A method and system for matching and distributing intelligent agents for space tasks based on spatial features

Abstract

This invention relates to the field of urban planning and management technology, specifically providing a method and system for spatial task intelligent agent matching and distribution based on spatial features. The method includes the following steps: receiving and parsing planning spatial tasks input by users, generating a standardized feature table containing entity information, task attributes, and spatial data, and simultaneously generating spatial scene tags and an associated task list; constructing an intelligent agent pool and an adaptation rule base, initially screening the intelligent agent pool based on the spatial scene tags and calling the adaptation rules to obtain a scene-adapted intelligent agent set; further screening the scene-adapted intelligent agent set according to the standardized feature table to obtain a target candidate set, forming a master-slave collaborative intelligent agent group and distributing tasks, establishing a data channel; monitoring the execution status in real time, restoring collaboration in case of anomalies, and linking with the upper-level planning database to verify the compliance of task results. This invention improves the accuracy and efficiency of spatial task matching and ensures compliance throughout the entire process through standardized parsing, precise screening, and master-slave collaboration.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban planning management, and particularly relates to a spatial task intelligent agent matching and distribution method and system based on spatial features. BACKGROUND

[0002] With the intelligent development of smart cities, land space planning and other fields, the complexity and collaboration requirements of spatial tasks (such as plot development planning, emergency disaster space deployment, historical block reconstruction, etc.) are increasingly improving. The traditional mode of relying on manual assignment or single intelligent agent processing has been difficult to meet the needs, and it is urgent to realize efficient matching and distribution through multi-agent collaboration technology. However, in the existing technology, there are some core defects in the matching and distribution of spatial tasks through multi-agent collaboration technology, such as:

[0003] Insufficient task analysis and feature standardization: Most solutions directly match intelligent agents based on task names or simple keywords, without structurally analyzing the core features of spatial tasks (such as entity information, task attributes, and required spatial data), and without forming a unified "standardized feature table". This leads to the inability of intelligent agents to accurately identify the spatial range (such as complex terrain, multi-plot linkage) and attribute constraints (such as environmental protection level, construction period requirement) of tasks, resulting in low matching accuracy.

[0004] Lack of associated task mining and collaboration: Spatial tasks often have strong correlation (such as "building planning" needing to be associated with "pipeline laying", "green design" and other pre / post tasks), but the existing technology does not establish an associated task mining mechanism based on spatial topological relationships (such as plot inclusion, adjacency) and functional dependency relationships, resulting in intelligent agents only being able to handle single tasks and being unable to handle associated tasks in a linked manner, which can easily result in task disconnection or repeated execution.

[0005] Inadequate intelligent agent collaboration group formation and data interaction: Even if multiple intelligent agents are selected, existing solutions mostly use a "parallel allocation" mode, without clearly defining the master-slave collaboration relationship (such as the division of labor between core task decision-making and associated task execution), and lacking real-time data sharing channels based on task attributes, resulting in data not being synchronized and high conflict rates when multiple intelligent agents collaborate, making it difficult to form efficient collaboration.

[0006] In summary, the existing technology has significant deficiencies in the standardized analysis of spatial tasks, associated task collaboration, and other aspects, resulting in low efficiency, high error rate, and poor adaptability in spatial task processing. Therefore, there is an urgent need for a method that can accurately match intelligent agents based on spatial features, efficiently collaborate, and ensure full-process compliance, to solve the above technical problems. SUMMARY

[0007] In order to overcome the shortcomings of the prior art, the present application provides a spatial task agent matching and distribution method and system based on spatial features to solve the problems in the prior art.

[0008] One embodiment of the present application provides a spatial task agent matching and distribution method based on spatial features, comprising the following steps:

[0009] S10, receiving and analyzing the user input planning spatial task, converting the planning spatial task into a standardized feature table containing entity information, task attributes and required spatial data, calling a preset GIS database based on the spatial range in the standardized feature table, generating a spatial scene label of the planning spatial task, and extracting associated tasks based on the preset association mining rules and the task attributes in the standardized feature table to obtain an associated task list;

[0010] S20, constructing an agent pool and an adaptation rule library, wherein the agent pool stores a plurality of special agents, each of which is configured with a preset agent attribute, and the adaptation rule library is preset with a plurality of matching rules of spatial scenes and agents; based on the spatial scene label of the planning spatial task, the corresponding filtering rules in the adaptation rule library are called to preliminarily screen the special agents in the agent pool to obtain a scene-adapted agent set;

[0011] S30, according to the standardized feature table, the scene-adapted agent set is re-screened according to a preset multi-dimensional screening rule to obtain a target agent candidate set; based on the associated task list, a master-slave collaborative agent group is formed from the target agent candidate set according to a preset collaborative group configuration rule, the standardized feature table and the associated task list are distributed to the corresponding master and slave agents, and a real-time data sharing channel between the master and slave agents is established;

[0012] S40, real-time collection of the execution and data sharing state of the master-slave collaborative agent group; if the state triggers a preset abnormal condition, the standardized feature table is synchronized with the processed data, a standby agent is dispatched or a standby data channel is switched to restore the collaboration; at the same time, the upper planning database is linked, the task processing result of the agent group is checked for compliance according to the standardized feature table, and it is ensured that the upper planning constraints are met.

[0013] The present application also relates to a spatial task agent matching and distribution system based on spatial features, comprising:

[0014] The task analysis and associated task generation module is configured to receive and analyze a user-input planning space task, convert the planning space task into a standardized feature table containing entity information, task attributes, and required space data, call a preset GIS database based on the space range in the standardized feature table, generate a space scene label for the planning space task, and extract associated tasks based on preset association mining rules and task attributes in the standardized feature table to obtain an associated task list.

[0015] The scenario-adaptive intelligent agent preliminary screening module is configured to construct an intelligent agent pool and an adaptive rule library, wherein the intelligent agent pool stores a plurality of special intelligent agents, each of which is configured with a preset intelligent agent attribute, and the adaptive rule library is preset with a plurality of matching rules for space scenarios and intelligent agents; based on the space scene label of the planning space task, the corresponding screening rule in the adaptive rule library is called to preliminarily screen the special intelligent agents in the intelligent agent pool to obtain a set of scenario-adaptive intelligent agents.

[0016] The special intelligent agent re-screening and collaborative construction module is configured to re-screen the set of scenario-adaptive intelligent agents according to the standardized feature table based on a preset multi-dimensional screening rule to obtain a target intelligent agent candidate set; based on the associated task list, a master-slave collaborative intelligent agent group is constructed from the target intelligent agent candidate set according to a preset collaborative group configuration rule, the standardized feature table and the associated task list are distributed to the corresponding master and slave intelligent agents, and a real-time data sharing channel between the master and slave intelligent agents is established.

[0017] The abnormality monitoring and compliance verification module is configured to collect the execution and data sharing state of the master-slave collaborative intelligent agent group in real time; if the state triggers a preset abnormal condition, the standby intelligent agent is dispatched or the standby data channel is switched to restore the collaboration based on the synchronized processed data of the standardized feature table; at the same time, the upper-level planning database is linked to verify the compliance of the task processing result of the intelligent agent group according to the standardized feature table, to ensure compliance with the upper-level planning constraints.

[0018] The 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, and the processor implements the steps of the above-mentioned space task intelligent agent matching and distribution method based on space features when executing the computer program.

[0019] The application also relates to a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the above-mentioned space task intelligent agent matching and distribution method based on space features when executed by a processor.

[0020] The space task intelligent agent matching and distribution method and system based on space features provided by the above embodiments have the following beneficial effects:

[0021] Through the scheme, the planning space task can be converted into a standardized feature table and a space scene label and an associated task list are generated, the problems of unstructured task analysis and insufficient associated task mining in the prior art are solved, and accurate basis is provided for agent matching; through initial screening based on the space scene label and multi-dimensional rescreening, combined with the establishment of the master-slave collaborative agent group and the establishment of the real-time data sharing channel, the defects of extensive agent screening and chaotic collaboration are overcome, and the matching accuracy of space tasks and agents and the multi-agent collaboration efficiency are significantly improved; at the same time, through real-time state monitoring, abnormal recovery mechanism and compliance verification linked with the upper planning database, the stability of the task execution process and the compliance of the results are ensured, the error rate and the risk of violation of the space task processing are effectively reduced, and the intelligent, accurate and efficient level of the space task processing in the field of urban planning and the like is overall improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 A flowchart of a space task agent matching and distribution method based on space features provided by the embodiment of the present application is shown in the figure.

[0023] Figure 2 A principle block diagram of a computer device provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings.

[0025] Referring to Figure 1 One of the embodiments of the present application provides a space task agent matching and distribution method based on space features, including the following steps:

[0026] S10, receiving and analyzing a planning space task input by a user, converting the planning space task into a standardized feature table containing entity information, task attributes and required space data, calling a preset GIS database based on the space range in the standardized feature table, generating a space scene label of the planning space task, and extracting associated tasks based on a preset associated mining rule and the task attributes in the standardized feature table to obtain an associated task list.

[0027] In this embodiment, step S10 is the basic link for standardized analysis and key feature extraction of user input planning space tasks. Among them, the planning space task refers to the planning or execution demand proposed by the user involving a specific space range, such as the construction planning of a certain plot in the city, the emergency space resource allocation of sudden disasters, etc.; receiving and analyzing refers to the system obtaining the task information input by the user (which may be in the form of unstructured text description, drawings or parameter table, etc.), and converting it into structured data that the system can recognize and process. The reason for this operation is that the original task form is diverse, and directly used for agent matching will lead to information transmission distortion. The effect is to unify the input format of the task, laying the foundation for subsequent processing.

[0028] The standardized feature table is a structured carrier integrating the core information of the task, including entity information, task attributes and required space data; among them, the entity information refers to the specific spatial objects involved in the task (such as buildings, roads, plots, etc.), the task attribute refers to the non-spatial characteristics of the task (such as task type, execution priority, time constraint, etc.), and the required space data refers to the geographic information on which the task depends (such as latitude and longitude range, terrain slope, surrounding facility distribution, etc.). Its role is to systematically integrate the multi-dimensional information of the task, avoiding the omission of key information.

[0029] The preset GIS database is a pre-constructed geographic information system database for storing various geographic spatial data related to the spatial task, mainly including spatial range data (such as plot boundary, latitude and longitude coordinates, regional division, etc.), terrain data (such as elevation, slope, slope direction, land cover type, etc.), spatial attribute data (such as land use type, planning district, historical protected block range, ecological sensitive area boundary, etc.). The database is pre-configured based on industry general geographic information standards, historical geographic data and dynamically updated spatial information to ensure the accuracy and timeliness of the data. In step S10, the purpose of calling the database is to extract the corresponding geographic environment feature data (such as whether the plot is located in a mountainous area, whether it belongs to a commercial planning area, etc.) through the spatial range in the standardized feature table (such as the latitude and longitude boundary of a plot), to provide data support for generating a spatial scene label that accurately reflects the spatial environment of the task, and to enable the subsequent agent to match the spatial constraint conditions that fully adapt to the task. It should be noted that the pre-constructed GIS database involves obtaining the corresponding data through the Internet from the corresponding official website, industry platform, etc., and the database construction method is also a conventional data processing method familiar to those skilled in the art. The present application does not improve the specific technical means of data integration and arrangement, which is clear to those skilled in the art, and therefore will not be described here.

[0030] The associated tasks are extracted based on the preset association mining rules and the task attributes in the standardized feature table, and an associated task list is obtained. Since there are often associated tasks (such as pipeline laying tasks associated with building planning) before, during or after the space task, these associated relationships can be mined through this operation, which avoids the disconnection of task execution caused by isolated task processing and provides associated task basis for multi-agent collaboration. The preset association mining rules are a set of rules that the system pre-sets to identify the associated relationship with the current planning space task, mainly including two types of space topology association rules and task function dependency rules. The space topology association rules are used to identify tasks with topological relationships such as containing, overlapping or adjacent to the spatial range of the current task (for example, a certain land construction planning task can be associated with the adjacent traffic dredging task through this rule) based on the spatial range features of the task (such as latitude and longitude boundaries, land contour). The task function dependency rules are used to identify tasks with pre-execution (such as geological survey before building construction), post-support (such as garbage disposal system after residential construction) or synchronous collaboration (such as synchronous road construction and underground pipeline laying) relationship with the current task based on task attributes (such as task type, execution target). These rules are pre-configured based on historical task association data, industry planning specifications, etc., to ensure that the associated tasks affecting the execution of the current task can be fully and accurately mined, providing complete task association basis for subsequent multi-agent collaboration processing.

[0031] S20, constructing an agent pool and an adaptation rule library, wherein the agent pool stores a plurality of special agents, each of the special agents is configured with a preset agent attribute, and the adaptation rule library is pre-configured with a plurality of matching rules of space scenes and agents; based on the space scene label of the planning space task, a corresponding filtering rule in the adaptation rule library is called to preliminarily screen the special agents in the agent pool, and a scene-adapted agent set is obtained.

[0032] In this embodiment, step S20 is a preliminary screening step for selecting an adapted agent for a planning space task, which aims to narrow the scope of agents and retain candidate objects that are basically adapted to the task space scene. The construction of the agent pool and the adaptation rule library is the basic preparation work of this step; the agent pool refers to a set of storing a plurality of special agents, and the special agent is a functional module with specific space task processing capability (for example, an agent good at historical street block reconstruction planning, an agent focusing on mountain area emergency deployment, etc.). The preset agent attributes configured for each special agent include the space scene type it is adapted to, the historical task success rate, the complexity of the space range it can handle, the average execution period, etc. These attributes are the core reference basis for matching agents and tasks, and their role is to standardize the capability characteristics of the agents, so as to facilitate attribute comparison and screening by the system.

[0033] The adaptation rule base is a pre-constructed rule set, which is pre-provided with corresponding matching rules of different space scene labels and agent attributes (for example, the "urban core area + commercial planning" scene corresponds to the agent attribute of "adaptation to commercial area space features, historical commercial planning task success rate ≥ 90%", and the "mountainous area + emergency rescue" scene corresponds to the agent attribute of "adaptation to complex terrain, response delay ≤ 10 minutes"). These rules are pre-configured based on industry task processing standards and historical matching data, to ensure the basic adaptability of the space scene and the agent.

[0034] Based on the space scene label of the planning space task, the corresponding screening rule in the adaptation rule base is called, that is, the space scene label (such as "historical and cultural street + flat land") generated by S10 is used to match the corresponding screening condition from the adaptation rule base, to compare the attributes of the special agent in the agent pool, to retain the agent whose attributes meet the screening condition, and to form a scene adaptation agent set. The reason for this operation is that the space scene is the most core constraint condition of the task, and through the preliminary screening, the obviously unsuitable agents (such as the agents suitable for flat land are excluded from the screening range of the mountainous area task) can be quickly excluded, reducing the calculation amount of subsequent rescreening; the effect is to ensure that the agents entering the next link all meet the basic space scene requirements of the task, laying a high-efficiency foundation for accurate matching. It should be noted that the construction of the agent pool (such as the access of the agent, the entry of the attribute parameter) and the rule configuration of the adaptation rule base all use conventional technical means in the field, and the underlying technology of the present application is not improved. The skilled person can realize it based on the existing technology, and will not be described here.

[0035] S30, according to the standardized feature table, the scene adaptation agent set is rescreened according to the pre-set multi-dimensional screening rule, to obtain a target agent candidate set; based on the associated task list, a master-slave collaborative agent group is formed from the target agent candidate set according to the pre-set collaborative group configuration rule, the standardized feature table and the associated task list are distributed to the corresponding master and slave agents, and a real-time data sharing channel between the master and slave agents is established.

[0036] In the embodiment, step S30 is the core link for realizing accurate matching and collaborative establishment of agents on the basis of preliminary screening, aiming to select optimal candidate objects from the scene-adaptive agent set and construct an efficient agent team. In this regard, re-screening of the scene-adaptive agent set according to the standardized feature table based on the preset multi-dimensional screening rule is a fine optimization of the preliminary screening result; the preset multi-dimensional screening rule refers to a multi-dimensional evaluation standard set based on the core requirements of the task, which usually includes a spatial adaptation dimension (such as the matching degree of the spatial range complexity handled by the agent and the task spatial range), a task capability dimension (such as the satisfaction degree of the agent to the task attributes, such as priority response speed and professional field matching degree), a time efficiency adaptation dimension (such as the matching degree of the historical execution cycle of the agent and the task time constraint), etc. The evaluation parameters of these dimensions are all from the standardized feature table generated in S10 (such as the priority in the task attribute and the range complexity in the required spatial data). Through the rule, the scene-adaptive agent set is quantitatively scored and compared, and the agent with the highest comprehensive score is retained to form a target agent candidate set. The reason for this operation is that the preliminary screening only meets the basic adaptation of the spatial scene, and the specific attributes and constraints of the task need to be further matched through multi-dimensional re-screening to improve the matching accuracy; the effect is to ensure that the candidate agent is highly adapted to the task in terms of space, capability, and time efficiency, providing high-quality objects for collaborative establishment.

[0037] Based on the associated task list, the master-slave collaborative agent group is established from the target agent candidate set according to the preset collaborative group configuration rule, which is the key to realize multi-task collaborative processing; the preset collaborative group configuration rule refers to the rule for dividing the roles of master and slave agents, which usually includes master agent selection criteria (such as being good at core task decision-making and having the highest historical collaborative success rate) and slave agent allocation criteria (such as matching corresponding professional agents according to the type of associated tasks and allocating them according to the task dependency order), for example, when the core task is "plot building planning", the master agent is responsible for overall scheme decision-making, and the slave agent is responsible for associated tasks such as "pipeline planning" and "green design". The standardized feature table and the associated task list are distributed to the corresponding master and slave agents to ensure that each agent is clear about the overall task and its own division of labor; a real-time data sharing channel is established between the master and slave agents to realize instant synchronization of task data (such as real-time synchronization of scheme adjustment of the master agent to the slave agent), and the connectivity of the channel is based on the preset verification rule of the task attribute (such as data sensitivity). The reason for this part of the operation is that the spatial task and the associated task need to be promoted collaboratively, and the clear master-slave relationship and data interaction mechanism can avoid confusion in division of labor and data islands; the effect is to form an efficient collaboration mode of core decision-making + associated execution, and to improve the coordination and synchronization of multi-task processing.

[0038] It should be noted that the specific quantification algorithm of multi-dimensional screening, the underlying technology (such as communication protocol) of data sharing channel, etc. adopt the conventional technical means in the art, and the present application does not improve them. The skilled person can realize based on the existing technology, and here is not expanded.

[0039] S40, real-time acquisition of the execution and data sharing state of the master-slave collaborative intelligent agent group; if the state triggers a preset abnormal condition, synchronizing the processed data based on the standardized feature table, scheduling a standby intelligent agent or switching a standby data channel to restore collaboration; at the same time, the upper planning database is linked, the task processing results of the intelligent agent group are checked for compliance according to the standardized feature table, and it is ensured that the upper planning constraints are met.

[0040] In the present embodiment, step S40 is the key link to ensure the stability of the task execution of the master-slave collaborative intelligent agent group and the compliance of the results, aiming to monitor the execution process in real time, handle exceptions in time and ensure that the task results meet the planning requirements. Real-time acquisition of the execution and data sharing state of the master-slave collaborative intelligent agent group is a dynamic tracking of the task advancement process; the contents of acquisition include the task execution progress of the master-slave intelligent agent (such as the core task completion ratio, the associated task node achievement situation), the integrity (such as whether there is data packet loss) and timeliness (such as whether the data synchronization delay is over-standard) of data sharing, etc. These state data are real-time returned through the preset monitoring interface, which provides real-time basis for abnormal identification and compliance checking.

[0041] If the state triggers a preset abnormal condition, synchronizing the processed data based on the standardized feature table, scheduling a standby intelligent agent or switching a standby data channel to restore collaboration is a safeguard mechanism to deal with the interruption of execution; the preset abnormal condition refers to the abnormal situation that may affect the task execution (such as master agent failure, data sharing channel interruption, slave agent execution timeout, etc.); synchronizing the processed data based on the standardized feature table is to ensure the continuity of the task progress and data state after recovery (such as synchronizing the "pipeline planning data" processed in the feature table to the standby intelligent agent when slave agent A fails); scheduling a standby intelligent agent or switching a standby data channel is a targeted measure according to the type of abnormality (such as enabling a pre-provisioned intelligent agent when an intelligent agent fails, switching to a redundant communication link when a channel is interrupted). The reason for this operation is that the multi-agent collaborative process is easily affected by hardware and software failures and data transmission problems, and needs to be quickly recovered to avoid task delay; the effect is to minimize the interference of abnormality on task execution and ensure the continuity of the collaborative process.

[0042] At the same time, the upper planning database is linked to check the compliance of the task processing result of the agent group according to the standardized feature table, which is a necessary link to ensure the legality of the task result; the upper planning database is a database for storing overall planning constraints of the region, including land use planning, ecological protection sensitive line, building density limit, historical and cultural protection requirements and other upper rules; the check basis is the task attribute (such as the allowed construction type of the planning plot) in the standardized feature table and the required spatial data (such as the planning partition corresponding to the plot coordinates), and whether it is compliant is judged by comparing the agent processing result (such as building height, volume rate) with the constraint condition in the database. The reason for this operation is that the spatial task needs to comply with the overall planning of the region to avoid conflicts between local tasks and upper requirements; the effect is to eliminate illegal results from the source and ensure the legality and rationality of task execution.

[0043] It should be noted that the sensor interface technology for state acquisition, the basic protocol for data synchronization, and the interface specification for database linkage all use conventional technical means in the art. The underlying technology of the present application is not improved, and the skilled person can implement it based on existing technology, which will not be described here.

[0044] In one embodiment, step S10 specifically includes the following steps:

[0045] S101, based on the required spatial data in the standardized feature table, calling a preset spatial topology association rule, identifying potential associated tasks with a topology relationship of containing, adjacent or overlapping with the spatial range of the planning spatial task, and generating a first potential associated task set;

[0046] S102, based on the task attribute in the standardized feature table, calling a preset task function dependency rule, identifying potential associated tasks with a function dependency relationship of pre-execution, post-support or synchronous collaboration with the planning spatial task, generating a second potential associated task set, and combining the second potential associated task set with the first potential associated task set to obtain a comprehensive potential associated task set;

[0047] S103, calling a preset upper planning check rule, linking an upper planning database to check the compliance of the comprehensive potential associated task set, eliminating tasks that do not comply with the upper planning constraints, and integrating the remaining tasks to obtain the associated task list.

[0048] In the present embodiment, step S101 is the core link for mining potential tasks associated with the current planning space task from the spatial range dimension, and the precise extraction of the first set of potential associated tasks is achieved through the process of "data calling → topology comparison → task labeling → set generation". Among them, the required spatial data (such as latitude and longitude boundary, land polygon coordinates, etc.) in the standardized feature table is the basis for triggering spatial association recognition. The system first extracts this data and converts it into a standardized spatial geometric model (such as a vector polygon); then calls the spatial range data (also converted into a geometric model) of other spatial tasks stored in the preset GIS database as the comparison object.

[0049] The preset spatial topology association rule is executed through specific logic in this link: the system performs topology relationship calculation (such as judging whether there is a containing relationship - the current land completely covers an auxiliary facility land, or a neighboring relationship - the distance between the boundaries of two lands is ≤50 meters, or an overlapping relationship - the spatial range coincidence degree of two lands is ≥30%) between the spatial geometric model of the current task and the spatial geometric model of other tasks, and filters out the tasks that meet the conditions according to the threshold (such as the adjacent distance threshold, the overlap threshold) preset by the rule. For example, if the current task is "construction planning of a certain commercial land", the system finds through topology calculation that "there is a residential land within 50 meters", and the "noise protection planning" task of the residential land is marked as a potential associated task because it meets the "adjacent relationship".

[0050] After the above comparison and screening, all the marked tasks are integrated into the first set of potential associated tasks, and this set only contains tasks associated by spatial topology relationship. The reason for this operation is that the quantitative comparison of spatial topology relationship can avoid the subjectivity of manual judgment and ensure the objectivity of associated tasks; the effect is that through the clear "data conversion → model comparison → threshold screening" process, the generation process of the first set of potential associated tasks is traceable and the results are verifiable, solving the problem of fuzzy spatial association task recognition and providing a precise spatial dimension basis for subsequent comprehensive mining.

[0051] Step S102 is the key link for supplementing associated tasks from the functional dependence dimension and integrating "spatial + functional" dual-dimension associated tasks, and the construction of the comprehensive set of potential associated tasks is achieved through the process of "task attribute extraction → functional dependence matching → second task set generation → dual-dimension task set merging". Among them, the task attributes in the standardized feature table need to extract key information related to functional association, including task type (such as building planning, pipeline laying, geological survey, etc.), execution target (such as land development, emergency support, etc.), professional field (such as civil engineering, environmental protection, transportation, etc.), which are the core basis for matching functional dependence relationship, for example, the task type of "building planning" directly corresponds to the association logic of "function tasks such as pipeline, greenery, etc. that need to be matched".

[0052] The preset task function dependency rule is executed in the manner of "attribute matching → rule calling → task screening": first, the system matches the corresponding function dependency items (such as the rule item of "pre-geological survey, synchronous-pipeline planning, post-landscaping design" matched with the "building planning" task) from the rule library according to the extracted key task attributes; then, based on these items, the system screens the tasks having "pre-execution (such as the geological survey needing to be completed before the building construction), post-support (such as the landscaping design needing to be carried out after the building is completed), synchronous coordination (such as the road construction and underground pipeline laying being carried out synchronously)" relationship with the current task from the system task library, and marks them as potential associated tasks of the function dimension.

[0053] After the above screening, all the marked function associated tasks are integrated into a second potential associated task set, which only covers the tasks having function logical dependency with the current task, and is complementary to the first potential associated task set (spatial dimension). In the merging link, the second potential associated task set needs to be compared with the first potential associated task set, first, the repeated tasks in the two sets are removed (such as a task meeting both the spatial adjacent relationship and the function synchronous coordination relationship, only one is retained), then the tasks after deduplication are integrated into a comprehensive potential associated task set, ensuring that the set covers both two-dimensional association and no redundant tasks.

[0054] The reason for this operation is that the associated tasks only relying on the spatial dimension may miss the key association in the function logical level (such as a certain adjacent task of a plot having no function with the current task, while a certain long-distance task needs to be coordinated synchronously), and direct merging may cause repeated tasks; the effect is that through "precise matching of function dependency + deduplication and merging", the comprehensive potential associated task set not only realizes the comprehensive coverage of "space + function", but also avoids task redundancy, provides a complete and simplified associated task basis for subsequent compliance verification, and makes the associated task mining process traceable and the results verifiable.

[0055] Step S103 is a key link of performing compliance verification on the comprehensive potential associated task set and generating a list of legal and executable associated tasks, which realizes the compliance screening of the associated tasks through the process of "rule calling → data linkage → task-by-task verification → list generation". Among them, the preset upper-level planning verification rule is a classification verification standard based on the constraint types in the upper-level planning database, mainly including three categories: land use type rules (such as "commercial planning area cannot be included in industrial production type associated tasks"), ecological protection type rules (such as "large-scale earthwork construction type associated tasks cannot be included in ecological protection sensitive areas"), and construction index type rules (such as "the volume rate of the building construction task associated with the residential plot cannot exceed the upper limit of 2.5"), these rules correspond one-to-one with the upper-level planning constraints, ensuring the pertinence of the verification standard.

[0056] When linking to the upper-level planning database, the system first extracts the corresponding upper-level planning constraint data (not global data) in the spatial range (such as latitude and longitude boundary) of the current planning space task based on the standardized feature table, for example, the upper limit of construction density corresponding to the "residential land planning district" to which the current land belongs, the specific range of the surrounding ecological protection sensitive area, etc. These data serve as the core basis for task-by-task verification, avoiding verification bias caused by irrelevant planning data.

[0057] The compliance verification is performed in a "task-by-task comparison" manner: first, extract the key information of each task in the comprehensive potential associated task set (including the spatial range of the task, the task type, the execution parameters, etc., such as the construction range and pipe diameter specification of a certain "pipeline laying" task); then match and verify these information with the extracted upper-level planning constraints, for example: if the construction range of a certain associated task is partially located within the ecological protection sensitive area and does not meet the "allowable small pipeline laying threshold (pipe diameter ≤ 300mm)" in the "ecological protection rules", it is determined as non-compliant; if the construction density (such as the auxiliary facility construction density) of a certain associated task exceeds the constraint of "residential land construction density ≤ 30%" in the land use rules, it is also determined as non-compliant. All non-compliant tasks will be directly excluded, and the remaining compliant tasks need to be classified and sorted according to "task type (core / associated)" or "execution order (pre / post / synchronous)", and finally integrated to form an associated task list, which also needs to mark the corresponding compliance verification result of each task (such as "comply with residential land construction indicators") for subsequent agent execution.

[0058] The reason for this operation is that the comprehensive potential associated task set only covers "space + function" association and does not consider upper-level planning constraints, which may have the risk of "task legal but conflicting with regional planning" if directly used (such as the associated task type does not match the land use planning); the effect is to ensure that the final associated task list not only fully covers the association requirements, but also fully meets the upper-level planning requirements, avoiding rework due to task violations during subsequent agent collaborative execution, and providing pre-conditions for the legal landing of tasks.

[0059] It should be noted that the specific identification algorithm of the spatial topological relationship, the logical matching bottom technology of the task function dependence, and the calling interface technology of the upper planning data, the basic comparison algorithm of the task-by-task verification, etc. All adopt the conventional technical means in the art, and the present application does not improve them. The skilled person can realize them based on the existing technology, and this will not be expanded here. Steps S101-S103 are mined by double-dimensional associated tasks through "spatial dimension topology mining + function dimension dependence matching + de-duplication and merging", combined with upper planning task-by-task accurate verification, screening out compliant tasks, solving the problems of incomplete, redundant and non-compliant associated task mining in S10, further improving the accuracy, simplicity and legality of the associated task list, and providing a more reliable associated task basis for subsequent master-slave collaborative agent group task allocation.

[0060] In one of the embodiments, step S20 specifically comprises the following steps:

[0061] S201, based on the space scene label, calling the space scene subdivision rule in the adaptive rule library, decomposing the space scene label into a combination label containing space function partition and space constraint level, to obtain a subdivided space scene label;

[0062] S202, calling the mapping rule of subdivided scene and agent attribute level in the adaptive rule library, screening out the special agent with special agent attribute matched with the subdivided space scene label from the agent pool, and generating a preliminary scene adaptive agent set;

[0063] S203, if the preliminary scene adaptive agent set is empty, triggering a pre-equipment mechanism, and determining the non-empty agent set obtained by the pre-equipment mechanism as the scene adaptive agent set, wherein the pre-equipment mechanism includes at least one of the following operations: sending an agent attribute upgrade prompt to the system and calling a cross-scene agent collaboration rule to generate a temporary collaborative adaptation scheme.

[0064] In the embodiment, step S201 is a step of fine disassembling the space scene label of the planning space task, aiming to improve the accuracy of the initial screening of the agent by subdividing the label, and the construction of the subdivided space scene label is realized through the process of "label calling → rule matching → dimension disassembling → subdivision generation". Among them, the space scene label refers to the characteristic identifier reflecting the task space environment generated by S10 (such as "urban core area + flat land", "mountainous area + ecological protection area", etc.), which is the basis for triggering the subdivision operation; the space scene subdivision rules in the adaptive rule base are pre-set rules for disassembling general scene labels into multi-dimensional sub-labels, which are designed based on the core constituent elements (functional attributes and constraint strength) of the space scene, for example, the urban core area can be subdivided into "commercial core area" and "administrative core area" according to function, and into "high constraint (including historical blocks)" and "medium constraint (regular commercial area)" according to constraint strength.

[0065] The space function partition refers to the subdivision of the space scene from the perspective of functional attributes (such as "commercial function area", "ecological function area", "traffic hub area", etc.), which reflects the dominant use of the space; the space constraint level refers to the subdivision from the perspective of constraint strength (such as "high constraint" involving historical protection, ecological sensitivity, high-density flow, etc.; "medium constraint" regular planning area; "low constraint" blank area to be developed, etc.), which reflects the degree of restriction of the space on task execution.

[0066] The specific disassembling process is: the system first calls the adaptive rule base, matches the corresponding subdivision rules based on the current space scene label (such as "urban core area + flat land"); then disassembles the label into a combination of "space function partition + space constraint level" according to the rules, for example, disassembles "urban core area + flat land" into a subdivided space scene label of "commercial core area (space function partition) + high constraint (space constraint level, because it contains historical commercial street)". The reason for this operation is that the original space scene label is general (such as "urban core area" cannot distinguish the functional differences between commercial and administrative areas, or whether it contains high constraint areas), which is directly used for agent screening and is easy to cause adaptive deviation; the effect is to make the characteristics of the space scene label more specific through multi-dimensional disassembly, to provide detailed basis for the accurate matching of agent attributes and subdivided scenes in S202, and to avoid initial screening errors caused by ambiguous scene description.

[0067] Step S202 is the core step of screening adaptive agents from the agent pool based on the subdivided space scene label. The accurate screening of the preliminary scene-adaptive agent set is achieved through the process of "subdivision label extraction → mapping rule calling → attribute comparison → set generation". Among them, the mapping rule of the subdivided scene and the agent attribute level in the adaptive rule library is a rule that predefines the corresponding association between the two dimensions of the subdivided space scene label (spatial function partition, spatial constraint level) and the agent attribute, including: the mapping of spatial function partition and agent professional field (such as "commercial function area" corresponding to "agent attribute with commercial planning and passenger flow analysis ability", "ecological function area" corresponding to "agent attribute with ecological protection and low-impact development ability"); the mapping of spatial constraint level and agent ability level (such as high constraint corresponding to "agent attribute level of historical protection project experience ≥ 5 items and compliance verification pass rate ≥ 95%", and medium constraint corresponding to "attribute level of regular planning project experience ≥ 10 items").

[0068] The specific screening process is as follows: the system first extracts the spatial function partition (such as "commercial core area") and the spatial constraint level (such as "high constraint") in the subdivided space scene label generated in S201; then calls the corresponding mapping rule in the adaptive rule library to extract the attributes (including professional field, historical task experience, constraint adaptation ability, etc.) of each special agent from the agent pool; through attribute comparison, the agents that meet the conditions of "professional field matching spatial function partition" and "ability level meeting spatial constraint level" are retained, for example, in the "commercial core area + high constraint" scene, only the agents that are good at commercial planning and have historical street reconstruction experience are retained.

[0069] After the above comparison, all agents that meet the conditions are integrated into the preliminary scene-adaptive agent set. The reason for this operation is that the subdivided space scene label has clearly defined the function and constraint details of the scene, and the accurate attribute mapping rule is needed to ensure that the agent's ability is highly matched with the scene's demand, avoiding the problem of "function adaptation but insufficient constraint adaptation" (such as being good at commercial planning but having no experience in high-constraint areas) in the preliminary screening; the effect is that through "double-dimensional attribute matching", the agents in the preliminary set meet the scene requirements in terms of function adaptation and constraint adaptation, providing high-quality candidates for subsequent re-screening, and making the screening process quantifiable and the results verifiable.

[0070] Step S203 is a special processing link to cope with the initial scene adaptation agent set being empty, aiming to ensure the effectiveness of the initial screening result of the agent through the pre-equipment mechanism, avoid task interruption due to no adaptation agent, and realize the bottom protection of the scene adaptation agent set through the process of "empty set judgment → mechanism triggering → scheme execution → set determination". Among them, the initial scene adaptation agent set being empty means that no special agent meeting the subdivided space scene label is obtained after S202 screening (for example, in the "high constraint + ecological function area" scene, there is no agent in the agent pool that meets the ecological professional ability and high constraint adaptation experience at the same time); the pre-equipment mechanism is a remedial scheme configured in advance to solve the empty set problem, including two core operations:

[0071] Sending agent attribute upgrade prompt to the system: referring to the system automatically pushing the attribute requirements that need to be supplemented (such as "need to add ecological high constraint area planning experience") to the agent management module, prompting to upgrade the ability of the existing agent (such as improving its constraint adaptation ability through algorithm iteration or data training), and after the upgrade is completed, it is included in the set;

[0072] Calling cross-scene agent collaboration rules to generate a temporary collaborative adaptation scheme: referring to selecting multiple agents adapted to different scenes to form a temporary collaboration group (such as "ecological function area adaptation agent + high constraint area adaptation agent" collaboration, the former is responsible for ecological function planning, and the latter is responsible for high constraint compliance verification), meeting the subdivided scene demand through complementary ability.

[0073] When executed specifically, the system can select one or a combination of operations according to the urgency of the task (such as calling cross-scene collaboration rules to quickly generate a temporary scheme for urgent tasks, and triggering attribute upgrade prompts to perfect the agent pool for non-urgent tasks), and finally determine the non-empty agent set obtained through the backup mechanism as the scene adaptation agent set. The reason for this operation is that the subdivided scene may result in no result from the initial screening due to strict constraints or insufficient coverage of the agent pool, which will directly block the subsequent task if not handled; the effect is to ensure that whether the initial screening is empty or not, an effective scene adaptation agent set can be obtained through the bottom-up scheme of "active upgrade + complementary collaboration", which not only guarantees the continuity of task advancement, but also expands the flexibility of agent adaptation through cross-scene collaboration, providing a necessary candidate basis for subsequent rescreening.

[0074] It should be noted that the space scene tag subdivision algorithm, the basic comparison logic of the agent attribute and the scene tag, and the agent upgrade communication interface in the pre-device mechanism and the basic protocol of cross-scene collaboration are all based on conventional technical means in the field, and the present application does not improve them. The skilled person can implement them based on the existing technology, and this will not be described in detail. Steps S201-S203 solve the problems of general scene tag screening, insufficient agent matching accuracy and no processing scheme for empty set in S20 through the progressive process of "scene tag fine disassembly → two-dimensional agent attribute matching → empty set bottom-up guarantee", which not only improves the accuracy of the scene adaptation agent set, but also ensures the effectiveness of the screening result, providing a high-quality and reliable candidate agent basis for subsequent multi-dimensional rescreening in S30.

[0075] In one of the embodiments, step S30 specifically includes the following steps:

[0076] S301, extracting rescreening core parameters from the standardized feature table, the rescreening core parameters at least including spatial range complexity of planning space tasks, task attribute weight and time constraint threshold;

[0077] S302, constructing a three-dimensional attribute evaluation matrix for each special agent in the scene adaptation agent set, the three-dimensional attribute evaluation matrix including a spatial adaptation dimension, a task capability dimension and a time efficiency adaptation dimension; the spatial adaptation dimension is the matching degree based on the historical service space range in the agent attribute and the current task space range complexity; the task capability dimension is the weighted matching based on the special task success rate in the agent attribute and the task attribute weight in the standardized feature table; the time efficiency adaptation dimension is the difference based on the average task completion cycle in the agent attribute and the time constraint threshold;

[0078] S303, applying a preset dynamic weight algorithm to quantitatively score the three-dimensional attribute evaluation matrix to generate a comprehensive adaptation score of each special agent;

[0079] S304, screening the special agent with a comprehensive adaptation score greater than or equal to a preset threshold to form a target agent candidate set; if there are special agents with the same comprehensive adaptation score, the special agent with the lowest historical collaborative conflict rate is preferentially retained, and the historical collaborative conflict rate is the proportion of the number of task conflicts when the special agent collaborates with other special agents to the total number of collaborations.

[0080] In this embodiment, step S301 is the parameter preparation stage of the rescreening link, aiming to extract the core quantitative indicators of multi-dimensional evaluation of the agent from the standardized feature table, to provide a data basis for the construction of the three-dimensional attribute evaluation matrix, and to realize the accurate acquisition of key parameters through the process of "feature table analysis → core parameter positioning → parameter extraction". Among them, the standardized feature table is the structured carrier of the integrated task core information generated by S10, and its spatial data and task attributes are the direct source of rescreening core parameters; the rescreening core parameters include spatial range complexity (based on terrain complexity such as slope ≥ 25° area proportion, spatial range area such as ≥ 100,000 square meters for large, surrounding facility density comprehensive quantification, describing the complexity of the task space environment, such as "mountainous area + large area + facility intensive" is judged as "high complexity"), task attribute weight (based on priority such as "emergency" 0.4, professional field such as "ecological protection" 0.3, execution target such as "compliance" 0.3 Set the weighting coefficient, the sum is 1, reflecting the focus of the task on the agent's ability), time constraint threshold (the latest node or the longest period that the task must be completed, such as "within 30 working days", based on the time constraint field of the task attribute extraction).

[0081] The specific extraction process is: the system first analyzes the structured fields of the standardized feature table, and locates the items corresponding to "spatial range complexity", "task attribute weight", and "time constraint threshold" (such as the "terrain and range description" field in the required spatial data, the "priority and time requirement" field in the task attribute); Then through the preset parameter extraction rule (such as quantifying "terrain and range description" to get the spatial range complexity, and assigning values to attributes such as "priority" according to the preset weight table), the non-structured description is converted into calculable quantitative parameters (such as spatial range complexity represented by 0-10 points, 10 points for the highest). The reason for this operation is that rescreening needs to develop quantitative evaluation standards based on the core constraints and requirements of the task, and the information in the original feature table needs to be converted into unified parameters to support the subsequent comparison of agent attributes; The effect is to provide clear quantitative benchmarks for the construction of the three-dimensional attribute evaluation matrix in S302 by extracting the core parameters of these three dimensions, ensuring the objectivity and pertinence of rescreening evaluation, and avoiding evaluation deviation caused by ambiguous parameters.

[0082] In this embodiment, step S302 is the core link of multi-dimensional capability evaluation of each special agent in the scene-adapted agent set, aiming to build a three-dimensional attribute evaluation matrix, integrate the adaptation characteristics of agents and task requirements, and provide structured basis for quantitative scoring. It is realized through the process of "agent traversal → dimension data extraction → matrix construction". The three-dimensional attribute evaluation matrix is a structured carrier containing spatial adaptation, task capability, and time efficiency adaptation dimensions, corresponding to the complex screening core parameters of S301 respectively, realizing the accurate benchmarking of agent capability and task demand: the spatial adaptation dimension takes "spatial range complexity" as the benchmark to calculate the matching degree of the agent's "historical service spatial range" and the current task (for example, if the task is "high complexity" and the agent's high complexity scene proportion is ≥70%, the matching degree is 8 / 10); the task capability dimension takes the task attribute weight as the weighted basis to calculate the weighted matching value of the agent's special task success rate (for example, if the task weight is 0.4 / 0.3 / 0.3 and the agent's corresponding success rate is 95% / 90% / 98%, then the weighted value = 95% × 0.4 + 90% × 0.3 + 98% × 0.3 = 94.4%); the time efficiency adaptation dimension takes "time constraint threshold" as the reference to calculate the difference between the agent's "average task completion period" and the threshold (for example, if the threshold is 30 working days and the agent's period is 25 days, the difference is -5, and if it is 35 days, the difference is +5).

[0083] The specific construction process is: the system first traverses each special agent in the scene-adapted agent set; for a single agent, extract the original data corresponding to the three dimensions in its attributes (such as historical service spatial complexity proportion, attribute-specific success rate, and average completion period); then combine the complex screening core parameters of S301 to get the quantitative results of each dimension according to the above calculation method; finally, integrate the quantitative results of the three dimensions to form the three-dimensional attribute evaluation matrix of the agent (for example: [spatial adaptation 8 points, task capability 94.4%, time efficiency adaptation -5 days]). The reason for this operation is that the scene-adapted agent only meets the basic scene demand, and needs to be further filtered through multi-dimensional quantitative evaluation to select the agent highly adapted to the task core constraints (space, capability, and time efficiency), and the three-dimensional matrix can systematically integrate evaluation indicators to avoid the one-sidedness of single-dimensional evaluation; the effect is to establish a multi-dimensional capability portrait for each agent, providing clear structured data support for the quantitative scoring of S303, ensuring the comprehensiveness and accuracy of the complex screening evaluation.

[0084] In the embodiment, step S303 is a key link for quantitatively integrating the three-dimensional attribute evaluation matrix and generating the comprehensive adaptation score of the agent, aiming to reflect the emphasis of the task on different dimensions through a dynamic weight algorithm, realize the pertinence and accuracy of the score, and realize the process through "dynamic weight determination → matrix value normalization → weighted calculation → score generation". Among them, the preset dynamic weight algorithm is an adaptive scoring rule, which dynamically adjusts the weight proportion of the three-dimensional dimension (non-fixed proportion) according to the type and core demand of the planning space task, for example, in "mountainous emergency rescue planning" (core demand fast response) Time adaptation 0.4, space adaptation 0.35, task ability 0.25, in "historical block reconstruction planning" (core demand space adaptation and compliance) Space adaptation 0.4, task ability 0.4, time adaptation 0.2; The dynamic weight depends on the task attributes such as "emergency" "routine" in the standardized feature table, spatial range complexity level, historical similar task optimal configuration, to ensure matching with the actual demand of the task.

[0085] The specific scoring process is: normalizing the values of each dimension in the three-dimensional attribute evaluation matrix (unified into standard scores of 0-100 points, such as "5 days" corresponding to 90 points, "+5 days" corresponding to 60 points in the time adaptation dimension); Based on the preset dynamic weight algorithm, determine the real-time weight of the three dimensions under the current task (such as "4:3.5:2.5" for emergency tasks); Calculate the weighted total score according to the formula "space adaptation score × space weight + task ability score × ability weight + time adaptation score × time weight", and generate the comprehensive adaptation score of each special agent (such as an agent with normalized scores of 80, 90 and 95, calculated according to the emergency task weight: 80×0.35+90×0.25+95×0.4=89.5 points). The reason for this operation is that the core constraints of different space tasks are different (such as emergency tasks with time efficiency and complex space tasks with adaptation), and fixed weight cannot reflect this difference, which will cause the score to be out of touch with the actual demand; The effect is that through the dynamic weight algorithm, the comprehensive adaptation score can accurately reflect the degree of satisfaction of the agent to the core demand of the current task, avoiding the selection of agents with "all-dimension average but core-dimension insufficient", and providing more targeted quantitative basis for the screening of S304.

[0086] In the embodiment, step S304 is the final screening stage of the secondary screening link, aiming to determine high-quality candidate agents based on the comprehensive adaptation score, solve the same score problem through the conflict rate priority rule, generate the target agent candidate set, and achieve the process of "score comparison threshold screening same score processing set generation". Among them, the preset threshold is the critical value of the secondary screening, and the basis includes the lowest comprehensive score of the selected agents of the same type of historical task (such as 80 points or more, the standard rate is greater than or equal to 90%), the importance of the current task (core task 80 points, regular task 70 points), and the overall score distribution of the scene adaptation agent (which can be dynamically adjusted).

[0087] The specific screening process is: the system first compares the comprehensive adaptation score (S303 generates) of each special agent with the preset threshold, retains all agents with a score greater than or equal to the threshold, and forms a preliminary candidate set; if there are agents with the same comprehensive adaptation score in the preliminary candidate set (such as two agents with 85 points), the same score processing mechanism is triggered - the "historical cooperative conflict rate" in the agent attribute is called for secondary screening. The historical cooperative conflict rate is a key indicator to measure the compatibility of agent cooperation, and its calculation method is "the number of task conflicts when the special agent cooperates with other agents / total cooperation times x 100%" (for example, an agent participates in 100 cooperation, and 5 task conflicts occur, then the conflict rate is 5%), the conflict scenarios include task division overlap, data interaction contradiction, and execution rhythm dislocation. The agent with the lowest conflict rate is preferentially retained, for example: agent A (conflict rate 3%) and agent B (conflict rate 8%) with a score of 85 are retained. After the above screening and processing, the agent set finally integrated is the target agent candidate set. The reason for this operation is: only through the comprehensive score screening, the agent with "score up to standard but poor cooperation compatibility" may be missed, and the lack of objective standard in the same score situation easily leads to random selection; the effect is to ensure that the target agent candidate set not only meets the comprehensive ability standard, but also has good cooperation potential through the double mechanism of "threshold screening + conflict rate priority", which provides "high ability + high compatibility" high-quality candidate objects for the formation of master-slave cooperative agent groups in S30, and reduces the potential conflict risk in the cooperation process.

[0088] It should be noted that the extraction algorithm of the re-screening core parameter, the three-dimensional matrix construction logic, the basic weighting method of the dynamic weight, the statistical technique of the historical coordination conflict rate, etc. are all conventional technical means in the field, and the present application does not improve them. The specific numerical values (such as weight, score, and conflict rate) in the explanation are only examples for assisting understanding, and are not a limitation on the present application. The above-mentioned technologies can be realized based on the existing technology, and will not be described here. Steps S301-S304 solve the problems of lack of quantitative standard, single dimension, and no objective basis for the same score in re-screening through the progressive process of "core parameter extraction → three-dimensional matrix construction → dynamic weighting scoring → threshold screening and same score processing". Both the adaptation of the agent to the core requirements of the task through multi-dimensional evaluation and the improvement of collaboration compatibility through conflict rate rules ultimately generate a target agent candidate set with "ability up to standard, accurate adaptation, and optimal collaboration", providing a high-quality basis for the accurate formation of the master-slave collaborative agent group.

[0089] In one of the embodiments, in step S30, the following steps are further included:

[0090] S305, based on the core task weight of the planning space task in the associated task list and the standardized feature table, calling the preset master agent selection rule, screening out the special agent with the highest core task processing efficiency and the historical coordination success rate greater than or equal to the preset threshold from the target agent candidate set, and determining it as the master agent;

[0091] S306, based on the dependency of each associated task in the associated task list and the core task, calling the preset slave agent allocation rule, allocating the associated task with a dependency greater than or equal to the preset dependency threshold to the special agent with a data synchronization delay less than or equal to the preset delay threshold, and allocating the associated task with a dependency less than the preset dependency threshold to the preset slave agent, and determining it as the slave agent;

[0092] S307, distributing the standardized feature table and the associated task list to the corresponding master agent and slave agent respectively, calling the preset channel effectiveness verification rule to build a real-time data sharing channel, and verifying the channel connectivity through bidirectional data transmission verification of the master-slave agent, and completing the formation of the master-slave collaborative agent group after the verification is passed.

[0093] In this embodiment, step S305 is the core link of the master-slave cooperative agent group, aiming to select a master agent with core task leading ability and high collaboration reliability from the target agent candidate set, and is specifically implemented through the process of "core parameter extraction → selection rule calling → dual evaluation of efficiency and success rate → master agent determination". Among them, the associated task list is the legal associated task set generated by S10, which clearly defines the boundaries of core and associated tasks; the core task weight of the standardized feature table quantifies the importance of the core task (such as the weight of "main building layout design" is 0.6); the preset selection rule contains two core indicators:

[0094] The core task processing efficiency is based on the comprehensive quantification of "historical completion efficiency, quality score, and resource matching degree", which is used to measure the core task processing ability; the historical collaboration success rate refers to the proportion of successful collaboration when the agent acts as a master agent (preset threshold 85% to ensure reliability).

[0095] The specific selection process is as follows: the system first extracts the core task definition and core task weight from the associated task list and the standardized feature table (clearly defines the tasks that the master agent needs to focus on); then calls the master agent selection rule to extract the core task processing efficiency data and historical collaboration success rate data of each agent from the target agent candidate set; through dual evaluation, the agent with the highest core task processing efficiency in the candidate set and the historical collaboration success rate ≥ the preset threshold (such as 85%) is selected, for example, the core task processing efficiency of agent A in the candidate set is 92 points (the highest), and the historical collaboration success rate is 90% (≥ 85%), which is determined as the master agent. The reason for this operation is that single-dimensional screening is easy to lead to "high ability and poor collaboration" or "good collaboration and insufficient ability"; the effect is to ensure that the master agent efficiently leads the core task and stably coordinates the slave agent, laying the foundation for subsequent collaboration.

[0096] In this embodiment, step S306 is the hierarchical allocation link of the slave agent in the master-slave cooperative agent group, aiming to match different priority slave agents according to the dependency of associated tasks and core tasks, and to ensure the efficient collaboration of key associated tasks and master agents, and is specifically implemented through the process of "dependency calculation → allocation rule calling → hierarchical matching → slave agent determination". Among them, the dependency of associated tasks and core tasks is a quantitative indicator based on execution timing (such as high pre-dependency and low post-dependency), data interaction frequency (such as high-frequency synchronous dependency), and functional coupling (such as direct input dependency), which is used to measure the influence of associated tasks on core tasks.

[0097] The preset slave agent allocation rule is based on the dependency level and the hierarchical standard of the agent cooperation performance, and includes two core logics: the high dependency associated task (dependency ≥ preset threshold, usually 60%) is allocated to the special agent (the lower the delay, the stronger the real-time cooperation ability) with the master agent "data synchronization delay ≤ preset threshold (such as 50 ms)"; the low dependency associated task (dependency < threshold) is allocated to the pre-equipped slave agent (the basic cooperation agent reserved by the agent pool, avoiding occupying the main cooperation resource).

[0098] The specific allocation process is: the system first calculates the dependency of each associated task in the associated task list to the core task (such as task A dependency 75%, task B dependency 40%); then the slave agent allocation rule is called, and the dependency threshold (60%) and the delay threshold (50 ms) are set; from the target agent candidate set, the agent (such as agent C delay 30 ms, agent D delay 45 ms) with the master agent data synchronization delay ≤ 50 ms is selected, and the high dependency task A is allocated to the agent C; the low dependency task B is allocated to the pre-equipped slave agent E; finally, the agents C and E are determined as slave agents. The reason for this operation is that the influence of associated tasks on core tasks is significantly different, and uniform allocation will lead to "insufficient key task cooperation" or "non-key task resource occupation", and the data synchronization delay directly affects the cooperation quality of high dependency tasks; the effect is to ensure efficient cooperation of high dependency tasks and take over by standby agents for low dependency tasks, which not only guarantees reliability, but also optimizes resource allocation.

[0099] In the embodiment, step S307 is the finishing link of the master-slave cooperative agent group establishment, which aims to ensure the information synchronization and efficient cooperation of the master-slave agents through accurate data distribution and real-time channel building, and is specifically implemented through the process of "data directional distribution → channel rule calling → connectivity verification → group formation". Among them, the standardized feature table and the associated task list are the core data basis: the standardized feature table is preferentially distributed to the master agent (including core task space data, attribute weight, etc.), and the associated task list is distributed to the corresponding slave agent according to the allocation result of S306 (high dependency task details to low delay slave agent, low dependency task to standby slave agent); the preset channel validity verification rule includes bandwidth threshold (such as ≥ 10 Mbps), data encryption protocol (such as TLS1.3), upper limit of packet loss rate (such as ≤ 0.1%), etc., which is used to guarantee the reliability of the channel.

[0100] The real-time data sharing channel is a special link for task state synchronization and data interaction of the master-slave agents, which is based on network identification (such as IP address, port number), and calls communication protocol (such as MQTT, WebSocket) to complete initialization configuration.

[0101] The specific execution process is: data distribution: the system pushes the data to the local database of each agent according to the rules of "master agent receiving standardized feature table + core task related data" and "slave agent receiving corresponding associated task list + supporting data" (for example, the master agent receives "commercial plot core building layout parameters", and the slave agent receives "adjacent plot traffic dispersion task details"); channel building: calling the preset channel validity verification rule, establishing a real-time data sharing channel based on the network parameters of the master-slave agent, configuring parameters such as bandwidth and encryption method (for example, configuring an encrypted channel for a high dependence slave agent to ensure real-time synchronization of geological survey data); connectivity verification: verifying the validity of the channel through bidirectional data test transmission of the master-slave agent - the master agent sends a "core task start signal" test package, and the slave agent returns a "associated task receiving confirmation" response package, the system checks the integrity (no packet loss), delay (≤ preset threshold, such as 50ms) and encryption validity of data transmission, and the channel is enabled after verification; group formation: after completing data distribution and channel verification, the master-slave collaborative agent group is formally formed and enters the task execution standby state. The reason for this operation is that data distribution errors will cause task understanding deviation, and channel problems will block collaboration (such as core task progress cannot be synchronized); the effect is to ensure that the agent has accurate data and real-time collaboration through "targeted distribution + regular channel + test transmission verification", providing bottom support for subsequent task execution and avoiding collaboration failure.

[0102] It should be noted that the quantitative algorithm of core task processing efficiency, the calculation logic of the dependence degree of associated tasks and core tasks, the detection technology of master-slave agent data synchronization delay, and the basic communication protocol of real-time data sharing channel (such as MQTT, WebSocket), the verification method of channel connectivity test transmission, etc. are all conventional technical means in the field, and the present application is not improved. The specific numerical values (such as dependence threshold, delay threshold, bandwidth threshold, etc.) involved in the explanation are only examples for understanding and are not limited to the present application. The skilled person can adjust according to actual needs. The above-mentioned technologies can be realized based on the existing technology, and will not be described here. Steps S305-S307 solve the problems of inaccurate matching of master-slave agents, no priority in associated task allocation, and lack of reliability guarantee for collaborative channel through the progressive process of "master agent double selection → slave agent hierarchical allocation → data targeted distribution and channel verification". Both the core leading ability of the master agent is ensured through "efficiency + success rate" screening, and the slave agent resource allocation is optimized through "dependence degree grading", and the stability of collaborative data transmission is ensured through "channel rule verification + bidirectional test transmission", and finally a "core clear, reasonable distribution, and reliable communication" master-slave collaborative agent group is built, providing complete agent organization and communication support for efficient collaborative execution of subsequent tasks.

[0103] In one of the embodiments, step S40 specifically includes the following steps:

[0104] S401, collect state data in multiple dimensions of data layer, execution layer and coordination layer in real time; wherein, the data layer collects packet loss rate and fragment integrity of data transmission between master agent and slave agent based on data segment identification of standardized feature table; the execution layer collects master agent core task progress and slave agent associated task completion degree based on task weight proportion of standardized feature table; the coordination layer collects connection stability and transmission delay of real-time data sharing channel;

[0105] S402, if any dimension state in the data layer, execution layer and coordination layer triggers a preset abnormal threshold, perform recovery operation according to a preset hierarchical recovery strategy, specifically including:

[0106] When the data layer is abnormal, call the directional supplement transmission rule based on the data segment identification, and only synchronize the target data segment lost in the standardized feature table;

[0107] When the execution layer is abnormal, schedule the standby agent with historical coordination success rate greater than or equal to a preset success rate threshold and adapt to the core task attribute, and synchronize the core task data processed in the standardized feature table;

[0108] When the coordination layer is abnormal, call the pre-device channel or start channel optimization based on the type of real-time data sharing channel, and synchronize the processed data to the new channel; wherein, if it is a wireless channel, switch to a preset wired standby channel, and if it is a wired channel, start channel bandwidth dynamic expansion;

[0109] S403, linkage upper planning database constructs compliance verification mechanism with phased and hierarchical, specifically including:

[0110] The phased verification triggers verification at three nodes of outputting task draft, task intermediate draft and task final draft of the master-slave coordination agent group respectively, and the verification trigger time of each node is bound with the task phase weight in the standardized feature table;

[0111] The hierarchical verification classifies the upper planning constraints in the upper planning database into strong constraints and weak constraints, wherein the strong constraint verification adopts parameter-by-parameter full amount verification, and the weak constraint verification adopts sampling verification based on task attribute weight;

[0112] S404, if the compliance verification does not pass, generate compliance correction guide, specifically including: associating the corresponding task attribute data in the standardized feature table with the upper planning constraint clause, explicitly prompting the parameters to be modified, the modification basis and the adjustable range, guiding the master-slave coordination agent group to accurately correct until the verification passes.

[0113] In this embodiment, step S401 is the basis for real-time monitoring of the task execution state of the master-slave collaborative agent group in S40, aiming to comprehensively master the core state data in the agent collaboration process through three-dimensional collection of "data + execution + collaboration", and provide data support for subsequent exception recovery and compliance verification. The process is implemented through "monitoring dimension determination → hierarchical collection rule calling → real-time collection of data at each level → state data integration". Among them, multi-dimensional real-time collection refers to synchronous coverage of "data transmission, task execution, channel collaboration" three core links, avoiding state misjudgment caused by single-dimensional collection; the collection logic of the three levels is bound with the standardized feature table or the constructed collaboration system, ensuring the pertinence and relevance of the collected data, as follows:

[0114] The data layer collection focuses on the data transmission quality between the master and slave agents based on the data segment identifier of the standardized feature table (i.e. the unique identifier of the feature table core data after splitting according to space, attribute, and task data, such as "space-longitude and latitude segment ID: 001"). Two core indicators are collected: packet loss rate (the proportion of lost segment number to total sent segment number, such as sending 100 segments and losing 2, the packet loss rate is 2%) and segment integrity (the proportion of complete received segments to required received segments, such as requiring 5 core segments and only receiving 4, the integrity is 80%). The execution layer collection focuses on task execution progress based on the task weight proportion of the standardized feature table (such as core task 0.6, high dependency associated task 0.3). Two core indicators are collected: master agent core task progress (calculated according to core task key node and weight proportion, such as core task total weight 0.6, complete proportion 40% of "scheme design", progress 24%) and slave agent associated task completion degree (calculated according to associated task weight and actual completion, such as high dependency task weight 0.3 and completed, completion degree 30%). The collaboration layer collection focuses on channel stability based on the real-time data sharing channel built in S307. Two core indicators are collected: connection stability (channel interruption times per unit time, such as 1 interruption in 1 hour, meeting the standard of "≤1 times / hour") and transmission delay (data transmission response time between master and slave agents, such as current 40ms, compared with S306 preset threshold of 50ms, meeting the standard).

[0115] The specific collection process is: the system first starts the multi-dimensional collection module, determines the collection objects and indexes of the data layer, the execution layer, and the coordination layer; then calls the collection rules of each level (such as collecting according to the "segment identification matching" rule in the data layer, and calculating according to the "task weight proportion" rule in the execution layer); synchronously collects each index data through real-time monitoring interface (such as collecting transmission delay every 5 seconds and task progress every 10 minutes); finally, the collection data of the three levels are integrated according to "time stamp + level identifier" to form a unified state data set. The reason for this operation is that in master-slave coordination, data transmission abnormalities, task progress disconnection, and unstable channels can all lead to coordination failure, and single-dimensional collection cannot fully capture the problem; the effect is to master the core state in real time through "three-dimensional full-coverage collection", and provide accurate data for S402 abnormal handling and S403 compliance verification, avoiding risk expansion.

[0116] Step S402 is a hierarchical recovery link for multi-dimensional state abnormalities, aiming to perform recovery operations according to different abnormal types of the data layer, the execution layer, and the coordination layer according to preset strategies, to avoid abnormal expansion. Specifically, it is implemented through the process of "abnormal dimension identification → corresponding strategy calling → targeted recovery → state correction". Among them, the preset abnormal threshold is the critical standard of each level state, and the hierarchical recovery strategy focuses on the core pain points of different abnormalities, as follows:

[0117] When the data layer is abnormal (such as packet loss rate ≥5% or segment integrity ≤80%), call the directional retransmission rule based on data segment identification, only synchronize the missing target data segments (non-full retransmission) in the standardized feature table, quickly recover the data integrity. When the execution layer is abnormal (such as core task progress lags behind the plan by 20%), schedule the standby agent with a historical coordination success rate ≥ a preset threshold (such as 85%) and adapt to the core task attributes, synchronize the processed core task data in the standardized feature table, and ensure seamless connection of tasks. When the coordination layer is abnormal (such as transmission delay ≥80ms or interruption more than 1 time within 10 minutes), handle based on channel type: switch wireless channel to preset wired backup channel, start bandwidth dynamic expansion in wired channel, reduce delay and improve stability.

[0118] The specific recovery process is: the system monitors the state data collected in real time S401, once the data of a certain level triggers the abnormal threshold (such as the data layer packet loss rate 6%), immediately identify the abnormal dimension (data layer); call the recovery strategy (directed supplement transmission rule) of the corresponding level; execute the recovery operation (locate the lost segment and supplement transmission); continuously collect the state data after recovery until the index returns to the normal range (such as the packet loss rate is reduced to 1%, the segment integrity is 100%), then determine that the recovery is completed. The reason for this operation is that the essence and influence of different levels of abnormality are different, and the unified recovery strategy is inefficient (such as full retransmission wastes resources); the effect is to quickly solve the abnormality through "layered accurate recovery", reduce the interference to the task, and ensure the continuity of master-slave cooperation.

[0119] Step S403 is the link of building a "staged + hierarchical" compliance verification mechanism for the upper planning database, aiming to realize differentiated compliance control of the whole process of the task, specifically through the process of "upper database linkage → staged verification trigger → hierarchical verification execution → result integration". Among them, the staged verification binds the task output node, and the hierarchical verification distinguishes the importance of constraints, as follows:

[0120] The staged verification triggers at the three nodes of the master-slave cooperative intelligent agent group output task draft, intermediate draft and final draft respectively, and the trigger time of each node is bound with the task phase weight in the standardized feature table (such as the final draft has the highest weight, and the verification is the most strict), to ensure that the achievements of each stage meet the upper planning guidance. The hierarchical verification classifies the constraints in the upper planning database into strong constraints and weak constraints: strong constraints (such as ecological red line, upper limit of volume rate) use parameter-by-parameter full verification to ensure that core compliance items are not missed; weak constraints (such as building style suggestions) use sampling verification based on task attribute weight (the higher the weight, the higher the sampling ratio), to balance efficiency and compliance.

[0121] The specific verification process is: the system first links to the upper planning database, extracts the strong and weak constraint clauses related to the current task (such as "commercial plot planning" associated with "plot ratio ≤3.0" (strong constraint), "parking ratio ≥1:1.2" (weak constraint)); When the agent outputs the first draft, the second draft and the final draft, trigger the verification of the corresponding stage according to the task stage weight (such as the final draft triggering full amount verification); For strong constraint clauses, compare the agent results with upper planning parameter by parameter (such as checking whether the height of each building is ≤100 meters); For weak constraint clauses, sample verification according to task attribute weight (such as "green building materials" weight 0.2, extract 20% of the building material list to check whether it meets the recommendation); Finally, integrate the verification results of each stage and each level to form a phased compliance report. The reason for this operation is that single time point verification may lead to later rework, and full amount verification of weak constraints will waste resources; The effect is to avoid large-scale rectification in the later stage through "staging", and to achieve "zero tolerance" for strong constraints and "high efficiency" for weak constraints through "hierarchical", providing basis for S404 correction.

[0122] Step S404 is the correction guidance link after the compliance verification fails, aiming to provide accurate correction direction for the master-slave collaborative agent group and avoid blind adjustment, which is realized through the process of "failed item positioning → data and clause association → correction guidance generation → re-passing". Among them, the compliance correction guidance focuses on "specific parameters, modification basis, adjustment range", ensuring that the correction is operable, as follows:

[0123] If the compliance verification fails, the system first locates the failed item (such as plot ratio exceeding the standard in strong constraint, green building material usage rate insufficient in weak constraint); Then associate the corresponding task attribute data in the standardized feature table (such as plot ratio calculation basis data) with the upper planning constraint clause (such as "plot ratio ≤2.5"); Finally, generate a correction guide containing three items: parameters that need to be modified (such as total building area), modification basis (quote planning clause original text), and adjustable range (combined with the legal interval of the standardized feature table).

[0124] The specific generation and execution process is: not through item positioning is that the system first extracts the specific items that do not pass from the check results of S403, and clearly whether it is a strong constraint or a weak constraint that does not pass, which link of the corresponding task (such as the "plot ratio parameter" in the core task "building layout"); data and clause association is to call the task attribute data (such as "plot area, total building area of plot ratio calculation basis data") corresponding to the item that does not pass in the standardized feature table, and at the same time associate the corresponding constraint clause (such as "Article X of XX Regional Regulatory Detailed Planning: The plot plot ratio shall not exceed 2.5") in the upper-level planning database, forming an association relationship of "task data + constraint clause"; correction content generation is to clearly three core guiding information based on the association result: the parameter that needs to be modified (directly positioning specific parameters, such as "total building area, plot ratio calculation result"), modification basis (quoting the original text of the upper-level planning constraint clause, such as "According to Article X of XX Planning, the upper limit of plot ratio is 2.5"), and adjustable range (combined with the task attribute weight in the standardized feature table, the legal adjustment interval of the parameter is given, such as "current total building area is 12000 square meters, plot area is 5000 square meters, plot ratio is 2.4, total building area needs to be reduced to ≤12500 square meters (5000×2.5), and the adjustable range is 10000-12500 square meters"), for example, when the check finds that "plot ratio 3.0 exceeds the standard", the guide is generated: "the parameter to be modified: total building area; modification basis: Article X of XX Planning (plot ratio ≤2.5); adjustable range: ≤12500 square meters (current 15000 square meters, need to reduce 2500-5000 square meters)"; after correction check is to push the compliance correction guide to the master-slave collaborative intelligent agent group, the master agent adjusts the core task parameters (such as reducing the total building area) combined with the guide, the slave agent adjusts the associated tasks (such as modifying the supporting facility area) at the same time, and after the correction is completed, the system triggers the compliance check of S403 again until the check passes. The reason for this operation is that only "not passed" is informed without guidance, which is easy to lead to repeated correction failure; the effect is to clearly "what to change, why to change, and how much to change", which greatly improves the correction efficiency and ensures that the final result fully meets the upper-level planning constraints.

[0125] It should be noted that the calculation method of data layer packet loss rate and segment integrity in step S401, the quantization logic of execution layer task progress, the detection technology of coordination layer transmission delay, the basic protocol of targeted retransmission in S402, the general algorithm of standby agent scheduling, the general technology of channel bandwidth expansion, the sampling algorithm of hierarchical verification in S403, the trigger timing control of phased verification, and the format generation logic of compliance correction guidance in S404, etc. are all adopted by the conventional technical means in the art, and the present application does not improve them. The specific threshold values (such as packet loss rate 5%, delay 80ms) involved in the explanation are only examples for assisting understanding, and are not a limitation of the present application. The skilled person can adjust them according to the actual scene. The above conventional technical means can be realized based on the prior art, and will not be described here. Steps S401-S404 form a closed-loop process of "multi-dimensional state monitoring → hierarchical abnormal recovery → full-process compliance verification → accurate correction guidance", which constructs an "execution- guarantee-verification" system of master-slave collaborative agent group. The three-dimensional collection of S401 provides a data basis for abnormal identification and compliance verification, the hierarchical recovery of S402 solves the efficient processing problem of different types of abnormalities, the "phased and hierarchical" verification of S403 realizes full-process compliance control, and the correction guidance of S404 ensures the accurate solution of compliance problems. The four work together to ensure the stability of the master-slave agent group (through real-time monitoring and abnormal recovery), and to ensure the compliance of the task results (through verification and correction), and finally to realize the efficient and compliant collaborative completion of the planning space task and the associated task.

[0126] In one of the embodiments, the method further comprises the following steps:

[0127] S501, during the execution of the task by the master-slave collaborative agent group, the preset GIS database and the upper planning database are linked every preset period, and the difference is detected based on the data segment identifier in the standardized feature table, and only the spatial data associated with the identified data segment in the planning space task and the upper planning constraint are detected whether they are updated;

[0128] S502, if the update is detected, the updated associated task list is regenerated based on the preset spatial topology association rule and the preset task function dependency rule, and the task attribute weight in the standardized feature table is used to determine that the task attribute weight greater than or equal to the preset weight threshold is the core associated task, and the task attribute weight less than the preset weight threshold is the general associated task, and the processed data belonging to the invalid task is marked at the same time.

[0129] S503, synchronize the updated associated task list to the master-slave collaborative agent group, and assign the newly added tasks by the master agent according to the priority, preferentially assign the core associated tasks to the idle slave agent with the highest historical collaboration success rate, assign the general associated tasks to the pre-equipment slave agent, and terminate the execution of the invalid tasks; at the same time, the master agent recycles the processed data of the invalid tasks, synchronously updates to the standardized feature table based on the data segment identifier, and ensures no redundancy conflict with the current task data.

[0130] In the embodiment, step S501 is a dynamic update detection link during the task execution of the master-slave collaborative agent group, which aims to timely capture the changes of the associated data and provide basis for task adjustment, and is specifically realized through the process of “period setting → database linkage → identifier matching → difference detection”. Among them, the preset period refers to the fixed detection interval (such as once every 2 hours), which ensures the timeliness of the update detection; the linked preset GIS database (storing spatial geographic data) and the upper planning database (storing planning constraints) are the data sources of the update detection, which are as follows:

[0131] Based on the data segment identifier (such as “space-plot boundary ID: 003” “constraint-volume ratio clause ID: 005”) in the standardized feature table, only the contents associated with these identifiers in the planning space task are detected, i.e. focusing on “associated spatial data” (such as plot boundary, terrain data) and “upper planning constraints” (such as volume ratio adjustment, ecological red line range change), rather than full detection, to avoid resource waste.

[0132] The specific detection process is: the system first automatically starts the update detection module according to the preset period (such as every 2 hours); links the preset GIS database and the upper planning database through the database interface, establishes a data access channel; calls the data segment identifier matching rule, locates the spatial data and planning constraints associated with the standardized feature table data segment identifier (such as “space-plot boundary ID: 003” “constraint-volume ratio clause ID: 005”) from the two databases; compare the current located data with the historical data record stored by the system field by field (such as plot boundary coordinates, volume ratio value, ecological red line range), and judge whether there is an update (such as the plot boundary coordinates from “X1, Y1-X2, Y2” to “X3, Y3-X4, Y4”, or the upper limit of the volume ratio in the constraint clause from “3.0” to “2.5”, which is determined as data update). The reason for this operation is that the spatial data (such as terrain) and the upper constraint (such as policy) may be dynamically updated in the planning space task execution, and full detection will occupy too many resources; the effect is to accurately capture the relevant updates through “identifier directional detection”, and provide timely and necessary basis for subsequent task list adjustment.

[0133] Step S502 is the reconstruction and classification of the updated associated task list, which aims to generate an adapted task list based on new data and determine the priority of tasks. This is achieved through the process of "update response → rule call → list generation → task classification → invalidation marking". The pre-set spatial topology association rules (such as "land boundary change associated with surrounding traffic planning tasks") and task function dependency rules (such as "volume rate increase associated with supporting facility expansion tasks") are the core basis for the reconstruction of the list, which is as follows:

[0134] If S501 detects an update, the system calls the above two rules to reorganize the association between tasks, generates an updated associated task list, and classifies it according to the task attribute weight in the standardized feature table (such as "traffic planning task weight 0.4" "green adjustment task weight 0.1") with a pre-set weight threshold (such as 0.3): tasks with attribute weight greater than or equal to the threshold are core associated tasks (such as traffic planning tasks with weight 0.4), and tasks with attribute weight less than the threshold are general associated tasks (such as green adjustment tasks with weight 0.1); at the same time, mark the tasks that are invalidated due to data update (such as the old planning tasks corresponding to the original land boundary) and the ownership of the processed data (such as "data to be recycled").

[0135] The specific execution process is as follows: after S501 detects data updates, the system immediately triggers the associated task list reconstruction mechanism and generates a reconstruction instruction; calls the pre-set spatial topology association rules and task function dependency rules (such as "land boundary change → associated surrounding traffic planning tasks" "volume rate increase → associated supporting facility expansion tasks"), reorganizes the association between core tasks and associated tasks based on updated data, and generates an updated associated task list; extract the task attribute weight in the standardized feature table (such as "traffic planning task 0.4, green adjustment task 0.1, pipeline reconstruction task 0.25"), set a pre-set weight threshold (such as 0.3), mark tasks with weight greater than or equal to 0.3 (such as traffic planning tasks) as core associated tasks, and mark tasks with weight less than 0.3 (such as green adjustment and pipeline reconstruction tasks) as general associated tasks; identify invalid tasks that are no longer adapted due to data updates (such as old traffic planning tasks corresponding to the original land boundary, supporting facility design tasks under the original volume rate standard), and mark the ownership of the processed data (such as "data to be recycled - original land planning" "data to be recycled - old volume rate supporting design"). The reason for this operation is that data updates may cause the original task association to be invalid or new necessary tasks to be added, which requires reorganization and priority differentiation; the effect is to generate a task list that adapts to new data, clearly defines core tasks and invalid data, and lays the foundation for subsequent task allocation.

[0136] Step S503 is the allocation and data synchronization link of the updated task, aiming to ensure the efficient execution of new tasks, the timely termination of invalid tasks, and the avoidance of data conflicts, which is achieved through the process of "inventory synchronization → task allocation → invalid termination → data recovery → feature table update". Among them, the master agent is the core of allocation, and the allocation strategy is formulated according to the task priority and the state of the slave agent, as follows:

[0137] The updated associated task list generated by S502 is synchronized to the master-slave collaborative agent group, and the master agent allocates according to priority: core associated tasks are preferentially allocated to "idle slave agents with the highest historical collaboration success rate" (such as slave agent A with a success rate of 95% and currently idle) to ensure the efficiency of core task collaboration, and general associated tasks are allocated to pre-equipment slave agents to avoid occupying master collaboration resources, while terminating the execution of invalid tasks to prevent invalid resource consumption; in terms of data processing, the master agent recovers the processed data of invalid tasks, updates it to the standardized feature table based on the data segment identifier (such as deleting old plot boundary data and replacing it with new data), ensuring that the data in the feature table matches the current task without redundancy or conflict.

[0138] The specific execution process is as follows: the master agent receives the updated associated task list generated by S502, parses the core associated task, general associated task, and invalid task information in the list; selects the slave agent with the highest historical collaboration success rate and currently in idle state from the target agent candidate set (such as slave agent A with a historical success rate of 95%), and allocates the core associated task (such as the updated traffic planning task) to this slave agent; at the same time, the general associated task (such as the green adjustment task) is allocated to the pre-equipment slave agent (such as the standby slave agent B in the agent pool); the task termination instruction is sent to the slave agent that is executing the invalid task (such as the original old traffic planning task), and the current task execution is stopped; the master agent recovers the processed data (such as the preliminary route design of the old traffic planning and the engineering quantity calculation data) from the slave agent executing the invalid task through the data interaction interface, and classifies and organizes the data; based on the data segment identifier, the organized invalid data is updated to the standardized feature table (such as deleting the old traffic planning data corresponding to "space-old plot boundary ID: 003" and replacing it with the new traffic planning basic data corresponding to "space-new plot boundary ID: 006"), ensuring that the data in the standardized feature table completely matches the current task requirements without redundancy or conflict. The reason for this operation is that new tasks need to be quickly implemented, and invalid tasks need to be terminated in a timely manner to avoid resource waste, and data inconsistency may lead to collaboration confusion; the effect is to ensure the efficient execution of core tasks through priority allocation, and to ensure the accuracy of task data without conflict through data recovery and update, realizing the dynamic adaptation of master-slave collaboration.

[0139] It should be noted that the conventional linkage technology of the GIS database and the upper planning database in step S501, the difference detection basic algorithm based on the data segment identifier, the basic logical framework of the spatial topology correlation rule and the task function dependence rule in S502, the conventional setting method of the task attribute weight, and the judgment technology of the idle state of the intelligent agent in S503, the conventional checking means of the data redundancy conflict, etc. are all adopted by the conventional technical means in the field, and the present application is not improved. The specific values involved in the explanation, such as the preset period (such as 2 hours) and the weight threshold (such as 0.3), are only examples for assisting understanding of the technical solution, and are not a limitation of the present application. The skilled in the art can adjust according to the actual scene. The above conventional technical means can be realized based on the prior art, and will not be described here. The process of “dynamic update detection → task list reconstruction → execution adaptation and data synchronization” in steps S501-S503 builds a dynamic response mechanism of the master-slave collaborative intelligent agent group to external data changes. The directional difference detection in S501 ensures timely capture of key data updates (rather than full detection), balancing timeliness and resource consumption; S502 reconstructs the task list based on the updated data and classifies it by weight, clearly defining the core tasks and invalid data, providing clear guidance for subsequent execution; The priority allocation and data recovery mechanism in S503 not only ensures the efficient execution of new core tasks, but also avoids data redundancy and conflicts by terminating invalid tasks and updating the standardized feature table. The three work together to enable the master-slave collaborative intelligent agent group to dynamically adapt to changes in GIS data and upper planning, upgrading from “static execution” to “dynamic response”, ensuring that the planning space task can still be efficiently and accurately promoted when external conditions change.

[0140] 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 by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0141] In one embodiment, a spatial task intelligent agent matching and distribution system based on spatial features is provided, which corresponds to the above-mentioned spatial task intelligent agent matching and distribution method based on spatial features. The spatial task intelligent agent matching and distribution system based on spatial features comprises:

[0142] A task analysis and associated task generation module is configured to receive and analyze user-input planning space tasks, convert the planning space tasks into a standardized feature table containing entity information, task attributes and required spatial data, call a preset GIS database based on the spatial range in the standardized feature table, generate a spatial scene label for the planning space task, and extract associated tasks based on the preset association mining rules and the task attributes in the standardized feature table to obtain an associated task list.

[0143] The scene-adapted intelligent agent preliminary screening module is configured to build an intelligent agent pool and an adaptation rule library, wherein the intelligent agent pool stores a plurality of special intelligent agents, each of which is configured with a preset intelligent agent attribute, and the adaptation rule library is preset with a plurality of matching rules of space scenes and intelligent agents; based on a space scene label of a planned space task, a corresponding screening rule in the adaptation rule library is called to preliminarily screen the special intelligent agents in the intelligent agent pool, and a set of scene-adapted intelligent agents is obtained;

[0144] The special intelligent agent re-screening and collaborative group building module is configured to re-screen the set of scene-adapted intelligent agents according to a standardized feature table based on a preset multi-dimensional screening rule, to obtain a target intelligent agent candidate set; based on an associated task list, a master-slave collaborative intelligent agent group is built from the target intelligent agent candidate set according to a preset collaborative group configuration rule, the standardized feature table and the associated task list are distributed to corresponding master and slave intelligent agents, and a real-time data sharing channel between the master and slave intelligent agents is established;

[0145] The abnormality monitoring and compliance verification module is configured to collect the execution and data sharing state of the master-slave collaborative intelligent agent group in real time; if the state triggers a preset abnormal condition, the standby intelligent agent is dispatched or the standby data channel is switched to restore the collaboration based on the synchronized processed data of the standardized feature table; at the same time, the upper planning database is linked to verify the compliance of the task processing result of the intelligent agent group according to the standardized feature table, to ensure compliance with the upper planning constraints.

[0146] The specific limitations of the space task intelligent agent matching and distribution system based on space features can be referred to the limitations of the space task intelligent agent matching and distribution method based on space features in the foregoing, which will not be repeated here. Each module in the space task intelligent agent matching and distribution system based on space features 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.

[0147] In one embodiment, a computer device, which can be a server, is provided, and an internal structure diagram of the computer device 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 operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is used to store data, data processing, data analysis, etc. The network interface of the computer device is used to communicate with the external terminal through the network connection. The computer program is executed by the processor to implement a space task agent matching and distribution method based on spatial features.

[0148] In one of the embodiments, a computer device is provided, including a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement a space task agent matching and distribution method based on spatial features.

[0149] In one of the embodiments, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement a space task agent matching and distribution method based on spatial features.

[0150] 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), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM) and the like.

[0151] 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.

[0152] 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 method for spatial task agent matching and distribution based on spatial features, characterized in that, Comprising the following steps: S10, receiving and analyzing the user input planning space task, converting the planning space task into a standardized feature table containing entity information, task attributes and required space data, calling a preset GIS database based on the space range in the standardized feature table, generating a space scene tag of the planning space task, and extracting associated tasks based on the task attributes in the standardized feature table and the preset association mining rules to obtain an associated task list; S20, constructing an agent pool and an adaptive rule library, wherein the agent pool stores a plurality of special agents, each of which is configured with a preset agent attribute, and the adaptive rule library is preset with a plurality of matching rules of space scenes and agents; based on the space scene tag of the planning space task, the corresponding filtering rules in the adaptive rule library are called to preliminarily screen the special agents in the agent pool to obtain a scene-adapted agent set; S30, according to the standardized feature table, the scene-adapted agent set is re-screened according to a preset multi-dimensional screening rule to obtain a target agent candidate set; based on the associated task list, a master-slave collaborative agent group is formed from the target agent candidate set according to a preset collaborative group configuration rule, and the standardized feature table and the associated task list are distributed to the corresponding master and slave agents, and a real-time data sharing channel between the master and slave agents is established; S40, real-time collection of the execution and data sharing state of the master-slave collaborative agent group; if the state triggers a preset abnormal condition, the processed data is synchronized based on the standardized feature table, a standby agent is dispatched or a standby data channel is switched to restore collaboration; at the same time, the upper planning database is linked, the task processing result of the agent group is checked for compliance according to the standardized feature table, and it is ensured that it conforms to the upper planning constraints.

2. The method of claim 1, wherein, In the step of extracting associated tasks based on the preset association mining rules and the task attributes in the standardized feature table to obtain an associated task list, the following steps are specifically included: S101, based on the required space data in the standardized feature table, a preset spatial topology association rule is called to identify potential associated tasks with a topology relationship of inclusion, adjacency or overlap with the spatial range of the planning space task to generate a first potential associated task set; S102, based on the task attributes in the standardized feature table, a preset task function dependency rule is called to identify potential associated tasks with a function dependency relationship of pre-execution, post-support or synchronous collaboration with the planning space task to generate a second potential associated task set, and the second potential associated task set is combined with the first potential associated task set to obtain a comprehensive potential associated task set; S103, calling a preset upper planning verification rule, linking an upper planning database to verify the compliance of the comprehensive potential associated task set, eliminating tasks that do not conform to the upper planning constraints, and integrating the remaining tasks to obtain the associated task list.

3. The method of claim 1, wherein, In the step of calling the adaptive rule library corresponding to the filtering rule based on the space scene tag of the planning space task, the special agents in the agent pool are preliminarily screened to obtain a scene-adapted agent set, the following steps are specifically included: S201, based on the space scene label, call the space scene subdivision rule in the adaptive rule library, decompose the space scene label into a combination label containing a space function partition and a space constraint level, and obtain a subdivided space scene label; S202, call the mapping rule of the subdivided scene and the agent attribute level in the adaptive rule library, filter out the special agent whose special agent attribute matches the subdivided space scene label from the agent pool, and generate a preliminary scene adaptive agent set; S203, if the preliminary scene adaptive agent set is empty, trigger a pre-equipment mechanism, and determine the non-empty agent set obtained through the pre-equipment mechanism as the scene adaptive agent set, wherein the pre-equipment mechanism includes at least one of the following operations: sending an agent attribute upgrade prompt to the system and calling a cross-scene agent cooperation rule to generate a temporary cooperation adaptation scheme.

4. The method of claim 3, wherein, In the step of screening the scene adaptive agent set according to the standardized feature table according to the preset multi-dimensional screening rule to obtain a target agent candidate set, the following steps are specifically included: S301, extract the screening core parameters from the standardized feature table, wherein the screening core parameters at least include the space range complexity of the planning space task, the task attribute weight and the time constraint threshold; S302, construct a three-dimensional attribute evaluation matrix for each special agent in the scene adaptive agent set, wherein the three-dimensional attribute evaluation matrix includes a space adaptation dimension, a task capability dimension and a time efficiency adaptation dimension; the space adaptation dimension is based on the matching degree of the historical service space range and the current task space range complexity in the agent attribute; the task capability dimension is based on the weighted matching of the special task success rate in the agent attribute and the task attribute weight in the standardized feature table; and the time efficiency adaptation dimension is based on the difference between the average task completion cycle in the agent attribute and the time constraint threshold; S303, apply a preset dynamic weight algorithm to quantitatively score the three-dimensional attribute evaluation matrix to generate a comprehensive adaptation score of each special agent; S304, screen the special agent with a comprehensive adaptation score greater than or equal to a preset threshold to form a target agent candidate set; wherein if there are special agents with the same comprehensive adaptation score, the special agent with the lowest historical cooperation conflict rate is preferentially retained, and the historical cooperation conflict rate is the proportion of the task conflict times to the total cooperation times when the special agent cooperates with other special agents.

5. The method of claim 4, wherein, In the steps of grouping the master-slave cooperative agent group from the target agent candidate set according to a preset cooperation group configuration rule based on the associated task list, distributing the standardized feature table and the associated task list to the corresponding master and slave agents, and establishing a real-time data sharing channel between the master and slave agents, the following steps are specifically included: S305, based on the core task weight of the planning space task in the associated task list and the standardized feature table, call a preset master agent selection rule, filter out a special agent with the highest core task processing efficiency and a historical cooperation success rate greater than or equal to a preset threshold from the target agent candidate set, and determine the special agent as a master agent; S306, based on the dependency of each associated task in the associated task list on the core task, a preset agent allocation rule is called to allocate the associated task with a dependency greater than or equal to a preset dependency threshold to a special agent with a master agent data synchronization delay less than or equal to a preset delay threshold, and to allocate the associated task with a dependency less than the preset dependency threshold to a preset slave agent, and determine the slave agent; S307, the standardized feature table and the associated task list are respectively distributed to the corresponding master agent and slave agent, a preset channel validity verification rule is called to build a real-time data sharing channel, and the channel connectivity is verified through bidirectional data transmission verification of the master-slave agent, and the construction of the master-slave collaborative agent group is completed after the verification is passed.

6. The method of claim 1, wherein, The execution and data sharing state of the real-time acquisition master-slave collaborative agent group is collected; if the state triggers a preset abnormal condition, the standby agent is dispatched or the standby data channel is switched to restore the collaboration based on the standardized feature table to synchronize the processed data; at the same time, the upper planning database is linked to verify the compliance of the task processing result of the agent group according to the standardized feature table, and ensure that the steps meet the upper planning constraints, which specifically include the following steps: S401, collect state data in multiple dimensions of data layer, execution layer and collaboration layer in real time; wherein, the data layer collects the packet loss rate and segment integrity of data transmission between the master agent and the slave agent based on the data segment identifier of the standardized feature table; the execution layer collects the progress of the master agent core task and the completion degree of the slave agent associated task based on the task weight proportion of the standardized feature table; the collaboration layer collects the connection stability and transmission delay of the real-time data sharing channel; S402, if any dimension state in the data layer, execution layer and collaboration layer triggers a preset abnormal threshold, execute the recovery operation according to the preset hierarchical recovery strategy, which specifically includes: When the data layer is abnormal, call the directional supplement rule based on the data segment identifier to synchronize only the target data segment missing in the standardized feature table; When the execution layer is abnormal, dispatch the standby agent with a historical collaboration success rate greater than or equal to a preset success rate threshold and adapt to the core task attribute to synchronize the processed core task data in the standardized feature table; When the collaboration layer is abnormal, call the preset channel or start the channel optimization based on the type of real-time data sharing channel, and synchronize the processed data to the new channel; if it is a wireless channel, switch to a preset wired standby channel, and if it is a wired channel, start the channel bandwidth dynamic expansion; S403, link the upper planning database to build a compliance verification mechanism with phased and hierarchical levels, which specifically includes: The phased verification triggers verification at three nodes of the master-slave collaborative agent group outputting the task draft, the task intermediate draft and the task final draft, respectively, and the verification trigger time of each node is bound to the task phase weight in the standardized feature table; The hierarchical verification classifies the upper planning constraints in the upper planning database into strong constraints and weak constraints, wherein the strong constraint verification adopts parameter-by-parameter full verification, and the weak constraint verification adopts sampling verification based on the task attribute weight; S404, if the compliance verification fails, a compliance correction guide is generated, specifically including: associating the corresponding task attribute data in the standardized feature table with the upper planning constraint clause, explicitly prompting the parameters to be modified, the basis for modification and the adjustable range, guiding the master-slave collaborative intelligent agent group to accurately correct until the verification passes.

7. The method of claim 2, wherein, Further comprising the following steps: S501, during the execution of the task by the master-slave collaborative intelligent agent group, the preset GIS database and the upper planning database are linked every interval, and based on the data segment identifier in the standardized feature table, difference detection is performed to detect whether the spatial data associated with the identified data segment in the planning space task and the upper planning constraint have been updated; S502, if the update is detected, the updated associated task list is regenerated based on the preset spatial topology association rule and the preset task function dependency rule, and according to the task attribute weight in the standardized feature table, the task attribute weight greater than or equal to the preset weight threshold is the core associated task, and the task attribute weight less than the preset weight threshold is the general associated task, and the corresponding processed data of the invalid task is marked; S503, the updated associated task list is synchronized to the master-slave collaborative intelligent agent group, and the master agent assigns the new task priority, preferentially assigns the core associated task to the idle slave agent with the highest historical collaboration success rate, assigns the general associated task to the pre-equipped slave agent, and terminates the execution of the invalid task; at the same time, the master agent recycles the processed data of the invalid task, synchronously updates to the standardized feature table based on the data segment identifier, and ensures no redundancy conflict with the current task data.

8. A spatial task agent matching and distribution system based on spatial features, for implementing the steps of a spatial task agent matching and distribution method based on spatial features according to any one of claims 1-7, characterized in that, It comprises: a task analysis and associated task generation module for receiving and analyzing user input planning space tasks, converting the planning space tasks into a standardized feature table containing entity information, task attributes and required spatial data, calling a preset GIS database based on the spatial range in the standardized feature table, generating a spatial scene label for the planning space task, and extracting associated tasks based on the preset association mining rules and task attributes in the standardized feature table to obtain an associated task list; a scene adaptation intelligent agent preliminary screening module for constructing an intelligent agent pool and an adaptation rule library, wherein the intelligent agent pool stores a plurality of special intelligent agents, each of which is configured with a preset intelligent agent attribute, and the adaptation rule library is preset with a plurality of matching rules for space scenes and intelligent agents; based on the spatial scene label of the planning space task, the corresponding screening rule in the adaptation rule library is called to preliminarily screen the special intelligent agents in the intelligent agent pool to obtain a scene adaptation intelligent agent set; a special intelligent agent rescreening and collaboration establishment module for rescreening the scene adaptation intelligent agent set according to the standardized feature table based on a preset multi-dimensional screening rule to obtain a target intelligent agent candidate set; based on the associated task list, a master-slave collaborative intelligent agent group is established from the target intelligent agent candidate set according to a preset collaboration group configuration rule, the standardized feature table and the associated task list are distributed to the corresponding master and slave agents, and a real-time data sharing channel between the master and slave agents is established; An abnormality monitoring and compliance checking module is configured to collect the execution and data sharing state of the master-slave collaborative intelligent agent group in real time; if the state triggers a preset abnormal condition, the processed data is synchronized based on a standardized feature table, and a standby intelligent agent or a standby data channel is dispatched to restore the collaboration; at the same time, the upper planning database is linked, the task processing result of the intelligent agent group is checked for compliance according to the standardized feature table, and it is ensured that the upper planning constraints are met.

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 processor executes the computer program to realize the steps of the space task intelligent agent matching and distribution method based on spatial features according to any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to realize the steps of the space task intelligent agent matching and distribution method based on spatial features according to any one of claims 1-7.

Images