fire climbing surface must be provided"; Correlation rules for cross-scale planning: refer to the rules for the connection of indicators between different planning levels (e.g., land parcel level, district level, city level), such as "when the total volume rate in a certain area of a certain city district control regulation is limited to 5.0, the volume rate of a single land parcel in that area must not exceed this total limit", "the building height of a land parcel must comply with the requirement of 'building height along the water corridor ≤ 80 meters' in the city design of the district"; Mapping relationships between cross-professional parameters: refer to the corresponding rules for the matching of parameters between different planning professions (e.g., architecture, municipal, landscape), such as "when the architecture profession determines the volume rate of a residential land parcel to be 2.0, the municipal profession must provide a parking space number ≥ 1.2 per hundred square meters", "when the landscape profession requires a green space rate ≥ 30%, the building density of the architecture profession must be ≤ 40%".

[0025] It should be noted that, in order to further improve the accuracy of subsequent reasoning and the comprehensiveness of verification, the planning domain knowledge graph constructed in this step S20 includes, in addition to the core "constraint relationship between planning indicators, correlation rules of cross-scale planning, and mapping relationship of cross-specialty parameters", the following: Historical planning case library: refers to a structured database storing key data of past planning projects of the same type and region, the data being derived from historical approval archives of planning authorities (such as approved plot planning points, planning scheme review opinions), containing fields such as "plot attribute (land use nature, region)-core planning indicator (plot ratio, building height, etc.)-approval result-adaptation scenario"; in the knowledge graph, this library is mainly used to provide "case reference basis" for correlation reasoning, for example, when reasoning the plot ratio range of a commercial plot, the plot ratio cases of the same region in the library can be called to assist in calibrating the rationality of the indicator adjustment range; Rail transit control area database: refers to a spatial database storing the safety control range of urban rail transit lines and corresponding parameters, the data being officially published by rail transit authorities (such as metro groups, planning and natural resources bureaus), containing information such as "rail line name-control area type (safety protection area, planning control area)-boundary coordinates-control parameters (such as underground engineering setback distance ≥ 5 meters, ground building height limit)"; in the knowledge graph, this database is linked with "correlation rules of cross-scale planning" to "quickly match whether the plot is located within the rail control area" during subsequent cross-scale verification, and to call corresponding control parameters to constrain the indicator range; Ecological data resource package: contains two types of core data, one is ecological sensitive area spatial data (such as natural protection area, waterfront protection zone, and boundary GIS data of ecological control area, derived from spatial planning results published by ecological environment departments), and the other is ecological control text rules (such as the clause "no new commercial buildings in ecological sensitive areas" in the Regional Ecological Protection Planning, derived from official ecological regulations); in the knowledge graph, this resource package is combined with "constraint relationship between planning indicators" to "extract implicit ecological constraints" during subsequent verification, for example, when a plot is located within a waterfront protection zone, the rule "building height in waterfront area ≤ 60 meters" in the package can be called to supplement the indicator verification dimension; Upper planning database: refers to the database storing the planning results of the upper level of the target land plot (such as the district level, the city level), and the data is derived from the upper planning text approved by the planning authority (such as the district regulatory detailed planning, the urban master plan), which contains "upper planning level - spatial control range (no construction area, limited construction area boundary) - total index constraint (upper limit of total plot ratio, lower limit of green space ratio) - index connection requirement" and the like; in the knowledge graph, this database is the core data support of the "cross-scale planning correlation rule", and in the subsequent cross-scale linkage verification, the index constraint of the corresponding upper planning in the library can be directly called to judge whether the plot index meets the level connection requirement; The above four types of data resources are maintained through official channels and regular updates (such as quarterly synchronization of the latest data from planning, ecological, and rail transit departments), and work together with the three types of core association relationships of the knowledge graph to form a rule + data driven planning field knowledge system, providing complete knowledge support for the reasoning and verification of subsequent steps.

[0026] It should be emphasized that the collection, selection, format standardization and other integration and arrangement operations of the above data are all conventional data processing methods familiar to planning field technical personnel (such as obtaining regulatory texts from official platforms and structuring historical approval archives), and these basic methods are common knowledge in the field. The present application does not improve the specific technical means of data integration and arrangement, which is clear to those skilled in the art.

[0027] S30, extracting target planning indexes from the structured instructions, calling the planning field knowledge graph based on the structured instructions to associate and reason the target planning indexes, and generating an index association set containing associated indexes and adjustment ranges; In this embodiment, step S30 is the core reasoning link of the method from "demand instructions" to "index generation", and the core function is to complete the extraction and association of target planning indexes based on structured instructions and the planning field knowledge graph, to provide a complete index system for subsequent collaborative verification, which specifically includes three progressive actions of extracting target planning indexes, associated reasoning, and generating an index association set: First, the target planning indexes are extracted from the structured instructions. The target planning indexes refer to the core planning parameters that the user focuses on or needs to generate in the user demand, which are selected and obtained from the structured instructions generated in step S10. For example, if the structured instruction is "target plot: certain district 09-05-04 (commercial land); instruction type: generate planning points report; core indexes: plot ratio, building height", then "plot ratio" and "building height" are extracted as the target planning indexes for this reasoning. The extraction process needs to accurately match the content of the "core index" field in the instruction to ensure that the user's core demand is focused on; Second, based on the structured instruction call planning domain knowledge graph to associate and reason the target planning index, "call knowledge graph" according to the target land attribute (such as land use property, the area where it is located) in the structured instruction and the target planning index, trigger the corresponding association in the knowledge graph constructed in step S20, for example, when taking "certain certain area 09-05-04 commercial land plot ratio" as the target planning index, call the knowledge graph "constraint relationship between planning indexes" (such as commercial land plot ratio ≥ 2.0, green space ratio needs to be ≥ 25%), "cross-scale planning association rule" (such as the upper limit of the total plot ratio of the area where the land is located is 5.0, the plot ratio of a single land cannot exceed this value), "cross-professional parameter mapping relationship" (such as plot ratio 2.5 corresponds to municipal professional parking space ≥ 1.2 / m2), and other indexes related to the target planning index are derived through semantic reasoning; Third, generate an index association set containing associated indexes and adjustment ranges. The index association set refers to a structured set integrating the target planning index, the associated index derived by reasoning, and the reasonable adjustment range of each index. For example, for the above-mentioned commercial land, the generated set may be "target planning index: plot ratio (2.0-3.5); associated index: green space ratio (25%-30%), number of parking spaces (≥ 32), building height (≤ 80 meters)". The "adjustment range" is determined based on the constraint rules between planning indexes and the cross-scale planning association rules in the knowledge graph, ensuring that each index meets the professional specifications and level requirements, and providing a complete and checkable index basis for the subsequent step S40 of collaborative verification.

[0028] S40, collaborative verification of the index association set, the collaborative verification includes cross-scale linkage verification and cross-professional parameter adaptation verification, wherein the cross-scale linkage verification is based on upper planning data and verifies the index association set through GIS spatial analysis, and the cross-professional parameter adaptation verification is based on the mapping relationship of cross-professional parameters in the planning domain knowledge graph to adaptively verify the index association set; In this embodiment, step S40 is the core verification link of the method to ensure the compliance and collaboration of the index association set. The core function is to perform multi-dimensional compliance check on the index association set generated in step S30, avoid cross-level conflicts or cross-professional contradictions, and provide a qualified index basis for finally generating a planning point report. Specifically, it includes three key parts of clearly defining the overall scope of collaborative verification, implementing cross-scale linkage verification, and implementing cross-professional parameter adaptation verification: First, clearly define the overall scope of collaborative verification, that is, through "cross-scale linkage verification" and "cross-professional parameter adaptation verification", cover "planning level connection" and "professional parameter matching" two core compliance dimensions, ensure that the index association set meets the requirements of different planning levels and satisfies the logic of multi-professional collaboration; Secondly, cross-scale linkage verification is implemented, which is based on higher-level planning data and carried out through GIS spatial analysis. The "higher-level planning data" refers to planning file data at a higher level than the target plot (such as the regulatory detailed planning of a certain area or the overall planning of a certain city), which contains the upper limit of the total plot ratio of the area, the spatial control range (the boundary of the prohibited construction area and the restricted construction area), etc. The "GIS spatial analysis" refers to the spatial relationship judgment carried out with the aid of geographic information system technology (such as superimposed analysis of the boundary of the target plot and the spatial control range of the higher-level planning). For example, if the higher-level planning of a certain area stipulates that "a certain area is a restricted construction area, and the plot ratio should not exceed 2.5", then through GIS analysis, it is determined whether the target plot is located in the restricted construction area. If it is, then it is checked whether the "plot ratio ≤ 2.5" in the verification index association set is true. If it is not, then the verification is carried out according to the upper-level index constraint of the area where the plot is located. Thirdly, cross-professional parameter adaptation verification is implemented, which is based on the "mapping relationship of cross-professional parameters" in the planning domain knowledge graph constructed in step S20, i.e. checking whether the parameters of different professions in the index association set meet the pre-set matching rules. For example, if the "building professional plot ratio = 2.0 (commercial land)" in the index association set, then the mapping relationship between "building - municipal" in the knowledge graph is called (such as plot ratio 2.0 corresponding to parking space ≥ 1.2 per hundred square meters), and it is checked whether the "municipal professional parking space quantity" in the verification set meets the requirement. If the "landscape professional green space ratio = 28%" in the set, then the mapping relationship between "building - landscape" is called (such as green space ratio 28% corresponding to building density ≤ 40%), and it is checked whether the "building professional building density" is adapted to ensure that there is no implicit contradiction between multi-professional parameters.

[0029] S50, generating a structured planning point report according to the index association set that passes the collaborative verification.

[0030] In this embodiment, step S50 is the final output link of the method from "qualified index system" to "available planning results". The core function is to convert the index association set that passes the collaborative verification in step S40 into a standardized report that meets the needs of planning management scenarios, providing users with clear and implementable planning basis. Specifically, it includes three progressive actions: obtaining a qualified index set, constructing a structured report framework, and integrating index and related information to generate a report. Firstly, the qualified index set is obtained, i.e. extracting the index association set that passes the cross-scale linkage verification and cross-professional parameter adaptation verification in step S40, excluding the indexes that do not pass the verification or need to be adjusted, to ensure that the indexes on which the report is based meet the planning specifications and collaborative requirements. For example, if it is determined through verification that "plot ratio (2.2-2.8), building height (≤75 meters), green space ratio (26%-30%), and parking space quantity (≥35)" are all compliant in a certain index association set, then this set is used as the basic data for report generation. Second, a structured report framework is constructed. Structured means that the report needs to be organized according to the logical modules commonly used in the planning industry, rather than unordered text. The framework needs to cover four core modules: land use basic information, core planning indicators, associated planning indicators, and basis traceability description. Among them, land use basic information corresponds to the plot attributes in the structured instruction (such as target plot: 09-05-04 in a certain area, land use property: commercial land, land area: 6800 square meters), core planning indicators correspond to the target planning indicators and adjustment range extracted in step S30 (such as plot ratio 2.2-2.8, building height ≤75 meters), associated planning indicators correspond to the associated indicators and range inferred (such as green space ratio 26%-30%, number of parking spaces ≥35), and basis traceability description briefly marks the compliance basis type of each indicator (such as "plot ratio range is subject to the total plot ratio constraint in the upper-level planning of a certain area", "number of parking spaces is subject to the building-municipal parameter mapping relationship in the knowledge graph"); Third, the indicators and associated information are integrated to generate a report. That is, the data in the qualified indicator set is filled in according to the structured framework to ensure that the content of each module is complete and the data is accurate. For example, in the "core planning indicators" module, it is clearly marked that "plot ratio: 2.2-2.8 (basis: total plot ratio upper limit 5.0 in a certain area, single plot plot ratio not exceeding 3.0, combined with the characteristics of commercial land to determine)", and in the "associated planning indicators" module, it is noted that "green space ratio: 26%-30% (associated basis: plot ratio 2.2-2.8 triggers planning indicator constraint relationship, green space ratio needs to be ≥25%, combined with landscape professional requirements to determine the upper limit)". Finally, a standardized document that can be directly used for planning approval and design reference is generated, completing the closed loop of the entire planning point generation process.

[0031] In one of the embodiments, in step S10, the following steps are specifically included: S101, receiving a natural language requirement input by a user, performing word segmentation and part-of-speech tagging on the text of the natural language requirement by using a pre-trained BERT model, and preliminarily identifying candidate entities contained in the text, wherein the candidate entities at least include a plot number, a land use property, and a planning indicator; S102, calling a customized entity vocabulary in the field of urban planning to preprocess the candidate entities, wherein the customized entity vocabulary in the field of urban planning at least includes a plot number format rule, a land use property standard term library, and a planning indicator name library; wherein the preprocessing includes: eliminating non-planning field entities in the candidate entities; correcting non-standard terms associated with the land use property in the candidate entities to corresponding standard terms in the land use property standard term library; verifying the plot number in the candidate entities according to the plot number format rule, and retaining the plot number that conforms to the rule; S103, calling a pre-trained intent determination model to analyze the pre-processed candidate entity association relationship, determining that the user intent is one of newly generated planning points, existing planning point adjustment or planning indicator compliance query, matching the user intent with a preset standard instruction template to generate a structured instruction executable by the system.

[0032] In the embodiment, the core of step S101 is to complete the technical processing of natural language text by pre-training the BERT model, realize the process of text splitting → part-of-speech tagging → preliminary identification of candidate entities, and provide a structured text basis for subsequent accurate extraction of core demand information (entity information + instruction intent), which specifically includes: First, receiving the natural language demand input by the user, i.e., the user presents the planning-related appeal in daily communication language (without fixed format), for example, “generate 09-05-04 commercial land planning points report for certain certain area, including plot ratio and building height” and “adjust the plot ratio of 07-02-01 residential land to 2.0 and check compliance”. The receiving action only obtains the original text without additional processing to ensure complete transmission of the demand; Second, using a pre-trained BERT model to perform word segmentation and part-of-speech tagging on the text. The “pre-trained BERT model” refers to a model trained on a large amount of text, with semantic understanding ability, and adapted to the text features of the planning field; word segmentation is to split continuous text into independent semantic units (such as “generate 09-05-04 commercial land planning points report for certain certain area” into “generate”, “09-05-04 certain certain area”, “commercial land”, “planning points report”, etc.); part-of-speech tagging is to tag the grammatical properties of the segmented results (such as “09-05-04 certain certain area” is tagged as “proper noun (land block number)”, “plot ratio” is tagged as “noun (planning indicator)”), which converts unordered text into structured units for analysis; Third, preliminary identification of candidate entities, i.e., after BERT tagging, preliminary determination of semantic units that are core elements in the planning field, at least including three types of information: land block number (such as “09-05-04 certain certain area”, “07-02-01”), land use nature (such as “commercial land”, “residential land”), and planning indicator (such as “plot ratio”, “building height”). “Preliminary identification” only selects possible core entities, and does not verify accuracy (such as not determining whether the land block number is valid) to narrow the range for S102 accurate extraction.

[0033] The core of step S102 is to filter, verify and classify the candidate entities preliminarily identified in step S101, to ensure that the entity information is accurate and effective, and to provide a reliable basis for subsequent identification of instruction intent and conversion of structured instructions, which specifically includes: One is to filter invalid candidate entities: based on the planning entity determination rules in the customized entity library in the field of urban planning, the units irrelevant to planning requirements are removed from the candidate entities in S101; for example, for "generate a certain certain area 09-05-04 commercial land planning points report", the mood or action guiding words such as "help me" and "generate" preliminarily identified by S101 are filtered through the preset "invalid word library in the field of planning" (including "help me", "please" and the like), and only the valid entities such as "certain certain area 09-05-04", "commercial land" and "planning points report" are reserved; Second, check the accuracy of the valid entity, that is, verify the authenticity and compliance of the entity through the authoritative data source in the field of planning: for the plot number (such as "certain certain area 09-05-04"), call the plot basic database to judge whether it exists; for land use (such as "commercial land"), check the land use standard term library in the customized entity library to judge the compliance of the expression; for planning index (such as "plot ratio"), confirm whether it is an industry general index through the planning index name library in the customized entity library to avoid the mixing of self-defined indexes such as "building congestion degree"; Third, classify and arrange entity information, that is, classify the verified entities into "plot basic information" and "planning target information": plot basic information includes plot number (such as "certain certain area 09-05-04") and land use (such as "commercial land"); planning target information includes planning result type (such as "planning points report") and core concerned index (such as "plot ratio" and "building height"), forming a structured list (for example: "plot basic information: certain certain area 09-05-04 (commercial land); planning target information: planning points report, plot ratio").

[0034] It should be noted that the customized entity library in the field of urban planning is based on the standard specifications commonly used in the field of planning, which includes plot number format rules (such as plot coding rules published by local planning management departments), land use standard term library (such as standard terms in planning land classification related standards), and planning index name library (such as general index names defined in urban and rural planning technology related specifications); for obtaining the above standard specifications, coding rules and other data resources, those skilled in the art can achieve it through channels such as industry supervisory departments, local public platforms, standard information databases and public technical texts, which is clear to those skilled in the art. The core function of the library is to provide standardized basis for preprocessing of candidate entities, which is not the innovation of the present scheme, and the innovation of the present step lies in the filtering, correction and verification logic of the candidate entities based on the library.

[0035] The core function of step S103 is to identify the user instruction intention first, and then integrate the classified entity information + instruction intention into a structured instruction executable by the system, to complete the final conversion of natural language demand → structured instruction, which specifically includes: First, a planning field instruction intent classification system is constructed, i.e., preset user core demand types are taken as identification standards, at least three high-frequency intents are covered: generating planning results (such as generating planning point reports, indicator lists), adjusting planning indicators (such as increasing / decreasing plot ratios), and checking compliance (such as checking whether indicators meet higher-level planning), and each intent corresponds to clear keywords (such as “generate” and “output” for generating results, “adjust” and “modify” for adjusting indicators, and “check” and “whether it meets” for checking compliance); Second, instruction intent is identified based on text features, i.e., entity information and verb phrases in natural language in S102 are matched with the above classification system. For example, for “adjust 07-02-01 residential plot plot ratio to 2.0 and check whether it meets higher-level planning”, “adjust” matches “adjust planning indicators” and “check whether it meets” matches “check compliance”, so it is determined as a compound intent; for “generate a certain planning point report for a certain plot”, “generate” directly matches “generate planning results”; Third, entity and intent are integrated to generate structured instructions, i.e., entity information in S102 and identified intents are filled in according to the system preset template. For example, the template for generating planning results is “target plot: [number] ([land use property]); instruction type: [intent]; planning result: [type]; core indicator: [focus item]”, and after filling, it becomes “target plot: a certain plot 09-05-04 (commercial land); instruction type: generate planning results; planning result: planning point report; core indicator: plot ratio, building height”; the compound intent template supplements the “adjustment parameter” field, for example: “target plot: 07-02-01 (residential land); instruction type: adjust planning indicators + check compliance; adjustment parameter: plot ratio (new value 2.0); check basis: higher-level planning”, so that the instruction can be directly called by the subsequent module.

[0036] In one of the embodiments, after step S30, the following steps are further included: S301, a preset historical planning case library is called, the historical planning case library stores planning cases of the same region and the same land use property within a preset time range, and the planning case information at least includes plot area, core planning indicators, and approval results; S302, the land use property and plot area of the target plot in the indicator association set are taken as screening conditions to screen a plurality of similar cases with a similarity to the target plot within a preset similarity range from the historical planning case library; S303, the average value and fluctuation range of the core planning indicators in the similar cases are calculated, the core planning indicators at least include plot ratio and building height, and are consistent with the core planning indicator types of the target plot in the indicator association set; S304, compare the core planning indicators of the target plot in the index association set with the fluctuation range of the core planning indicators in the similar cases; S305, based on the comparison result, if the core planning indicators of the target plot exceed the fluctuation range, re-call the planning domain knowledge graph to check whether the regional development restriction clauses in the cross-scale planning association rules are missed, and adjust the core planning indicators of the target plot to a reasonable range based on the checking result; if the fluctuation range is not exceeded, the original index association set is retained.

[0037] In the embodiment, steps 301~S305 core are to verify and adjust the core planning indicators of the target plot in the way of "case reference + rule checking" by calling the historical planning case library, to further improve the rationality and landing of the index association set. The overall logic revolves around "calling case library → screening similar cases → calculating case indicators → comparing and adjusting indicators", and each step is as follows: Step S301 core is to call the preset historical planning case library in the planning domain knowledge graph to provide targeted data basis for subsequent screening of similar cases. The cases in the called library need to meet the basic conditions of "preset time range, same region, and same land use property", and each case at least contains three types of key information: plot area, core planning indicators, and approval results, to provide accurate data support for the subsequent screening of similar cases in S302.

[0038] Step S302 core is to screen similar cases matching the target plot from the historical planning case library to ensure the pertinence of case reference. Specifically, the two core attributes of the target plot in the "index association set" generated in step S30 are used as screening conditions: land use property (which must be completely consistent with the case land use property) and plot area (for example, if the target plot area is 6800 square meters, cases in the range of "6000-7500 square meters" are screened, which is the "preset similarity range" and can be preset according to actual scenarios), to screen multiple similar cases (such as 8 historical cases in the same district, with the same commercial land use, and close to the area) that meet the conditions from the library, and exclude cases with inconsistent land use properties and large area differences to avoid cases with low reference value interfering with subsequent analysis.

[0039] Step S303 core is to calculate the core planning indicators of similar cases to provide reference standards for subsequent index comparison, which specifically includes: First, determine the type of core indicators to be calculated: it must be completely consistent with the type of core planning indicators of the target plot in the "index association set", and at least cover the plot ratio and building height (for example, if the core indicators of the target plot are "plot ratio and building height", only these two types of indicators in the cases are calculated); Second, calculate the index data: for the selected similar cases, respectively, the "average" of the core indicators (such as 8 cases of average volume rate 2.5) and "fluctuation range" (such as the fluctuation range of volume rate 2.2-2.8), among which, the fluctuation range usually takes the minimum and maximum values of the case indicators, forming a reference interval that can be directly used for comparison.

[0040] Step S304 is to complete the core index comparison between the target plot and the similar cases, to judge whether the target index is in the reasonable reference interval, which specifically includes: comparing the core planning indicators of the target plot (such as the target plot volume rate 3.0, building height 80 meters) in the "index association set" generated in step S30 with the fluctuation range of similar case indicators (such as volume rate 2.2-2.8, building height ≤75 meters) calculated in step S303 one by one, to determine whether the target index exceeds the fluctuation range (such as in the above example, the volume rate 3.0 exceeds the range of 2.2-2.8, and the building height 80 meters exceeds the range of ≤75 meters).

[0041] Step S305 is to adjust or retain the core planning indicators of the target plot according to the comparison results, to ensure that the indicators meet the case reference and rule requirements, which specifically includes: First, the processing of indicators exceeding the fluctuation range: if the target index exceeds the fluctuation range of similar cases (such as volume rate 3.0 exceeds 2.2-2.8), then re-call the planning domain knowledge graph constructed in step S20, and focus on checking whether the "regional development restriction clause" in the "cross-scale planning association rule" is missed (such as the clause "commercial land volume rate upper limit 2.8" in the district control regulation), and based on the checking result (such as confirming that the clause is missed), adjust the target index to a reasonable range (such as reducing the volume rate from 3.0 to 2.8); second, the processing of indicators not exceeding the fluctuation range: if the target index is within the fluctuation range of similar cases (such as the target volume rate 2.6, within the range of 2.2-2.8), then there is no need to adjust, and the original index association set generated in step S30 is directly retained, to ensure that the subsequent verification process can be directly used.

[0042] In one embodiment, the cross-scale linkage verification of step S40 also includes a special verification of the rail transit safety protection zone for the index association set, which specifically includes the following steps: S411, call the preset rail transit control area database, extract the boundary coordinates of the target plot in the index association set, and judge whether the target plot is located within the rail transit operation safety protection zone range based on the rail line and station location data in the rail transit control area database; S412, if the target plot is located in the range of the rail operation safety protection zone, extract the rail safety protection parameters from the cross-scale planning correlation rules of the planning domain knowledge graph, and check whether the plot vertical control indicators in the index correlation set meet the requirements based on the rail safety protection parameters; S413, if the plot vertical control indicators in the index correlation set do not meet the requirements of the rail safety protection parameters, generate targeted optimization suggestions and update to the index correlation set; if they meet the requirements, mark the rail safety compliance in the collaborative verification result.

[0043] In this embodiment, the core is to check whether the target plot planning indicators meet the rail operation safety requirements by calling the rail transit control data and knowledge graph rules, to ensure that the index correlation set takes into account the planning rationality and rail safety. The overall logic revolves around "judging whether the plot is in the protection zone → checking the vertical indicators → processing the verification results", and each step is explained as follows: Step S411 is to judge whether the target plot is located in the rail operation safety protection zone, which provides a premise for subsequent special verification. Specifically, it includes: First, call the preset rail transit control area database (this database stores structured data such as rail line direction, station specific location, and safety protection zone boundary, which comes from industry public rail transit planning materials); Second, extract the boundary coordinates of the target plot in the index correlation set, and compare them with the location data of the rail line and station in the database to determine whether the plot is within the range of the rail operation safety protection zone (for example, if a plot is within 30 meters of the center line of the subway line 1, it is determined to be within the protection zone).

[0044] Step S412 is to check whether the vertical control indicators of the target plot located in the protection zone meet the rail safety requirements, which specifically includes: First, if the target plot is located in the rail operation safety protection zone, extract the rail safety protection parameters from the "cross-scale planning correlation rules" of the planning domain knowledge graph (such as "the excavation depth of the plot in the protection zone shall not exceed 5 meters", "the ground elevation shall not be lower than the rail bed elevation by 1.2 meters", and other restrictive parameters for vertical control); Second, based on the above parameters, check whether the plot vertical control indicators (such as pit depth, ground design elevation, etc.) in the index correlation set meet the requirements, and clarify the matching degree of the indicators and the safety parameters.

[0045] Step S413 is to process the index correlation set according to the vertical indicator verification result to ensure rail safety compliance, which specifically includes: One is if the vertical control index does not meet the requirements of the track safety protection parameters (such as the foundation pit depth exceeding the limited 5 meters), a targeted optimization suggestion (such as "suggesting the foundation pit depth to be adjusted to 4.5 meters") is generated, and the suggestion is updated to the index association set; Two is if the vertical control index meets the requirements, the track safety compliance is explicitly marked in the cross-scale linkage verification result, ensuring that the land planning index passes the track safety special verification.

[0046] In one of the embodiments, the cross-scale linkage verification in step S40 further includes an ecological sensitive area special verification of the index association set, specifically including the following steps: S421, a preset ecological data resource package is called, the ecological data resource package includes a structured ecological sensitive area GIS database and an unstructured regional ecological management and control text; a pre-trained BERT model is used to extract the implicit ecological constraint rules of the unstructured regional ecological management and control text and convert them into structured implicit ecological constraint rules; S422, the structured implicit ecological constraint rules, the boundary data in the ecological sensitive area GIS database, and the ecological rule sub-module contained in the cross-scale planning associated rule in the planning domain knowledge graph are fused to infer and generate a unique ecological verification parameter set for the target land block in the index association set, wherein the ecological rule sub-module stores ecological management and control rules of different levels and a preset priority, and if there is a rule conflict, the preset priority is used for automatic resolution; S423, based on the unique ecological verification parameter set, the ecological sensitive type index in the index association set is double-verified, specifically including: Verifying whether the target land block boundary exceeds the management and control range in the ecological sensitive area GIS database; Verifying whether the ecological sensitive type index violates the structured implicit ecological constraint rules; S424, if any verification item does not meet the requirements, the planning domain knowledge graph is called again, and a targeted optimization suggestion is generated based on the cross-scale planning associated rule and updated to the index association set; if all meet the requirements, the ecological sensitive area compliance is marked in the collaborative verification result.

[0047] In the embodiment, the core is to generate verification parameters by calling ecological data and fusing multiple source rules to carry out ecological sensitive area special verification on the index association set, ensuring that the index takes into account the planning demand and ecological protection, and the overall logic revolves around "obtaining structured ecological constraints → fusing to generate unique verification parameters → double verification → processing results", and each step is specifically explained as follows: The core of step S421 is to convert ecological data resources into structured constraint rules to provide standardized basis for subsequent verification, specifically including: One is to call the preset ecological data resource package, including: structured ecological sensitive area GIS database storing the boundary coordinates, control level and other spatial data of ecological sensitive areas such as wetlands, nature reserves and waterfront belts; unstructured regional ecological control text including local ecological protection regulations, ecological explanations in district planning, etc. (such as "no new industrial projects within 500 meters of XX wetland"); Two is to use the pre-trained BERT model to process unstructured text - identify implicit constraints not directly listed in the text through semantic analysis (such as deducing "the waterfront area needs to retain sufficient green buffer zone" to "the building setback waterfront boundary should not be less than 30 meters"), and convert these implicit rules into a structured format of "constraint object + limitation condition" (such as "building setback: waterfront boundary ≥ 30 meters").

[0048] Step S422 is the core of fusing multiple source ecological rules to generate a unique check standard for the target plot, specifically including: One is to fuse three types of data: structured implicit ecological constraint rules generated by S421, boundary data of ecological sensitive areas GIS database (such as wetland core area coordinate range), and ecological rule submodule in the planning domain knowledge graph "cross-scale planning correlation rules" (this module stores rules by level, such as superior ecological protection rules and local district ecological details, and presets priority - superior rules take precedence over local rules); Two is to infer and generate a unique set of ecological check parameters: if there is a rule conflict in the fusion (such as local rules allowing building height of 60 meters and superior rules limiting 50 meters), automatically adopt the stricter superior rules according to the preset priority, and finally form a unique parameter set containing "ecological control range boundary, ecological sensitive type index limit value" (such as "wetland within 500 meters is the control range, building height ≤ 50 meters").

[0049] Step S423 is the core of confirming the ecological compliance of the target plot through double-checking, specifically including: One is spatial boundary checking: comparing the boundary coordinates of the target plot in the index association set with the "ecological sensitive area GIS boundary data" in the unique ecological check parameter set to determine whether the plot is partially or entirely outside the control range (such as if 1 / 3 of the plot area is located within the 500-meter control area of the wetland, it is determined to be outside); Two is index compliance checking: comparing the ecological sensitive type index (such as building height, land use, green ratio, etc.) in the index association set with the structured implicit ecological constraint rules (such as checking whether "building height 55 meters" violates the rule "≤ 50 meters").

[0050] Step S424 is the core of optimizing the index or labeling compliance according to the double-checking results, specifically including: One is that if any check item does not meet the requirements (such as the plot exceeding the control range or the building height exceeding the standard), the planning domain knowledge graph is called again, the cross-scale planning association rule is generated based on the targeted optimization suggestions (such as "suggesting the plot boundary to be shrunk to the control range outside, and the building height to be adjusted to 45 meters"), and the suggestions are updated to the index association set; Two is that if the space boundary and the ecological sensitive type index both meet the requirements of the unique ecological check parameter set, the cross-scale linkage check result is marked as "ecological sensitive area compliance" to ensure that the plot planning index passes the ecological special check.

[0051] In one of the embodiments, in the step of performing collaborative checking on the index association set, the specific steps of the cross-scale linkage checking based on the upper planning data and through GIS spatial analysis on the index association set include: S431, call the preset upper planning database, extract the upper planning data corresponding to the planning jurisdiction region to which the target plot in the index association set belongs, and the upper planning data at least includes the upper planning space control range and the upper planning index constraint; S432, perform spatial overlay analysis on the boundary of the target plot in the index association set and the upper planning space control range through the GIS spatial analysis tool, and determine whether the target plot is completely located within the space range allowed for development by the upper planning, or whether it is partially overlapped with the forbidden construction area / limited construction area; S433, if the target plot is completely located within the allowed development range, based on the upper planning index constraint, check whether the core planning index in the index association set meets the upper planning index threshold requirement; if the target plot is partially overlapped with the forbidden construction area / limited construction area, directly determine that the development type index related to the overlapped area in the index association set does not meet the requirement; S434, if the core planning index meets the upper planning index threshold requirement, mark the upper planning compliance in the collaborative checking result; if the core planning index does not meet the upper planning index threshold requirement, generate a space range adjustment suggestion or an index down adjustment suggestion based on the upper planning data, and update it to the index association set.

[0052] In this embodiment, the core is to match the upper planning requirements, analyze the spatial relationship and check the index threshold, to ensure that the index association set meets the space and index constraints of the upper planning, and the overall logic is developed around "calling upper planning data→spatial overlay analysis→checking index in different scenes→processing checking result", and each step is specifically explained as follows: Step S431 is to accurately obtain the upper planning data corresponding to the target plot, to provide a standardized basis for subsequent space and index checking, which specifically includes: One is to call the preset upper planning database, which stores structured upper planning results according to the planning jurisdiction area classification structure, covering planning files of different levels such as overall planning of the whole space and control detailed planning of the area, and the data comes from the industry public planning compilation results; Two is to extract the corresponding upper planning data according to the "planning jurisdiction area" (such as XX area, XX new city) of the target plot in the index association set, which mainly includes two types of information: upper planning space control range (to clearly allow the development area, forbidden area (such as ecological protection core area), limited area (such as geological disaster prone area) in the spatial range with coordinate boundary), and upper planning index constraint (index threshold for different land use properties, such as "XX area residential land volume ratio upper limit 2.0", "industrial land building height ≤24 meters").

[0053] The core of step S432 is to determine the positional relationship between the target plot and the upper planning space control range through GIS spatial analysis, which provides the premise for scene-based verification, which specifically includes: through GIS spatial analysis tools (such as overlay analysis, buffer analysis function), the boundary coordinates (such as polygon vertex coordinates) of the target plot in the index association set and the vector layer of the upper planning space control range are overlaid in space - that is, the spatial range layers of the two are overlapped through coordinate matching, the overlapping area and proportion of the target plot and the allowed development area, forbidden area, limited area are calculated, and finally it is judged: whether the target plot is located in the allowed development range; or whether there is a part of the area (such as 30% area) overlapping with the forbidden area / limited area (such as the east part of the plot falling into the ecological protection core area).

[0054] The core of step S433 is to verify the compliance of the index association set according to the spatial relationship, which ensures that the index meets the spatial control requirements and does not exceed the upper index constraint, which specifically includes: One is that if the target plot is completely located in the upper planning allowed development range (without any area overlapping the forbidden area / limited area), then based on the upper planning index constraint, the core planning index (such as volume ratio, building density, land use property) in the index association set is verified whether it meets the threshold requirement (such as verifying whether "target plot volume ratio 2.2" meets the constraint of "residential land upper limit 2.0"); two is that if the target plot partially overlaps the forbidden area / limited area (such as 15% area located in the ecological protection core area), then directly determine that the development-related index (such as the construction scale, building base area, excavation depth of the area) in the index association set does not meet the requirement, because the forbidden area / limited area has strict restrictions on development behavior.

[0055] The core of step S434 is to optimize the index or mark the compliance according to the verification result, which ensures that the index association set completely meets the upper planning requirements, which specifically includes: One is if the core planning indicators meet the threshold requirements of the upper planning indicators (such as the plot ratio 1.8≤2.0), the upper planning compliance is marked in the coordination checking result, indicating that the planning indicators of the plot pass the checking of the upper planning level; Two is if the core planning indicators do not meet the threshold requirements (such as the plot ratio 2.3>2.0), two types of targeted suggestions are generated based on the upper planning data: spatial range adjustment suggestions (such as “shrink the south boundary of the plot by 30 meters to the north and include it in the low constraint area”) or indicator reduction suggestions (such as “reduce the plot ratio from 2.3 to 1.9”), and the suggestions are updated to the indicator association set to ensure that the adjusted indicators meet the upper planning constraints.

[0056] In one of the embodiments, in the step of performing coordination checking on the indicator association set, the specific steps of the cross-professional parameter adaptation checking based on the mapping relationship of the cross-professional parameters in the planning domain knowledge graph include: S441, call the planning domain knowledge graph and extract the mapping relationship of the cross-professional parameters, the mapping relationship of the cross-professional parameters at least includes the parameter mapping of the building profession and the municipal profession, the parameter mapping of the building profession and the landscape profession; S442, filter out the cross-professional associated indicator group from the indicator association set, the cross-professional associated indicator group at least contains planning indicators of two different professions; S443, based on the mapping relationship of the cross-professional parameters, adaptively check each cross-professional associated indicator group, including: checking whether the pre-sequence indicators trigger the constraint threshold of the post-sequence indicators; checking whether there is an implicit conflict in the parameters in the indicator group; S444, if the checking meets the mapping relationship requirements of the cross-professional parameters, mark the cross-professional parameter adaptation in the coordination checking result; if there are unmatched items, call the knowledge graph reasoning to generate targeted optimization suggestions and update them to the indicator association set.

[0057] In this embodiment, the core is to extract the mapping relationship of cross-professional parameters, filter the associated indicator group, and check the adaptability, to ensure that there is no conflict between planning indicators of different professions and they are matched with each other, and the overall logic revolves around “extract mapping relationship→filter associated indicator group→adaptability checking→handle checking result”, and each step is explained as follows: Step S441 is to obtain the mapping relationship of cross-professional parameters, which provides a rule basis for subsequent adaptability checking, and specifically includes: calling the planning domain knowledge graph, and extracting the preset mapping relationship of cross-professional parameters from it, that is, the association rules and constraint logic between planning indicators of different professions, which at least covers two types of core mappings: Parameter mapping between building and municipal professions (e.g., "building height > 24 meters → fire access width ≥ 4 meters", "total building area > 50,000 square meters → municipal water supply pipe diameter ≥ 200 mm"); Parameter mapping between building and landscape professions (e.g., "building density > 30% → landscape green area ratio ≥ 25%", "building setback distance < 5 meters → street landscape belt width ≥ 8 meters").

[0058] Step S442 core is to filter out the index combination that needs to be checked across professions from the index association set, and to clearly identify the checking object, which specifically includes: identifying and filtering "cross-professional associated index groups" from the index association set, i.e., associated combinations composed of two or more different professional planning indexes, for example: "Building height (30 meters)" of the building profession and "fire access width (3.5 meters)" of the municipal profession form a group; "Building density (35%)" of the building profession and "green area ratio (20%)" of the landscape profession form a group.

[0059] Step S443 core is to perform adaptability checking on cross-professional associated index groups based on mapping relationships, to determine whether the indexes match, which specifically includes: First, check whether the pre-sequence index triggers the constraint threshold of the post-sequence index: taking the value of the pre-sequence index (e.g., building height) as the benchmark, determine whether the constraint condition (e.g., building height 30 meters > 24 meters) of the post-sequence index (e.g., fire access width) is met according to the mapping relationship (e.g., building height 30 meters > 24 meters triggers "fire access width ≥ 4 meters", if the actual width is 3.5 meters, it does not meet the standard); Second, check whether there is an implicit conflict between the parameters in the index group: identify the situation where there is no direct association between the indexes but there is a logical contradiction (e.g., building density 35% corresponds to a requirement of green area ratio ≥ 25%, if the actual green area ratio is 20%, there is an implicit conflict between the two).

[0060] Step S444 core is to process the index association set according to the adaptability checking result to ensure the adaptation of cross-professional parameters, which specifically includes: First, if all cross-professional associated index groups meet the requirements of the mapping relationship (e.g., building height 30 meters corresponds to fire access width 4.5 meters, and there is no implicit conflict), clearly mark "cross-professional parameter adaptation" in the collaborative checking result; Second, if there are unmatched items (e.g., fire access width does not meet the standard or green area ratio is insufficient), call the planning domain knowledge graph reasoning to generate targeted optimization suggestions (e.g., "adjust the fire access width to 4.2 meters", "reduce the building density to 30% to meet the green area ratio requirement"), and update the suggestions to the index association set.

[0061] According to the need, in one embodiment, the step S40 of performing the consistency check on the index association set further includes dynamic checking based on the space-time coupling constraint, and specifically includes the following steps: S451, extracting a planning period parameter of the target land plot from the index association set, the planning period parameter including a near-term planning interval, a medium-term planning interval and a long-term planning interval divided according to time stages; S452, obtaining a planning and construction time sequence parameter of an associated facility within a preset range of the target land plot, the associated facility at least including a transportation hub, an infrastructure and a service facility, and the planning and construction time sequence parameter at least including a construction planning start time, a facility put-into-use time and a phased construction scale; S453, based on the planning period parameter and the planning and construction time sequence parameter of the associated facility within the preset range of the target land plot, constructing a space-time coupling constraint rule including a time dimension constraint and a space association constraint, wherein the time dimension constraint is that different threshold values of a core planning index of the target land plot correspond to different time intervals, and the threshold values of each interval increase or decrease in a ladder form as the planning period advances; and the space association constraint is that a phased development index of the target land plot matches the facility put-into-use time of the associated facility within the preset range; S454, based on the space-time coupling constraint rule, performing dynamic checking on a phased index in the index association set, the phased index being a phased core planning index corresponding to the near-term planning interval, the medium-term planning interval and the long-term planning interval; S455, if the phased index does not meet the constraint conditions of the time dimension constraint or the space association constraint, generating a time sequence adjustment suggestion and updating to the index association set, the time sequence adjustment suggestion including a stage threshold value adjustment of the core planning index or a phased development progress adjustment; and if the phased index meets the time dimension constraint and the space association constraint, marking space-time coupling compliance in the consistency check result.

[0062] To avoid the problem that the existing planning index checking only adopts “static threshold value comparison” and does not combine the planning period stage of the target land plot and the construction time sequence of the surrounding associated facility, resulting in that the index is out of step with the dynamic development rhythm of the city (for example, early planning index is too strict to cause low efficiency of land use, or surrounding facilities are not put into use but set too high development index to cause overloading), in the embodiment, a “dynamic checking mechanism of space-time coupling constraint” is constructed through steps S451-S455, so that the planning index is dynamically adapted to the time stage and the progress of the surrounding facility, and it is ensured that the index meets the current demand and matches long-term development, which is specifically explained as follows: Step S451 core is to extract the planning cycle parameters of the target plot from the index association set, to split the overall planning period into different time stages, and to clarify the time interval boundaries of "short term, medium term, and long term" to lay the foundation for setting index thresholds in subsequent stages. Specifically includes: Extract the planning cycle parameters from the index association set. The parameters are divided into short-term planning intervals (such as 1-3 years, corresponding to the early development of the plot), medium-term planning intervals (such as 4-6 years, corresponding to the gradual improvement of surrounding facilities), and long-term planning intervals (such as 7-10 years, corresponding to the mature stage of the region). For example: a new city plot has a total planning period of 10 years, the short term is 2025-2027 (first build residential phase), the medium term is 2028-2030 (build commercial support), and the long term is 2031-2034 (build high-end industrial building). By clarifying the interval, the index can be adjusted by stage.

[0063] Step S452 core is to obtain the planning and construction time sequence parameters of the associated facilities within the preset range of the target plot (such as 3-5 kilometers from the plot center, determined by facility service radius), to clarify the facilities "when to build, when to use, and how many steps to build", and to ensure that the plot development index matches the facility carrying capacity. Specifically includes: Obtain the planning and construction time sequence parameters of the associated facilities, including at least transportation hubs (such as subway stations and bus hubs, affecting plot traffic accessibility), infrastructure (such as sewage treatment plants and substations, affecting plot resource carrying capacity), and service facilities (such as schools and community hospitals, affecting plot supporting perfection). The planning and construction time sequence parameters include at least construction plan start time (such as subway line 2 starting construction in 2026), facility commissioning time (such as opening to traffic in 2030), and phased construction scale (such as sewage treatment plant phase I commissioned in 2027, with a treatment capacity of 50,000 tons / day).

[0064] It should be noted that the above-mentioned planning and construction time sequence parameters of the associated facilities belong to the conventional public information in the field of urban planning, and those skilled in the art can obtain them through relevant department-issued urban master planning texts, special facility construction plan announcements (such as annual construction plan of the housing and construction department), and industry public databases (such as urban infrastructure construction information platform). The acquisition method is clear to those skilled in the art.

[0065] Step S453 core is to construct spatio-temporal coupling constraint rules based on planning cycle parameters (time dimension) and associated facility time sequence parameters (space dimension), which not only specify index thresholds in different stages, but also require index matching facility progress to avoid "time disconnection" or "space mismatch". Specifically includes: The rules are constructed based on planning period and facility timing, wherein the time dimension constraint is that the core planning index (such as the plot ratio, building height) corresponds to different thresholds in the near / mid / long term, and is adjusted step by step with the period (such as the plot ratio near term ≤2.0, mid term ≤2.5, long term ≤3.0, and gradually relaxed with the maturity of the region); the space correlation constraint is that the phased development index (such as the construction area of each stage and the sewage discharge) is matched with the facility commissioning time (such as the sewage treatment plant commissioned in 2027, the near-term sewage discharge ≤500 tons / day, and the mid-term can be relaxed to ≤1000 tons / day after commissioning).

[0066] Step S454 is to check the "phased index" (the phased core index corresponding to the near / mid / long term) in the index correlation set according to the space-time coupling constraint rule, to see if the index meets the time threshold of the corresponding stage and matches the surrounding facility progress. Specifically, it includes: The phased index (such as near-term "plot ratio 1.8, sewage discharge 400 tons / day", mid-term "plot ratio 2.3, sewage discharge 800 tons / day") is selected from the index correlation set, and is checked one by one based on the space-time coupling constraint rule - whether the near-term plot ratio 1.8 is ≤2.0 (time constraint), and whether the sewage discharge 400 tons / day is ≤500 tons / day (space constraint, corresponding to the facility not commissioned); whether the mid-term plot ratio 2.3 is ≤2.5 (time constraint), and whether the sewage discharge 800 tons / day is ≤1000 tons / day (space constraint, corresponding to the facility commissioned).

[0067] Step S455 is to generate adjustment suggestions for the indexes that do not meet the constraints according to the dynamic checking results, and to label the compliance for the indexes that meet the constraints, so as to finally adapt the index correlation set to the urban dynamic development rhythm. Specifically, it includes: If the phased index does not meet the constraint (such as mid-term plot ratio 2.6>2.5, or near-term sewage discharge 600 tons / day>500 tons / day), generate timing adjustment suggestions and update to the index correlation set - if the plot ratio exceeds the threshold, suggest to reduce to 2.4 (adjust the stage threshold), and if the sewage exceeds, suggest to delay part of the residential construction to after the sewage treatment plant is commissioned in 2027 (adjust the development progress of each stage); if all phased indexes meet the time dimension constraint and the space correlation constraint, label "space-time coupling compliance" in the collaborative checking result, indicating that the index can match the planning period and the surrounding facility progress.

[0068] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0069] In one of the embodiments, a cross-scale planning point generation system based on NLP and knowledge graph reasoning is provided, which corresponds to the above-mentioned cross-scale planning point generation method based on NLP and knowledge graph reasoning. The cross-scale planning point generation system based on NLP and knowledge graph reasoning comprises: A demand extraction module is configured to receive and analyze a natural language demand input by a user, extract core demand information including entity information and instruction intention in the natural language demand, and convert the core demand information into a structured instruction executable by a system. A knowledge graph construction module is configured to construct a planning domain knowledge graph, which comprises associated relationship information of constraint relationships between planning indicators, associated rules of cross-scale planning, and mapping relationships of cross-professional parameters. An associated reasoning module is configured to extract a target planning indicator from the structured instruction, perform associated reasoning on the target planning indicator based on the structured instruction and the planning domain knowledge graph, and generate an indicator association set comprising associated indicators and adjustment ranges. A coordination verification module is configured to perform coordination verification on the indicator association set, which comprises cross-scale linkage verification and cross-professional parameter adaptation verification. The cross-scale linkage verification is based on upper planning data and performs verification on the indicator association set through GIS spatial analysis. The cross-professional parameter adaptation verification is based on the mapping relationships of cross-professional parameters in the planning domain knowledge graph to perform adaptability verification on the indicator association set. A report generation module is configured to generate a structured planning point report according to the indicator association set that passes the coordination verification.

[0070] The specific limitations of the cross-scale planning point generation system based on NLP and knowledge graph reasoning can be referred to the limitations of the cross-scale planning point generation method based on NLP and knowledge graph reasoning in the above, which will not be repeated here. Each module in the above cross-scale planning point generation system based on NLP and knowledge graph reasoning can be realized by software, hardware, and combinations thereof, in whole or in part. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.

[0071] In one of the embodiments, a computer device, which can be a server, is provided, and its internal structure diagram can be as shown in Figure 2As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for storing data, data processing, data analysis, etc. The network interface of the computer device is used for communication connection with the external terminal through the network. The computer program is executed by the processor to implement a cross-scale planning point generation method based on NLP and knowledge graph reasoning.

[0072] In one of the embodiments, a computer device is provided, including a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement a cross-scale planning point generation method based on NLP and knowledge graph reasoning.

[0073] In one of the embodiments, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by the processor to implement a cross-scale planning point generation method based on NLP and knowledge graph reasoning.

[0074] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM) and the like.

[0075] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0076] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A cross-scale planning point generation method based on NLP and knowledge graph reasoning, characterized in that, Comprise the following steps: S10, receiving and analyzing the user input natural language requirements, extracting the core requirement information containing entity information and instruction intention in the natural language requirements, and converting the core requirement information into structured instructions executable by the system; S20, constructing a planning domain knowledge graph, the planning domain knowledge graph contains constraint relationship between planning indicators, correlation rule of cross-scale planning and mapping relationship of cross-professional parameters; S30, extracting the target planning indicator from the structured instruction, calling the planning domain knowledge graph based on the structured instruction to associate and reason the target planning indicator, and generating an indicator association set containing associated indicators and adjustment range; S40, the synergistic verification of the indicator association set, the synergistic verification includes cross-scale linkage verification and cross-professional parameter adaptation verification, wherein the cross-scale linkage verification is based on the upper planning data and verified by GIS spatial analysis on the indicator association set, and the cross-professional parameter adaptation verification is based on the mapping relationship of cross-professional parameters in the planning domain knowledge graph to adaptively verify the indicator association set; S50, generating a structured planning point report according to the indicator association set passed by the synergistic verification.

2. The cross-scale planning bullet generation method based on NLP and knowledge graph reasoning according to claim 1, wherein, In the step S10, the following steps are specifically included: S101, receiving the user input natural language requirements, using the pre-trained BERT model to perform word segmentation and part-of-speech tagging on the text of the natural language requirements, and preliminarily identifying candidate entities contained in the text, the candidate entities at least including plot number, land use property and planning indicator; S102, calling a customized entity word library in the field of urban planning to preprocess the candidate entities, the customized entity word library in the field of urban planning at least including plot number format rules, land use property standard term library and planning indicator name library; wherein the preprocessing includes: Eliminate non-planning domain entities in the candidate entities; Correct non-standard terms associated with land use properties in the candidate entities to corresponding standard terms in the land use property standard term library; According to the plot number format rule, the plot number in the candidate entity is verified, and the plot number with the format conforming to the rule is retained; S103, calling a pre-trained intention determination model to analyze the associated relationship of the preprocessed candidate entities, determining that the user intention is one of newly generated planning points, existing planning point adjustment or planning indicator compliance query, matching the user intention with a pre-set standard instruction template to generate a structured instruction executable by the system. 3.The NLP and knowledge graph reasoning based cross-scale planning point generation method of claim 1, wherein, After the step S30, the following steps are further included: S301, calling a pre-set historical planning case library, the historical planning case library stores planning cases with the same region and land use property within a pre-set time range, and the planning case information at least includes plot area, core planning indicator and approval result; S302, using the land use property and plot area of the target plot in the indicator association set as the screening conditions, screening a plurality of similar cases with a similarity to the target plot within a pre-set similarity range from the historical planning case library; S303, calculating the average value and fluctuation range of the core planning indicators in the similar cases, the core planning indicators at least including the plot ratio and the building height, and being consistent with the core planning indicator types of the target plot in the index association set; S304, comparing the core planning indicators of the target plot in the index association set with the fluctuation range of the core planning indicators in the similar cases; S305, based on the comparison result, if the core planning indicators of the target plot exceed the fluctuation range, re-calling the planning domain knowledge graph, checking whether the regional development restriction clauses in the associated rules of the cross-scale planning are missed, and adjusting the core planning indicators of the target plot to a reasonable range based on the checking result; If the fluctuation range is not exceeded, the original index association set is retained. 4.The NLP and knowledge graph reasoning based cross-scale planning point generation method of claim 1, wherein, In the cross-scale linkage verification of the step S40, the special verification of the rail transit safety protection zone in the index association set is further included, specifically including the following steps: S411, calling a preset rail transit control zone database, extracting the boundary coordinates of the target plot in the index association set, judging whether the target plot is located within the rail transit operation safety protection zone range based on the rail line and station location data in the rail transit control zone database; S412, if the target plot is located within the rail transit operation safety protection zone range, extracting the rail safety protection parameters from the associated rules of the cross-scale planning in the planning domain knowledge graph, and checking whether the plot vertical control indicators in the index association set meet the requirements based on the rail safety protection parameters; S413, if the plot vertical control indicators in the index association set do not meet the requirements of the rail safety protection parameters, generating a targeted optimization suggestion and updating to the index association set; if the requirements are met, marking the rail safety compliance in the collaborative verification result.

5. The NLP and knowledge graph reasoning based cross-scale planning point generation method of claim 1, wherein, In the cross-scale linkage verification of the step S40, the special verification of the ecological sensitive area in the index association set is further included, specifically including the following steps: S421, calling a preset ecological data resource package, the ecological data resource package including a structured ecological sensitive area GIS database and an unstructured regional ecological control text; using a pre-trained BERT model to extract the implicit ecological constraint rules of the unstructured regional ecological control text and convert them into structured implicit ecological constraint rules; S422, fusing the structured implicit ecological constraint rules, the boundary data in the ecological sensitive area GIS database, and the ecological rule sub-module contained in the associated rules of the cross-scale planning in the planning domain knowledge graph, and reasoning to generate a unique ecological verification parameter set for the target plot in the index association set, wherein the ecological rule sub-module stores different levels of ecological control rules and preset priorities, and if there is a rule conflict, it is automatically resolved according to the preset priorities; S423, based on the unique ecological verification parameter set, double-checking the ecological sensitive type indicators in the index association set, specifically including: checking whether the boundary of the target plot exceeds the control range in the ecological sensitive area GIS database; checking whether the ecological sensitive type indicators violate the structured implicit ecological constraint rules; S424, if any of the check items does not meet the requirements, re-call the planning domain knowledge graph, generate targeted optimization suggestions based on the correlation rules of cross-scale planning and update to the index correlation set; if all meet the requirements, mark the ecological sensitive area compliance in the collaborative verification result.

6. The cross-scale planning bullet generation method based on NLP and knowledge graph reasoning according to claim 1, wherein, In the step of collaborative verification of the index correlation set, the specific steps of the cross-scale linkage verification based on the upper planning data and through GIS spatial analysis to verify the index correlation set include: S431, call the preset upper planning database, extract the upper planning data corresponding to the planning jurisdiction region to which the target land plot in the index correlation set belongs, and the upper planning data at least includes the upper planning space control range and the upper planning index constraint; S432, through the GIS spatial analysis tool, the boundary of the target land plot in the index correlation set is subjected to spatial overlay analysis with the upper planning space control range, to determine whether the target land plot is completely located within the space range allowed for development by the upper planning, or whether it is partially overlapped with the forbidden construction area / limited construction area; S433, if the target land plot is completely located within the allowed development range, based on the upper planning index constraint, it is verified whether the core planning index in the index correlation set meets the upper planning index threshold requirement; if the target land plot is partially overlapped with the forbidden construction area / limited construction area, it is directly determined that the development type index in the index correlation set related to the overlapped area does not meet the requirements; S434, if the core planning index meets the upper planning index threshold requirement, mark the upper planning compliance in the collaborative verification result; if the core planning index does not meet the upper planning index threshold requirement, generate a space range adjustment suggestion or an index down-regulation suggestion based on the upper planning data, and update to the index correlation set.

7. The NLP and knowledge graph reasoning based cross-scale planning point generation method of claim 1, wherein, In the step of collaborative verification of the index correlation set, the specific steps of the cross-professional parameter adaptation verification based on the mapping relationship of cross-professional parameters in the planning domain knowledge graph to adaptively verify the index correlation set include: S441, call the planning domain knowledge graph and extract the mapping relationship of the cross-professional parameters, and the mapping relationship of the cross-professional parameters at least includes the parameter mapping of the building profession and the municipal profession, the parameter mapping of the building profession and the landscape profession; S442, filter out the cross-professional correlation index group from the index correlation set, and the cross-professional correlation index group at least includes planning indexes of two different professions; S443, based on the mapping relationship of the cross-professional parameters, adaptively verify each cross-professional correlation index group, including: verify whether the pre-sequence index triggers the constraint threshold of the post-sequence index; verify whether there is an implicit conflict in the parameters in the index group; S444, if the verification meets the requirements of the mapping relationship of the cross-professional parameters, mark the cross-professional parameter adaptation in the collaborative verification result; if there is a mismatch, call the knowledge graph reasoning to generate targeted optimization suggestions and update to the index correlation set.

8. A cross-scale planning point generation system based on NLP and knowledge graph reasoning, for implementing the steps of a cross-scale planning point generation method based on NLP and knowledge graph reasoning according to any one of claims 1-7, characterized in that, including: a demand extraction module, configured to receive and parse a natural language demand input by a user, extract core demand information including entity information and instruction intention in the natural language demand, and convert the core demand information into a structured instruction executable by a system; The knowledge graph construction module is configured to construct a planning domain knowledge graph, which contains constraint relationships between planning indexes, correlation rule information of cross-scale planning, and mapping relationships of cross-professional parameters. The correlation reasoning module is configured to extract a target planning index from the structured instruction, perform correlation reasoning on the target planning index based on the planning domain knowledge graph, and generate an index correlation set containing correlated indexes and adjustment ranges. The collaborative verification module is configured to perform collaborative verification on the index correlation set, which includes cross-scale linkage verification and cross-professional parameter adaptation verification. The cross-scale linkage verification is based on higher-level planning data and performed by GIS spatial analysis. The cross-professional parameter adaptation verification is based on the mapping relationships of cross-professional parameters in the planning domain knowledge graph. The report generation module is configured to generate a structured planning point report based on the index correlation set that passes the collaborative verification.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The computer program is executed by the processor to implement the steps of the cross-scale planning point generation method based on NLP and knowledge graph reasoning according to any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: The computer program is executed by the processor to implement the steps of the cross-scale planning point generation method based on NLP and knowledge graph reasoning according to any one of claims 1-7.

Images