Land resource asset checking method and system based on multi-source data analysis

Abstract

The invention relates to the field of multi-source data analysis, in particular to a land resource asset checking method and system based on multi-source data analysis. The method comprises the steps that boundary data and checking rules are acquired, asset unit segmentation and data source capability matching are carried out, and acquisition configuration is generated; ground survey data and historical data are collected according to configuration, coordinate and timestamp unified processing is carried out, and a multi-source fragment evidence packet is generated; constructing geometric, time sequence and semantic consistency vectors based on the evidence packet to form a consistency conflict graph structure; and finally, through fusion of constraint rule loading and evidence priority ranking, a closed-loop collection session configuration is generated to realize continuous checking. According to the invention, through consistency fusion and conflict graph analysis of multi-source data, the automatic processing capability, traceability and decision reliability of the checking result are significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of multi-source data analysis, in particular to a land resource asset inventory method and system based on multi-source data analysis. BACKGROUND

[0002] In the field of multi-source data analysis, the existing scheme for land resource asset inventory around the inventory target domain boundary data, land asset object list, multi-source data source list and asset unit division rule usually implements asset unit segmentation, multi-source data source capability field registration and anchor point candidate matching processing separately from remote sensing image data and laser radar point cloud data collection, ground survey data collection and historical inventory data loading, and adopts a loose association mode to link between each segment output, which has limitations such as difficulty in keeping consistent reference of collection session configuration structure when multi-source data source list changes, easy occurrence of field caliber difference and discontinuous source tracing in multi-source fragmented evidence package collection structure when cross-source aggregation, and lack of unified organization carrier in cross-link transmission of consistency conflict graph structure. The existing method relies on scattered collection session configuration and data loading rules, and geometric consistency sub-vector construction, time sequence consistency sub-vector construction and semantic consistency sub-vector construction as independent processing links, which is prone to inconsistent sub-vector input boundary and incomplete conflict relationship expression in the scene of multi-source fragmented evidence package collection structure with diverse sources and frequent updates, making it difficult to meet the continuous generation and repeatable call of consistency conflict graph structure. For joint processing of multi-source fragmented evidence package collection structure, the existing technology generally lacks a through link constraint between fusion constraint rule set loading, evidence priority sequence generation and fusion session primary key registration, making it difficult to form the same process from collection session configuration structure to multi-source fragmented evidence package collection structure, to consistency conflict graph structure and back to collection session configuration structure in the process of land resource asset inventory, resulting in repeated loading, repeated comparison and missing version association of multi-source data analysis process corresponding to the same inventory target domain boundary data, which has adverse effects on subsequent inventory data processing, inventory process trace and process traceability. SUMMARY

[0003] To solve the above technical problems, the present application provides a land resource asset inventory method based on multi-source data analysis, comprising:

[0004] Obtaining inventory target domain boundary data, land asset object list, multi-source data source list and asset unit division rule, performing asset unit segmentation, multi-source data source capability field registration and anchor point candidate matching processing, and generating collection session configuration structure;

[0005] Based on the collection session configuration structure, performing ground survey data collection and historical inventory data loading processing, and generating multi-source fragmented evidence package collection structure;

[0006] Based on the multi-source fragmented evidence package set structure, geometric consistency sub-vector construction, temporal consistency sub-vector construction and semantic consistency sub-vector construction are performed to generate a consistency conflict graph structure.

[0007] Based on the consistency conflict graph structure, the fusion constraint rule set is loaded, the evidence priority sequence is generated, and the fusion session primary key registration is processed to generate the collection session configuration structure.

[0008] Furthermore, the process of dividing asset units also includes:

[0009] The asset unit segmentation process includes sorting candidate boundary sources by the parcel boundary priority field, performing boundary trimming or pruning actions based on the validity status of the boundary version identifier using the parcel boundary or grid skeleton, and performing void area threshold comparison and connectivity merging based on the suspected void repair rule field; and performing multi-source data source capability field registration processing.

[0010] Furthermore, the process of registering multi-source data source capability fields and handling anchor point candidate matching also includes:

[0011] The multi-source data source capability field registration process includes reading the data source identifier, source type identifier, coverage identifier, refresh interval field, acquisition method identifier, and interface stability identifier from the multi-source data source list and generating capability field records by combining them with historical fetch logs; performing anchor point candidate matching processing, which includes constructing an anchor point candidate matching request for the primary key field of each asset unit, filtering data source identifiers in the capability field records whose coverage identifier field and spatial range identifier field have an overlap relationship, generating candidate sorting by spatial resolution field priority order field, point cloud density level field priority order field, and interface stability level field priority order field, and performing time availability verification; and generating the acquisition session configuration structure.

[0012] Furthermore, the process of collecting ground survey data and loading and processing historical inventory data also includes:

[0013] The ground survey data acquisition and processing includes reading the sampling evidence chain field from the acquisition session configuration structure to generate an acquisition instruction package, sending it to the mobile acquisition terminal to start the acquisition session, acquiring the positioning trajectory record index, the field image index and the set of mandatory attribute fields, and performing form field integrity verification and basic legality verification; performing historical inventory data loading processing, which includes retrieving historical inventory result files by sampling spatial range field, filtering historical result batches by sampling time window field and historical version identifier field falling into the window, and performing file index loading; aggregating the acquisition results into multi-source data packets; and performing coordinate projection standard unification processing based on multi-source data packets, which includes reading coordinate reference labels from remote sensing image data fragments, lidar point cloud data fragments, ground survey data fragments and historical inventory data fragments and performing projection transformation to generate unified coordinate reference labels.

[0014] Furthermore, the process of constructing geometrically consistent subvectors also includes:

[0015] The geometric consistency sub-vector construction process includes reading the coordinate system identifier and projection parameter identifier in the aperture field, uniformly projecting geometric fragments from different sources to the standard coordinate frame, generating a standardized geometric contour set, and calculating the spatial overlap field, morphological offset field, and boundary fit field; and performing temporal consistency sub-vector construction process.

[0016] Furthermore, the processes for constructing temporally consistent subvectors and semantically consistent subvectors also include:

[0017] The temporal consistency sub-vector construction process includes extracting the collection time, submission time, and historical version identifier fields from the evidence chain field and converting them into a standard timeline format; aggregating time series fragments to calculate the sampling density field, time drift field, and rate of change field; and performing semantic consistency sub-vector construction process, which includes reading the source identifier, sampled evidence type, investigator identifier, and historical attribute labels from the evidence chain field; uniformly mapping them to a vocabulary standard set; establishing an attribute correspondence matrix based on the terminology dictionary entry set; merging synonym items; performing conflict labeling on contradictory items to generate a semantic difference marker field; and calculating the consistency distribution field and semantic confidence field.

[0018] Furthermore, the process of loading the fusion constraint rule set also includes:

[0019] The fusion constraint rule set loading process includes reading the set of rule entries that match the rule version identifier field from the rule repository service, filtering constraint subsets by the applicable conflict type field, adjusting the loading order of constraint subsets according to the conflict intensity field, and registering hard constraint entry fields and soft constraint entry fields.

[0020] Furthermore, the process of generating the evidence priority sequence also includes:

[0021] The evidence priority sequence generation process includes initiating evidence retrieval requests to the evidence index service for the graph node set elements, extracting the evidence score field from the quality field and the evidence chain field, and sorting the evidence list according to the evidence score field to generate an evidence priority sequence.

[0022] Furthermore, the process of integrating session primary key registration also includes:

[0023] The fusion session primary key registration process includes reading the conflict graph batch identifier field, rule version identifier field, boundary version identifier field, and evidence priority sequence field index to generate session registration input records, performing uniqueness verification, and returning the fusion session primary key.

[0024] Furthermore, a land resource asset inventory system based on multi-source data analysis includes: an asset unit primary key table generation unit, an anchor point candidate list generation unit, a data acquisition session configuration structure generation unit, a multi-source data packet generation unit, a multi-source fragmented evidence packet set structure generation unit, a consistency conflict graph structure generation unit, a consistency fusion result set generation unit, and a closed-loop disposal and result version chain generation unit; the units are connected in sequence to implement the method described in any of the above-mentioned methods.

[0025] The key innovations of this invention include:

[0026] (1) Incorporate the boundary data of the target domain, the list of land asset objects, the list of multi-source data sources and the asset unit division rules into the same processing link. Based on the asset unit segmentation, perform multi-source data source capability field registration and anchor point candidate matching processing to generate the acquisition session configuration structure and use it as a unified entry point for subsequent remote sensing image data and lidar point cloud data acquisition, ground survey data acquisition and historical inventory data loading and processing.

[0027] (2) Based on the acquisition session configuration structure, remote sensing image data and lidar point cloud data acquisition, ground survey data acquisition and historical inventory data loading and processing are organized to generate a multi-source fragmented evidence package set structure, so that the data loading and fragmented organization corresponding to the multi-source data source list are carried by the same data set structure and connected to the subsequent consistency analysis link.

[0028] (3) Using the multi-source fragmented evidence package set structure as the input carrier, the geometric consistency sub-vector construction, temporal consistency sub-vector construction and semantic consistency sub-vector construction processes are completed within the same link to generate a consistency conflict graph structure. The fusion constraint rule set loading, evidence priority sequence generation and fusion session primary key registration processes are executed on the consistency conflict graph structure to form a closed-loop configuration update link that returns the acquisition session configuration structure.

[0029] The following are its main beneficial effects:

[0030] (1) To address the inconsistency in referencing caused by the segmented implementation of acquisition session configuration and acquisition tasks and the loose association of outputs in the existing scheme, the asset unit segmentation, multi-source data source capability field registration and anchor point candidate matching processing are unified and converged into an acquisition session configuration structure. This allows remote sensing image data and lidar point cloud data acquisition, ground survey data acquisition and historical inventory data loading processing to be controlled and triggered under the same configuration entry, reducing the repeated configuration and loading of the acquisition link when the multi-source data source list changes.

[0031] (2) To address the issues of field caliber differences and discontinuous source tracing in the cross-source aggregation stage of the existing scheme, a multi-source fragmented evidence package set structure is generated by driving the acquisition session configuration structure. This allows remote sensing image data and lidar point cloud data, as well as ground survey data and historical inventory data, to be loaded and organized within the same set structure, facilitating consistent referencing and link tracing of fragmented data of the same land asset object list in subsequent consistency analysis.

[0032] (3) To address the problem that the consistency conflict graph structure is difficult to continuously generate and reuse due to the independent construction of geometric consistency sub-vectors, temporal consistency sub-vectors and semantic consistency sub-vectors in the existing scheme, the three types of consistency sub-vector construction processes are integrated on the multi-source fragmented evidence package set structure to generate a consistency conflict graph structure. Then, the fusion constraint rule set loading, evidence priority sequence generation and fusion session primary key registration processing are combined to form a write-back link of the collection session configuration structure, so that the multi-source data analysis around the boundary data of the same investigation target domain in the subsequent investigation process has a continuous link organization and session-level configuration update basis. Attached Figure Description

[0033] Figure 1 A flowchart illustrating a land resource asset inventory method based on multi-source data analysis, provided for an embodiment of this application;

[0034] Figure 2 This is a structural block diagram of a land resource asset inventory system based on multi-source data analysis, provided as an embodiment of this application. Detailed Implementation

[0035] Example 1: Refer to Figure 1 This is a flowchart illustrating a land resource asset inventory method based on multi-source data analysis provided by an embodiment of the present invention. The process may include at least steps S100-S400:

[0036] S100: Obtain the boundary data of the target domain, the list of land asset objects, the list of multi-source data sources and the asset unit division rules, perform asset unit segmentation, register the capability fields of multi-source data sources and perform anchor point candidate matching, and generate the collection session configuration structure.

[0037] S200, based on the acquisition session configuration structure, performs ground survey data acquisition and historical inventory data loading and processing, and generates a multi-source fragmented evidence package set structure;

[0038] S300: Based on the multi-source fragmented evidence package set structure, geometric consistency sub-vector construction, temporal consistency sub-vector construction and semantic consistency sub-vector construction are performed to generate a consistency conflict graph structure;

[0039] S400: Based on the consistency conflict graph structure, load the fusion constraint rule set, generate the evidence priority sequence and register the fusion session primary key to generate the collection session configuration structure.

[0040] Step S100 includes at least steps S110-S130:

[0041] S110. Obtain the boundary data of the target domain, the list of land asset objects, the list of multi-source data sources and the asset unit division rules, perform terminology dictionary entry set registration and asset unit segmentation processing, and obtain the asset unit primary key table.

[0042] In this embodiment, the boundary data of the target domain is retrieved from the spatial data platform of the natural resources authority by the data access gateway, or the historical survey data loading module loads existing boundary files from the archive and writes them to the boundary cache. The boundary data of the target domain is defined as a set of boundary records describing the target survey area. Each boundary record includes at least a boundary line string, coordinate reference identifier, boundary version identifier, acquisition time identifier, and data source identifier. The boundary line string consists of several sequentially connected coordinate points. The coordinate reference identifier declares the coordinate system caliber. The boundary version identifier distinguishes between boundary adjustment and revision batches. The acquisition time identifier marks the time the boundary data was entered into the database. The data source identifier marks the source channel of the boundary data. The land asset object list is generated by the survey business configuration service and written to the object list repository. The land asset object list is defined as a set of records of survey object categories and object attribute fields. It includes at least an object type identifier, land category code field, use description field, ownership attribute field set, object priority field, and object version identifier. The object priority field is used to enable stricter segmentation constraints for boundary-sensitive objects during subsequent asset unit segmentation. The multi-source data source list is maintained by the data access gateway and reviewed by the metadata registration service. The multi-source data source list is defined as a directory of data sources participating in the cleanup, and includes at least the data source identifier, source type identifier, acquisition method identifier, coverage identifier, refresh interval field, authorization credential index field, and interface stability identifier. The source type identifier covers remote sensing image data, lidar point cloud data, ground survey data, and historical cleanup data. The acquisition method identifier describes four methods: interface pull, batch loading, offline import, and manual entry. The coverage identifier describes the spatial boundary of the data source coverage. The refresh interval field registers the update cycle caliber. The authorization credential index field points to the credential primary key of the authorization credential library. The interface stability identifier registers the interface availability status and fluctuation level. The asset unit partitioning rules are provided by the rule repository and jointly loaded by the inventory business configuration service and the version management service. The asset unit partitioning rules are defined as a set of rule records used to divide the boundary data of the inventory target domain into several asset units. The rule records contain at least the following fields: land parcel boundary priority field, grid size field, boundary trimming rule field, suspected void repair rule field, void area threshold field, connectivity merging rule field, and rule version identifier field. Among them, the land parcel boundary priority field is used to declare the priority order of land parcel boundaries, administrative boundaries, and task temporary boundaries; the grid size field is used to generate a grid skeleton in the case of missing land parcel boundaries; the boundary trimming rule field is used to declare the enabled status of self-intersection detection, short edge removal, duplicate point removal, and gap closure; the suspected void repair rule field is used to declare the void identification and repair action criteria; the void area threshold field is used to declare the void merging conditions; the connectivity merging rule field is used to declare the merging adjacency relationship determination criteria; and the rule version identifier field is used to establish an association with the boundary version identifier and for subsequent version comparison and calling.

[0043] The terminology dictionary entry set registration is performed by the terminology dictionary service module. A terminology dictionary entry set is defined as a unified set of entries with consistent field definitions and terminology spellings throughout the entire process. Each entry must include at least the term name, a set of field aliases, field type, value constraints, source identifier, version identifier, and effective status fields. Specifically, after receiving the target domain boundary data, land asset object list, multi-source data source list, and asset unit division rules, the terminology dictionary service module first parses the field names, field aliases, and enumeration definitions of the input records to generate a terminology candidate set. This terminology candidate set is then written to the candidate cache table, and a candidate generation session identifier is generated and written to the audit log. Next, conflict detection is performed on the terminology candidate set. Conflict detection covers name conflicts, synonym conflicts, field type conflicts, and value constraint conflicts. The conflict detection output is written to the terminology conflict record table, along with conflict level and conflict location fields. Subsequently, the terminology dictionary service module triggers two paths based on the conflict level field: When the conflict level field meets the automatic decision-making conditions, the terminology dictionary service module generates a decision result based on the source identifier priority field and the version identifier new / old order field, writes the decision result into the terminology dictionary entry set and updates the version identifier, and simultaneously writes the change summary field into the audit log; when the conflict level field indicates manual confirmation, the terminology dictionary service module generates a work order to be confirmed and writes it into the work order queue, and writes the freeze identifier into the terminology conflict record table, so that subsequent steps can skip frozen entries and record the skipping reason when reading the terminology dictionary entry set. Further, when writing back the work order, the terminology dictionary service module performs an unfreezing action on the freeze identifier, writes the unfreezing action work order number field into the audit log, and simultaneously writes the term version identifier into the terminology dictionary entry set for subsequent steps to load according to the version identifier. Thus, the output product registered in the terminology dictionary entry set is the terminology dictionary entry set, whose version identifier field is written into the terminology version identifier field of the segmented context record in this step, and is read as the basis for caliber verification in the subsequent S220 coordinate projection caliber unification processing and timestamp caliber unification processing stages.

[0044] The asset unit segmentation process is run by the asset unit segmentation service module under the scheduling of the task orchestration engine. Specifically, the asset unit segmentation service module reads the boundary line string of the target domain boundary data and loads the asset unit division rules. It also reads the object type identifier and object priority field of the land asset object list simultaneously and generates a segmentation context record. The segmentation context record includes the boundary version identifier, rule version identifier, object version identifier, segmentation session identifier, and terminology version identifier. Subsequently, the asset unit segmentation service module sorts the candidate boundary sources according to the parcel boundary priority field. If a parcel boundary source exists and the boundary version identifier is valid, the parcel boundary is used as the segmentation skeleton and the trimming action corresponding to the boundary trimming rule field is executed. The trimming action includes boundary self-intersection detection, short side segment removal, duplicate point removal, and micro-gap closure, and the abnormal segments generated by the trimming are marked and written into the boundary anomaly record table. If the parcel boundary source is missing or the boundary version identifier is invalid, a mesh skeleton is generated within the boundary of the target domain according to the mesh size field and the boundary trimming action is executed. The boundary trimming action includes intersection trimming of the mesh boundary with the boundary of the target domain, deletion of the mesh outside the boundary, and boundary fitting and trimming. The deviation record generated by the boundary fitting and trimming is written into the trimming deviation record table. Furthermore, if the suspected void repair rule field exists and the boundary anomaly record table contains void annotations, the asset unit segmentation service module extracts the void boundary and performs a void area threshold field comparison. Void units that meet the threshold conditions are merged into adjacent asset units through the connectivity merging rule field. Void units that do not meet the threshold conditions are written into the pending review record table and a review task identifier field is generated. The review task identifier field is bound to the segmentation session identifier field and written to the audit log. The review conclusion of the pending review record table is unfrozen by the work order write-back and merged into the merged record table. The work order number field and change summary field of the merged record table are written to the audit log. In this segmentation link, the parcel boundary priority field, grid size field, and rule version identifier field constitute the minimum parameter set for asset unit segmentation processing. The boundary trimming rule field, suspected void repair rule field, void area threshold field, and connectivity merging rule field are read as an extended parameter set and written to the segmentation context record, so that the primary key field of the same asset unit can reuse the same set of extended parameter calibers in the recalculation scenario. When the task orchestration engine detects a change in the boundary version identifier or rule version identifier, it triggers a recalculation of the split session. During the recalculation, the identifier of the previous split session is written to the preceding session field of the split context record, and the difference summary field is written to the audit log.

[0045] The asset unit primary key table, as an output of this step, is generated by the asset unit splitting service module after splitting and written into the primary key table repository. The asset unit primary key table is defined as the table structure of an asset unit along with its primary key identifier and associated metadata, and includes at least the following fields: asset unit primary key field, spatial range identifier field, object type field, time identifier field, splitting session identifier field, boundary version identifier field, and rule version identifier field. The asset unit primary key field is generated by the primary key generator, which combines the spatial range identifier field with the time identifier field as the primary key seed and generates a fixed-length primary key identifier using an irreversible digest algorithm. The spatial range identifier field records the boundary line string index of the asset unit; the object type field is taken from the object type identifier in the land asset object list; the time identifier field records the splitting completion time; and the splitting session identifier field records the current splitting session. The asset unit primary key table is called as the input object of S120 at the end of this step, and serves as the source of values ​​for the asset unit primary key binding fields in the subsequent multi-source data packet generation stage of S210. Simultaneously, the boundary version identifier field and rule version identifier field within the asset unit primary key table are referenced by the subsequent result version chain structure registration and processing, forming a version traceability chain across main steps. In a specific embodiment, when the target domain for investigation is a county-level administrative area, the boundary data of the target domain is taken from the county-level administrative boundary revision batch. The asset unit segmentation service module prioritizes loading the county-level land parcel boundaries according to the parcel boundary priority field and completes boundary trimming. Hollow units recorded in the pending review record table are merged into adjacent asset units after being determined by the hollow area threshold field. Subsequently, the primary key generator generates the asset unit primary key field and writes it into the asset unit primary key table. The spatial range identifier field and the object type field are backfilled by the segmentation context record, enabling the asset unit primary key table to be directly used by the spatial indexing service for primary key sequence extraction.

[0046] S120. Extract the primary key sequence of the asset unit from the primary key table of the asset unit, perform multi-source data source capability field registration and anchor point candidate matching processing, and generate an anchor point candidate list.

[0047] In this embodiment, the asset unit primary key sequence is extracted and sorted by the spatial index service from the asset unit primary key table. Specifically, the spatial index service reads the spatial range identifier field of the asset unit primary key table and establishes a spatial index. The spatial index uses a grid index code or a parcel index code as the index key, and then outputs the asset unit primary key field sequence as the asset unit primary key sequence according to the index key order. When there are asset unit tags to be reviewed in the asset unit primary key table, the spatial index service synchronously writes the asset unit tags to be reviewed into the sequence attribute field of the asset unit primary key sequence, so that subsequent anchor point candidate matching processing can enable stricter matching constraints for the asset units to be reviewed, and writes the matching constraints into the matching session log.

[0048] The registration of capability fields for multi-source data sources is completed by the metadata registration service in collaboration with the data access gateway. The metadata registration service reads the data source identifier, source type identifier, coverage identifier, refresh interval field, acquisition method identifier, and interface stability identifier from the multi-source data source list, and combines this with the availability, latency statistics, and failure reason code fields from historical fetch logs to generate capability field records. Capability field records must include at least the following fields: spatial resolution field, time refresh interval field, coverage identifier field, acquisition method identifier field, point cloud density level field, image cloud cover level field, interface stability level field, and availability level field. The spatial resolution field is used for registering the pixel size of remote sensing image data; the point cloud density level field is used for registering the point density of lidar point cloud data; the image cloud cover level field describes the occlusion status of remote sensing image data; the interface stability level field is generated by segmenting and mapping the interface stability identifier; and the availability level field is generated by segmenting and mapping the availability field. The metadata registration service writes capability field records into the capability field repository and generates capability field version identifiers and registration session identifiers. The registration session identifier is bound to the data source identifier and recorded in the audit log. When the interface stability identifier or authorization credential index field changes, the metadata registration service triggers capability field recalculation and generates a new capability field version identifier, and writes the capability field version identifier into the capability field repository for the anchor matching service module to load in the matching session.

[0049] During operation, the data access gateway performs round-robin fetching of each data source identifier according to the refresh interval field. When the round-robin fetching fails, the failure reason code field is written to the historical fetching log and the interface stability identifier is triggered to update. After the interface stability identifier is updated, the metadata registration service writes the interface stability level field to the capability field record and updates the capability field version identifier, so that the anchor candidate matching processing reads the capability field record with the same version identifier in the same matching session.

[0050] Anchor point candidate matching is performed by the anchor point matching service module. This module receives the asset unit primary key sequence and loads the capability field records from the capability field repository. It then generates a matching session primary key and writes it to the matching session log. The anchor point matching service module constructs an anchor point candidate matching request for each asset unit primary key field. This request includes at least the asset unit primary key field, spatial extent identifier field, time identifier field, object type field, boundary version identifier field, and rule version identifier field. Within the capability field records, the anchor point matching service module filters data source identifiers whose coverage identifier field overlaps with their spatial extent identifier field. It then generates a candidate ranking based on the spatial resolution field priority, point cloud density level field priority, and interface stability level field priority. When the object type field is labeled as a boundary-sensitive object such as forest land or construction land, the candidate ranking rule increases the point cloud density level field priority and writes it to the parameter tuning summary. Subsequently, the anchor point matching service module performs time availability verification on the candidate data source identifiers. This verification includes comparing the time refresh interval field with the current time window configuration and filtering the failure reason code field of historical fetch logs. Data source identifiers that pass the time availability verification are marked as available anchor point sources, while those that do not are marked as anchor point sources to be supplemented, and this is written to the supplementary sampling reason field. If the sequence attribute field of the asset unit's primary key sequence contains an asset unit marker to be verified, the anchor point matching service module adds an anchor point redundancy quantity field and an anchor point cross-verification marker field within the matching constraints. The anchor point redundancy quantity field is mapped from the anchor point redundancy quantity field in the asset unit partitioning rules, and the anchor point cross-verification marker field is used to declare that the same asset unit needs cross-verification between remote sensing image data and lidar point cloud data. After candidate sorting, the anchor point matching service module extracts the first few anchor point source identifier fields based on the anchor point redundancy quantity field and writes them into the candidate anchor point source set field of the anchor point candidate list. Simultaneously, it writes the generated summary of the candidate anchor point source set field into the matching session log.

[0051] The anchor candidate list, as an output of this step, is generated by the anchor matching service module after completing the full matching of the primary key sequences of asset units and written to the anchor list repository. The anchor candidate list is defined as a list structure where the anchor source selection for an asset unit corresponds to the anchor availability status. It includes at least the following fields: spatial baseline anchor field, time baseline anchor field, anchor source identifier field, anchor availability identifier field, reason for supplementary collection field, and asset unit primary key field. Specifically, the spatial baseline anchor field is defined as a spatial anchor record bound to the asset unit's spatial range identifier field, containing the anchor boundary index and coordinate reference identifier; the time baseline anchor field is defined as a time anchor record bound to the asset unit's time identifier field, containing the time window identifier and refresh cycle caliber; the anchor source identifier field records the identifier of the selected data source; the anchor availability identifier field records the time availability verification result; and the reason for supplementary collection field records the summary of the failure reason code. The anchor point candidate list is invoked as input to S130 at the end of this step and is used to generate the sampling space range field and sampling time window field in the subsequent acquisition session configuration structure generation stage. At the same time, the anchor point source identifier field and anchor point availability identifier field are also used for subsequent evidence chain field registration and processing, forming an evidence tracing chain across the main steps. Meanwhile, the reason field for supplementary sampling in the anchor point candidate list is read by the subsequent handling strategy matching process to generate supplementary sampling task parameters and write back to the acquisition session configuration structure.

[0052] S130. Perform baseline anchor point layer registration processing, verification sampling layer task orchestration processing, and change capture layer encrypted sampling task orchestration processing on the anchor point candidate list to generate a data acquisition session configuration structure.

[0053] In this embodiment, the data acquisition session configuration structure is generated by the data acquisition orchestration service module. This module is connected to the task orchestration engine, rule repository, and terminology dictionary service module, and is used to complete anchor point registration, sampling task generation, trigger condition fixing, session audit registration, and version field binding. After receiving the anchor point candidate list, the data acquisition orchestration service module generates an orchestration session primary key and reads the version identifier of the terminology dictionary entry set and the rule version identifier field of the asset unit primary key table. Then, it groups the anchor point candidate list into asset unit anchor point candidate groups according to the asset unit primary key field. These asset unit anchor point candidate groups serve as the grouping input for the subsequent three-layer data acquisition orchestration structure fields, and a merging statistics field is registered in the orchestration session log. This merging statistics field records the number of candidate anchor point sources corresponding to each asset unit primary key field.

[0054] The baseline anchor point layer registration process extracts spatial and temporal baseline anchor point fields for each asset unit's anchor point candidate group and generates a baseline anchor point registration record. The baseline anchor point registration record is defined as the registration carrier of the binding relationship between the anchor point source and the asset unit, and includes at least the following fields: anchor point source identifier, anchor point availability identifier, coordinate reference identifier, time window identifier, registration time identifier, orchestration session primary key, boundary version identifier, rule version identifier, and capability version identifier. When the anchor point availability identifier field is marked as an anchor point source to be supplemented, the acquisition orchestration service module writes an anchor point supplementation flag field and writes the supplementation reason field into the supplementation reason record. The supplementation reason record is written to the supplementation queue and bound to the orchestration session primary key. After reading the supplementation queue, the task orchestration engine generates a supplementation scheduling task and writes it to the acquisition queue, enabling subsequent acquisition actions to be automatically triggered. The data acquisition and orchestration service module writes the baseline anchor point registration record into the anchor point registration repository, and writes the change summary field, version identifier, and orchestration session primary key of the writing action into the audit log. The audit log is bound to the orchestration session primary key to make the anchor point registration process auditable.

[0055] The verification sampling layer task orchestration process is executed sequentially after the baseline anchor point layer registration process is completed. The acquisition orchestration service module reads the object type field from the asset unit primary key table, synchronously reads the ownership attribute field set and use description field from the land asset object list, and generates a verification sampling template record. The verification sampling template record is defined as a mapping template from asset object category to sampling action set, and at least includes a set of mandatory sampling attribute fields, a sampling frequency caliber field, a sampling carrier identifier field, and a sampling evidence chain field template. The set of mandatory sampling attribute fields, the sampling frequency caliber field, and the sampling carrier identifier field constitute the minimum parameter set for verification sampling layer task orchestration. The sampling evidence chain field template is used to solidify the field caliber for subsequent evidence chain field registration processing. In extended implementations, the verification sampling template record is also allowed to include a survey form version identifier field, a field image index rule field, and a positioning accuracy constraint field. The data acquisition and orchestration service module generates verification sampling task records for each asset unit's primary key field according to the verification sampling template record. Each verification sampling task record includes at least the following fields: sampling spatial range, sampling time window, sampling carrier identifier, sampling evidence chain, task status, task triggering condition, task priority, and task retry strategy. The sampling spatial range field originates from the anchor point boundary index associated with the spatial benchmark anchor point field; the sampling time window field originates from the time window identifier associated with the time benchmark anchor point field; and the task triggering condition field includes a combination of anchor point availability identifier, object type, boundary version identifier, rule version identifier, and capability version identifier. When the combined conditions are met and the sampling time window field is in an executable state, the task orchestration engine writes the verification sampling task record to the acquisition queue and sets the task status field to "issued." Subsequently, the acquisition execution node retrieves the task according to the sampling carrier identifier field and writes the execution receipt to the task status field. When the execution receipt indicates failure, the task retry strategy field drives the task orchestration engine to generate a retry task and write it to the acquisition queue, while simultaneously writing the failure reason summary field to the audit log.

[0056] The change capture layer encrypted sampling task orchestration process is executed after the verification sampling layer task orchestration process is completed. The acquisition orchestration service module reads the difference information between adjacent versions from the change summary field of the subsequent result version chain structure and compares it with the spatial range identifier field and object type field of the asset unit primary key table. If the spatial range identifier field or object type field corresponding to the primary key field of the same asset unit changes in adjacent versions, the asset unit is marked as a change capture candidate object. At the same time, the acquisition orchestration service module reads the anchor availability identifier field and the merging statistics field from the anchor candidate list. If the anchor availability identifier field is marked as a source of anchors to be supplemented or the number of candidate anchor sources marked in the merging statistics field is lower than the anchor redundancy number field, the asset unit is marked as a high uncertainty object. For candidate objects and highly uncertain objects in change capture, the acquisition orchestration service module generates encrypted sampling task records. These records, based on the fields in the verification sampling task record, are supplemented with an encrypted sampling level field, an encrypted sampling time window extension field, an anomaly reuse trigger identifier field, and an encrypted task reason summary field. The encrypted sampling level field is obtained by matching the object type field with the change level mapping table. The encrypted sampling time window extension field is obtained by matching the time window identifier with the extended rule record. The anomaly reuse trigger identifier field is associated with the anomaly type identifier field of the subsequent anomaly event set investigation, enabling a new round of encrypted sampling task orchestration and write-back to the acquisition queue after an anomaly event is generated. After generating the encrypted sampling task record, the acquisition orchestration service module writes it to the encrypted task log. The encrypted task log includes a trigger basis summary field, a change level mapping table version identifier field, an extended rule record version identifier field, and an orchestration session primary key, and is associated with the task status field of the encrypted sampling task record for subsequent result version chain structure registration and processing.

[0057] The acquisition session configuration structure, as an output of this step, is generated and written to the acquisition session configuration repository by the acquisition orchestration service module after completing the baseline anchor point layer registration processing, verification sampling layer task orchestration processing, and change capture layer encrypted sampling task orchestration processing. The acquisition session configuration structure is defined as the session configuration carrier that drives subsequent multi-source data acquisition and aggregation. It includes at least the following fields: sampling carrier identifier field, sampling time window field, sampling spatial range field, sampling evidence chain field, three-layer acquisition orchestration structure field, orchestration session primary key, boundary version identifier field, rule version identifier field, capability field version identifier field, and terminology version identifier field. The three-layer acquisition orchestration structure field consists of baseline anchor point registration records, verification sampling task records, and encrypted sampling task records, and is bound to the asset unit primary key field. The sampling carrier identifier field describes the subject of the data acquisition, which includes remote sensing image data access tasks, lidar point cloud data access tasks, ground survey data acquisition tasks, and historical inventory data loading tasks. The sampling time window field describes the time window during which the task is allowed to execute. The sampling spatial range field describes the set of spatial anchor point boundary indexes covered by the task. The sampling evidence chain field is used for subsequent evidence chain field registration and processing to write the source identifier, session identifier, and sampling action summary. This acquisition session configuration structure is called as the input object of S210 at the end of this step, allowing S210 to trigger the acquisition and processing of remote sensing image data and lidar point cloud data, ground survey data, and historical inventory data loading. It is also used as the configuration input for generating the preprocessing fragment set in the subsequent coordinate projection standardization and timestamp standardization stages of S220. At the same time, the sampling evidence chain field and the orchestration session primary key in the acquisition session configuration structure are called by the evidence chain field registration and processing in the subsequent S230, and an association record is established with the fragment unique identifier field generated in the fragment unique identifier registration and processing stage. Furthermore, the sampling space range field and sampling time window field in the acquisition session configuration structure are expanded into a data source retrieval window by the S210 acquisition execution node. Multi-source data packets under the same orchestration session primary key are written into the binding field of the asset unit primary key binding fragment when the preprocessed fragment set is generated in S220. Subsequently, when the asset unit consistency fingerprint vector is constructed in S310, the anchor source identifier field in the sampling evidence chain field is read and written into the source caliber field of the consistency sub-vector construction processing.

[0058] Step S200 includes at least steps S210-S230:

[0059] S210. Obtain the acquisition session configuration structure, perform remote sensing image data and lidar point cloud data acquisition and processing, ground survey data acquisition and processing, and historical inventory data loading and processing to obtain multi-source data packets;

[0060] In this embodiment, the acquisition session configuration structure comes from the output of the preceding S130. The acquisition session configuration structure serves as a unified entry point for acquisition scheduling and data access. It is written into the acquisition session configuration repository by the acquisition orchestration service module and then read by the task orchestration engine. Specifically, the task orchestration engine reads the sampling carrier identifier field, sampling time window field, sampling spatial range field, sampling evidence chain field, three-layer acquisition orchestration structure field, orchestration session primary key, boundary version identifier field, rule version identifier field, capability field version identifier field, and terminology version identifier field from the acquisition session configuration structure. It then decomposes tasks under the same orchestration session primary key into four task units: remote sensing image data acquisition task, lidar point cloud data acquisition task, ground survey data acquisition task, and historical inventory data loading task. Each task unit establishes a binding relationship with the asset unit primary key field. This binding relationship is determined by the spatial anchor point boundary index pointed to by the sampling spatial range field, and the time window boundary is determined by the sampling time window field. The binding relationship is then written into the acquisition task registration table, which includes the task primary key field, task category identifier field, data source identifier field, asset unit primary key field, sampling spatial range field, sampling time window field, task status field, and task trigger condition field. The task trigger condition field is composed of the anchor point availability identifier field, the interface stability level field, the authorization credential index field status, and the sampling time window field status. When the task trigger condition field meets the executable state, the task orchestration engine pushes the task to the collection queue and registers the orchestration session primary key field, the task primary key field, the trigger condition summary field, and the distribution time identifier field in the audit log to form a traceable scheduling link.

[0061] The remote sensing image data acquisition and processing is performed by the remote sensing access unit on the cloud-based acquisition execution node, which is connected to the data access gateway. The remote sensing image data is defined as a raster image set record with georeferenced information. This image set record includes at least an original image file index, image metadata records, and an image acquisition time identifier. The image metadata records include coordinate reference identifiers, image resolution identifiers, band set identifiers, cloud cover level fields, and coverage area identifier fields. Specifically, the remote sensing access unit reads the data source identifier field from the acquisition task registration table and loads the credential record pointed to by the authorization credential index field. Then, it generates spatial search conditions based on the spatial anchor point boundary index given by the sampling spatial range field and generates temporal search conditions based on the sampling time window field. The remote sensing access unit writes the spatial search conditions and temporal search conditions into the search request record and calls the data source access adapter to retrieve the image list. For the retrieved image list, the remote sensing access unit performs filtering based on the spatial resolution field and image cloud cover level field corresponding to the capability version identifier field. During the filtering process, a filtering summary field is generated and written to the image filtering log. When the image cloud cover level field exceeds the allowable threshold or the coverage identifier field does not meet the coverage relationship with the spatial anchor point boundary index, the remote sensing access unit marks the image entry as unavailable and writes it to the reason for re-acquisition in the reason field. At the same time, the reason for unavailability is bound to the task primary key field and written to the audit log. When the filtering is successful, the remote sensing access unit performs block download or object storage mounting on the image file and writes the original image file index field to the original file index field. At the same time, the image acquisition time identifier is written to the acquisition time field.

[0062] The lidar point cloud data acquisition and processing is run on the cloud-based acquisition execution node by the point cloud access unit, which is connected to the data access gateway. The lidar point cloud data is defined as a point cloud set record containing 3D point records. This set record includes at least an original point cloud file index, point cloud metadata records, and a point cloud acquisition time identifier. The point cloud metadata records include a coordinate reference identifier, a point cloud density level field, a scan trajectory index field, and a coverage area identifier field. Specifically, the point cloud access unit reads the data source identifier field, sampling spatial range field, and sampling time window field from the acquisition task registration table, loads the authorization credential index field, generates a point cloud retrieval request, and calls the data source access adapter to retrieve the point cloud list. The point cloud access unit filters candidate point cloud entries according to the point cloud density level field and performs coverage relationship verification on the scan trajectory index field. The coverage relationship verification record is written to the point cloud coverage verification log and bound to the task primary key field. For point cloud entries that pass the screening, the point cloud access unit performs point cloud file retrieval or object storage mounting, generates the original point cloud file index field and writes it to the original file index field, and records the point cloud acquisition time as the acquisition time field; for point cloud entries that fail the screening, the point cloud access unit writes the reason field to be supplemented and triggers the supplementary acquisition queue write action. The supplementary acquisition queue write action is bound to the orchestration session primary key for subsequent acquisition session configuration structure writeback link reading.

[0063] The ground survey data acquisition and processing is run by the ground acquisition access unit on the edge-cloud collaborative acquisition execution node. The ground acquisition access unit is connected to the mobile acquisition terminal, the field acquisition service, and the data access gateway. The ground survey data is defined as a set of field survey records. Each field survey record includes at least a survey form version identifier field, a location trajectory record index, a field image index, a survey time identifier, an investigator identifier, and an asset unit primary key field. The location trajectory record index points to a location trajectory set, which consists of several location point records. Each location point record includes a coordinate reference identifier, a location time identifier, and a location accuracy field. Specifically, the ground acquisition access unit reads the sampling evidence chain field from the acquisition session configuration structure and generates an acquisition instruction package. The acquisition instruction package includes an asset unit primary key field, a sampling space range field, a sampling time window field, and a set of mandatory attribute fields. After the field acquisition service sends the acquisition instruction package to the mobile acquisition terminal, the mobile acquisition terminal starts the acquisition session when the sampling time window field is valid and writes the survey form version identifier field into the session header information. The mobile data acquisition terminal collects the location trajectory record index, on-site image index, and mandatory attribute field set within the sampling space range field coverage area, and writes the submission time identifier and terminal identifier field each time it is submitted; after receiving the submission, the on-site data acquisition service performs form field integrity verification and basic legality verification. Records that pass the verification are written to the ground survey data temporary storage area, and records that fail the verification are written to the record table to be corrected and the terminal prompt information is written back. At the same time, the task primary key field, failure reason summary field, and write-back time identifier field are recorded in the audit log.

[0064] The historical data loading and processing is carried out by the historical data loading unit on the cloud-based data collection and execution node. This unit is connected to the historical data archive, metadata registration service, and data access gateway. The historical data is defined as a collection of historical data collection result files and their metadata records. These files include vector boundary files, attribute table files, and thematic map output files. Metadata records include at least the following fields: historical version identifier, boundary version identifier, rule version identifier, coordinate reference identifier, time identifier, and source identifier. Specifically, the historical data loading unit retrieves historical data collection result files according to the sampling spatial range field, filters historical result batches whose historical version identifier fields fall within the sampling time window field, and writes the filtered summary field to the historical filtering log. For the filtered historical result batches, the historical data loading unit performs file index loading, writing the historical result file index to the original file index field and simultaneously writing the historical version identifier field to the source version identifier field for reference in subsequent evidence chain field registration and processing.

[0065] After completing the above four types of data acquisition and processing, the acquisition execution node will aggregate the remote sensing image data acquisition results, lidar point cloud data acquisition results, ground survey data acquisition results, and historical inventory data loading results into a multi-source data package and write it into the multi-source data package repository. The multi-source data package, as the output of this step, is defined as an encapsulation structure of multiple source raw inputs under the same orchestration session primary key. The multi-source data package includes at least the following fields: source type identifier field, data source identifier field, acquisition method identifier field, original file index field, acquisition time field, asset unit primary key binding field, boundary version identifier field, rule version identifier field, capability field version identifier field, and terminology version identifier field. The asset unit primary key binding field is filled back through the binding relationship in the acquisition task registration table. After this step, the multi-source data packet is called as the input object of S220. Subsequently, S220 extracts the asset unit primary key binding fragment from the multi-source data packet and performs unified processing of coordinate projection caliber and unified processing of timestamp caliber. At the same time, the original file index field and collection time field in the multi-source data packet are also read and written into the multi-source fragment evidence packet set structure by the evidence chain field registration processing of subsequent S230, forming a traceable link across steps.

[0066] S220. Extract asset unit primary key binding fragments from multi-source data packets, perform unified processing of coordinate projection caliber and timestamp caliber, and generate a preprocessed fragment set.

[0067] In this embodiment, the input source for this step is the multi-source data packet output from S210. Preprocessing is executed by the preprocessing orchestration unit on a cloud processing node. The preprocessing orchestration unit is connected to the data access gateway, spatial reference registration service, and time-based registration service. The asset unit primary key binding fragment is defined as a set of original data fragments obtained by merging the multi-source data packet according to the asset unit primary key field. The original data fragments include four categories: remote sensing image data fragments, lidar point cloud data fragments, ground survey data fragments, and historical inventory data fragments. For remote sensing image data fragments, the original data fragment consists of image blocks pointed to by the original image file index field and carries image metadata records. For lidar point cloud data fragments, the original data fragment consists of point cloud blocks pointed to by the original point cloud file index field and carries point cloud metadata records. For ground survey data fragments, the original data fragment consists of a set of survey records and carries a survey form version identifier field and a positioning trajectory record index. For historical inventory data fragments, the original data fragment consists of a set of historical result file indexes and carries a historical version identifier field and metadata records. The preprocessing orchestration unit first reads the asset unit primary key binding field from the multi-source data packet, aggregates the original file index fields corresponding to the same asset unit primary key field into a shard index group, and records the shard index group as a shard index field. The shard index field, along with the source type identifier field, the data source identifier field, and the collection time field, together form the shard location key. After the shard location key is generated, the preprocessing orchestration unit writes it into the preprocessing session log. The preprocessing session log records the orchestration session primary key, the asset unit primary key field, the shard index field, and the generation time identifier field.

[0068] The unified processing of coordinate projection aperture is achieved by the spatial reference registration service providing an aperture mapping table and the preprocessing orchestration unit executing the mapping action. The minimum parameter set for unified coordinate projection aperture processing includes a coordinate reference identifier, a projection parameter identifier field, and a spatial anchor point boundary index and boundary version identifier field. The coordinate reference identifier comes from image metadata records, point cloud metadata records, positioning trajectory set records, or historical inventory data metadata records. The projection parameter identifier field is mapped by the spatial reference registration service according to the coordinate reference identifier. The spatial anchor point boundary index and boundary version identifier field comes from the version field carried in the multi-source data packet. Specifically, the preprocessing orchestration unit reads the coordinate reference identifier from the image metadata record of the remote sensing image data segment and obtains the image geographic transformation record. The image geographic transformation record includes pixel positioning parameters and rotation parameters. The preprocessing orchestration unit writes the image geographic transformation record and the projection parameter identifier field into the image projection conversion request and executes a raster reprojection action. The raster reprojection action outputs the reprojected image segment index and writes it into the segment index field, while simultaneously updating the coordinate reference identifier to a unified coordinate reference identifier. For LiDAR point cloud data segments, the preprocessing and orchestration unit reads the coordinate reference identifier from the point cloud metadata record and obtains the trajectory coordinate caliber pointed to by the scan trajectory index field. Then, it performs coordinate transformation on the point coordinate record according to the projection parameter identifier field. After the coordinate transformation, a transformed point cloud segment index is generated and written to the segment index field. Simultaneously, the point cloud coordinate reference identifier is updated to the unified coordinate reference identifier. For ground survey data segments, the preprocessing and orchestration unit reads the coordinate reference identifier from the positioning trajectory set record and performs coordinate transformation. The coordinate transformation result is written to the positioning coordinate field and associated with the asset unit primary key field. When the positioning trajectory set record lacks a coordinate reference identifier, the preprocessing and orchestration unit reads the terminal coordinate caliber marker from the sampling evidence chain field of the acquisition session configuration structure and writes it to the caliber missing record table. The caliber missing record table is bound to the task primary key field and written to the audit log. For historical data segments, the preprocessing and orchestration unit reads the coordinate reference identifier and projection parameter identifier fields from the metadata record. If the coordinate reference identifier corresponding to the historical version identifier field is inconsistent with the unified coordinate reference identifier, a vector geometric reprojection action is performed and written to the reprojection result index. After completing the unified processing of coordinate projection aperture, the preprocessing and arrangement unit writes the unified coordinate system identifier field and projection parameter identifier field into the preprocessing segment header information. The preprocessing segment header information is bound to the segment index field for reference in the subsequent quality self-inspection field generation and processing.

[0069] The unified timestamp caliber processing is achieved by a caliber mapping table provided by the timestamp caliber registration service and the mapping action executed by the preprocessing orchestration unit. The minimum parameter set for unified timestamp caliber processing includes an acquisition time field, a time window identifier, and a timestamp format caliber field. The acquisition time field originates from the multi-source data packets, the time window identifier comes from the sampling time window field of the acquisition session configuration structure, and the timestamp format caliber field is provided by the timestamp caliber registration service. Specifically, the preprocessing orchestration unit reads the image data segments to obtain time identifiers and parses them with the timestamp format caliber field, generating a unified timestamp field; it reads the point cloud data segments to obtain time identifiers and generates a unified timestamp field; it reads the survey time identifiers from ground survey data segments and combines them with the terminal submission time identifier to generate a unified timestamp field; and it reads the time identifier field corresponding to the historical version identifier field from historical inventory data segments and generates a unified timestamp field. The preprocessing orchestration unit merges the unified timestamp field with the time window identifier, outputs the time window marker field, and writes it into the preprocessing segment header information. If the time field is missing or fails to be parsed, the preprocessing orchestration unit writes the time anomaly marker field into the time anomaly record table and binds it to the shard index field. At the same time, the time anomaly record table is written into the audit log. The time anomaly marker field is read as a component of the quality field in the subsequent S230 quality self-inspection field generation process, and is referenced as the source caliber of the differential dimension identifier field in the subsequent S320 cross-source differential field extraction stage.

[0070] After completing the unified processing of coordinate projection and timestamp standards, the preprocessing orchestration unit generates a preprocessed fragment set and writes it to the preprocessed fragment repository. The preprocessed fragment set is the output of this step, defined as a set structure of preprocessed fragments merged according to the asset unit's primary key field. The preprocessed fragment set includes at least the following fields: coordinate system identifier, projection parameter identifier, timestamp, fragment index, source type identifier, data source identifier, asset unit primary key, boundary version identifier, rule version identifier, capability field version identifier, and terminology version identifier. The preprocessed fragment set is called as the input object of S230 after this step. Subsequently, S230 performs quality self-inspection field generation processing, evidence chain field registration processing, and fragment unique identifier registration processing on the preprocessed fragment set, and finally generates a multi-source fragmented evidence package set structure. At the same time, the coordinate system identifier field, projection parameter identifier field, and timestamp field in the preprocessed fragment set are also read by the caliber field of the geometric consistency sub-vector construction processing and temporal consistency sub-vector construction processing in the subsequent S310, forming a connection relationship across the main steps.

[0071] S230. Perform quality self-inspection field generation processing, evidence chain field registration processing, and fragment unique identifier registration processing on the preprocessed fragment set to generate a multi-source fragment evidence package set structure.

[0072] In this embodiment, the input source for this step is the preprocessed fragment set output by S220. This step is run by the evidence encapsulation unit on a cloud processing node. The evidence encapsulation unit is connected to the acquisition session configuration repository, audit log service, quality rule repository, and identifier generation service. The input for the quality self-inspection field generation process includes a fragment index field, a source type identifier field, a coordinate system identifier field, a projection parameter identifier field, a timestamp field, and an asset unit primary key field. The minimum parameter set for the quality self-inspection field generation process includes a fragment integrity judgment rule field, a time window marker field, and a caliber consistency judgment rule field. The fragment integrity judgment rule field and the caliber consistency judgment rule field come from the quality rule repository, and the time window marker field comes from the preprocessed fragment set. Specifically, the evidence encapsulation unit reads the cloud cover level field and image resolution identifier from the raster reprojection action log of remote sensing image data segments, generates an image integrity marker field according to the segment integrity judgment rule field, generates an image aperture consistency marker field according to the aperture consistency judgment rule field, and combines the image integrity marker field and image aperture consistency marker field with the time window marker field to generate an image quality field; for lidar point cloud data segments, it reads the point cloud density level field, coverage verification log and coordinate transformation log, generates a point cloud integrity marker field and a point cloud aperture consistency marker field, and combines them to generate a point cloud quality field; for ground survey data segments, it reads the form field integrity verification record, positioning accuracy field and positioning coordinate field, generates a survey integrity marker field and a survey aperture consistency marker field, and combines them to generate a survey quality field; for historical inventory data segments, it reads the historical screening log, historical version identifier field and vector geometric reprojection log, generates a historical integrity marker field and a historical aperture consistency marker field, and combines them to generate a historical quality field. The evidence encapsulation unit uniformly writes various quality fields into the quality field and binds the same source type identifier field of the quality field to the shard quality record table. When a quality field triggers an unavailable state, the evidence encapsulation unit writes the unavailable reason summary field into the shard quality record table and into the audit log. The audit log records the orchestration session primary key field, asset unit primary key field, shard index field and unavailable reason summary field, which are then read and processed by the subsequent S430 handling strategy to generate supplementary collection task parameters.

[0073] The evidence chain field registration process is completed by the evidence encapsulation unit calling the acquisition session configuration repository. The registration objects are the sampled evidence chain fields in the acquisition session configuration structure and the orchestration session primary key. Specifically, the evidence encapsulation unit reads the asset unit primary key field from the preprocessed shard set and looks up the sampled evidence chain fields bound to the asset unit primary key field in the acquisition session configuration repository. The sampled evidence chain fields include at least the anchor source identifier field, data source identifier field, acquisition method identifier field, task primary key field, terminal identifier field or execution node identifier field, submission time identifier field and boundary version identifier field, rule version identifier field, capability field version identifier field, and terminology version identifier field. The evidence encapsulation unit writes the sampled evidence chain fields into the evidence chain fields and binds the evidence chain fields to the shard index fields and writes them into the evidence chain registration table. At the same time, it registers the evidence chain registration action summary field and the registration time identifier field in the audit log. For ground survey data fragments, the evidence encapsulation unit also reads the investigator identifier and form version identifier fields from the field acquisition service and writes them into the evidence chain field, ensuring that ground survey evidence under the same asset unit's primary key field establishes a consistent session tracing standard with remote sensing, point cloud, and historical evidence. For historical inventory data fragments, the evidence encapsulation unit writes the historical version identifier field into the evidence chain field and establishes a link with the arrangement session primary key, allowing subsequent result version chain structure registration and processing to directly reference it. If the acquisition session configuration repository fails to be checked or the evidence chain field is missing, the evidence encapsulation unit writes the evidence chain missing marker field into the evidence chain registration table and the audit log, and simultaneously adds the fragment index field to the pending review record table. The pending review record table is bound to the task primary key field and is read for subsequent handling strategy matching and processing, forming an automated abnormal closed-loop trigger entry point.

[0074] The fragment unique identifier registration process is completed collaboratively by the identifier generation service and the evidence encapsulation unit. The inputs to this process include the asset unit primary key field, source type identifier field, data source identifier field, fragment index field, timestamp field, and orchestration session primary key. The fragment unique identifier is defined as a non-conflicting identifier record for a fragment at the session dimension. Specifically, the identifier generation service generates an identifier seed by combining the asset unit primary key field, data source identifier field, timestamp field, fragment index field, and orchestration session primary key according to a fixed concatenation rule, and then performs an irreversible digest algorithm on the identifier seed to generate the fragment unique identifier field. The evidence encapsulation unit writes the fragment unique identifier field into the fragment identifier record table and establishes a related record with the evidence chain registration table. Simultaneously, it writes the fragment unique identifier field into the evidence encapsulation header information, which also includes a quality field, a caliber field, and an evidence chain field. The caliber field is composed of a coordinate system identifier field, a projection parameter identifier field, a timestamp field, a boundary version identifier field, a rule version identifier field, a capability field version identifier field, and a terminology version identifier field. The evidence encapsulation unit writes the caliber field into the evidence encapsulation header information and binds it to the fragment unique identifier field, so that the subsequent S310 consistency sub-vector construction process can directly obtain a unified caliber input when reading the multi-source fragment evidence package set structure. If the fragment unique identifier field fails to generate or a conflict occurs, the identifier generation service writes the conflict flag field into the fragment identifier record table and triggers a recalculation action. The recalculation action registers the recalculation count field, the recalculation reason summary field, and the generation time identifier field in the audit log. When the recalculation count field exceeds a preset threshold, the evidence encapsulation unit writes the fragment index field into the pending review record table and writes it back to the collection queue. After the subsequent collection session configuration structure write-back link reads the data, it generates a supplementary collection task record and updates the sampling time window field.

[0075] After completing the quality self-inspection field generation, evidence chain field registration, and fragment unique identifier registration, the evidence encapsulation unit generates a multi-source fragment evidence package set structure and writes it into the evidence package repository. This multi-source fragment evidence package set structure, as the output of this step, is defined as an evidence package set structure merged according to the asset unit primary key field. The multi-source fragment evidence package set structure includes at least a fragment unique identifier field, a quality field, a caliber field, an evidence chain field, an asset unit primary key field, a source type identifier field, a data source identifier field, a fragment index field, and a timestamp field. The fragment unique identifier field and the evidence chain field constitute a traceable primary key; the quality field and the caliber field constitute a recalcible caliber; and the fragment index field establishes a jump relationship with the original file index field. The multi-source fragmented evidence package set structure is called as the input object of S310 after this step. Subsequently, S310 obtains the multi-source fragmented evidence package set structure and performs geometric consistency sub-vector construction processing, temporal consistency sub-vector construction processing and semantic consistency sub-vector construction processing. At the same time, the quality field and evidence chain field in the multi-source fragmented evidence package set structure are referenced in the conflict residual field calculation processing and conflict attribution label field generation processing stages of S420, forming a closed-loop link across main steps from the acquisition session configuration structure to the consistency fusion result set.

[0076] Step S300 includes at least steps S310-S330:

[0077] S310. Obtain the multi-source fragmented evidence package set structure, perform geometric consistency sub-vector construction processing, temporal consistency sub-vector construction processing and semantic consistency sub-vector construction processing to obtain the asset unit consistency fingerprint vector;

[0078] In this embodiment, the multi-source fragmented evidence package set structure is derived from the output of preceding step S230 and includes a fragment unique identifier field, a quality field, a caliber field, an evidence chain field, an asset unit primary key field, a source type identifier field, a data source identifier field, and a timestamp field. This step is run by the consistency calculation unit on a cloud computing node. The consistency calculation unit is connected to the spatial analysis service, the time series service, and the semantic parsing engine to construct a comprehensive fingerprint description at the asset unit level in the multi-source data consistency analysis layer. Specifically, the acquisition module first aggregates and groups the multi-source fragmented evidence package set structure based on the asset unit primary key field to establish an asset unit index table. Under the same asset unit index table, it reads the fragment unique identifier field and its associated coordinate system identifier, projection parameter identifier, timestamp field, and semantic tag field to form the input set for the construction of the consistency sub-vector.

[0079] The geometric consistency sub-vector construction process is performed by the spatial analysis service, defined as calculating geometric overlap, positional deviation, and boundary fit for spatial morphology information from various sources. Specifically, the spatial analysis service reads the coordinate system identifier and projection parameter identifier from the caliber field, projects geometric fragments from different sources onto a standard coordinate frame, and generates a standardized geometric contour set through the spatial vector overlay module. The geometric contour set includes a set of boundary points, center coordinate points, and boundary difference records. Subsequently, the spatial analysis service calculates the spatial overlap field based on the boundary difference records and constructs a morphological offset field according to the directional distribution of the boundary differences. The geometric consistency sub-vector is composed of the spatial overlap field, morphological offset field, and boundary fit field, and is bound to the corresponding asset unit primary key field record. If geometric contour data is missing or overlap anomalies occur, the system records an anomaly marker field and a fragment unique identifier field in the audit log. Subsequent S320 differential sorting processing will read this anomaly marker as a priority comparison dimension.

[0080] The construction of the time-series consistency sub-vector is performed by the time series service, defined as time-series alignment and change stability modeling for multi-source data acquisition time. Specifically, the time series service extracts the acquisition time, submission time, and historical version identifier fields from the evidence chain field, and converts them into a standard timeline format. Then, it aggregates time series segments based on the asset unit primary key field, calculating the sampling density field, time drift field, and update interval field between time series. The system further extracts the rate of change field between consecutive versions through a sliding window mechanism to characterize data time consistency. The time-series consistency sub-vector is formed by combining the sampling density field, time drift field, and rate of change field, where the rate of change field indicates the stability of multi-source sampled data in the time dimension. When any source data has an anomaly in the timestamp field or a version gap, the system marks the anomaly time point in the time-series anomaly record table and writes the record index into the differential comparison list for use by the S320 cross-source differential field extraction.

[0081] The semantic consistency sub-vector construction process is executed by the semantic parsing engine, defined as the consistency modeling of asset units at non-spatial levels such as attributes, categories, and business tags. Specifically, the semantic parsing engine reads source identifiers, sampled evidence types, investigator identifiers, and historical attribute tags from the evidence chain field, uniformly mapping semantic descriptions from different sources under the primary key field of the same asset unit to a lexical standard set. Within the lexical standard set, the semantic parsing engine builds an attribute correspondence matrix based on the terminology dictionary entry set, merges synonyms, marks conflicts for contradictory terms, and generates a semantic difference marker field. Further, the semantic parsing engine calculates a consistency distribution field for similar tag values ​​and outputs a semantic confidence field. The semantic consistency sub-vector consists of an attribute consistency field, a semantic difference marker field, and a semantic confidence field, and is bound to the asset unit's primary key field. If semantic mapping is missing or parsing fails, the semantic parsing engine writes the erroneous entry marker to a semantic anomaly record table, with the record index bound to the asset unit's primary key field.

[0082] After constructing the geometric, temporal, and semantic consistency sub-vectors, the consistency calculation unit merges the three types of sub-vectors to generate an asset unit consistency fingerprint vector and writes it into the consistency fingerprint repository. The asset unit consistency fingerprint vector is the output of this step, containing geometric, temporal, and semantic sub-vector fields, along with a comprehensive consistency weight field. The comprehensive consistency weight field records the contribution ratio of each sub-vector and is used as a weighted input in subsequent difference calculations. After this step, the asset unit consistency fingerprint vector is used as the input object for S320. Simultaneously, the geometric and semantic sub-vector fields of this fingerprint vector are also read and processed by the conflict residual field in S420, forming a data consistency closed loop across the main steps.

[0083] S320. Extract cross-source differential fields from the asset unit consistency fingerprint vector, perform differential sorting and conflict evidence pair generation processing, and generate a differential candidate set.

[0084] In this embodiment, the input source for this step is the asset unit consistency fingerprint vector output by S310. The differential calculation is performed by the differential analysis unit on a cloud analysis node, which is connected to the conflict identification service and evidence association repository. Specifically, the differential analysis unit first reads the geometric sub-vector field, temporal sub-vector field, and semantic sub-vector field from the asset unit consistency fingerprint vector, and establishes cross-source comparison groups based on the data source identifier field. Each comparison group contains sub-vector records from at least two sources, and a comparison mapping table is established by matching the asset unit primary key field. The comparison mapping table records the sub-vector indices and weight parameter fields of source A and source B.

[0085] During the cross-source differential field extraction phase, the differential analysis unit calculates the geometric offset field for the geometric sub-vector field, which is generated by the difference between the spatial overlap field and the morphological offset field; it calculates the time drift difference field and the rate of change difference field for the time-series sub-vector field; and it calculates the semantic deviation field and the confidence difference field for the semantic sub-vector field. All differential results are combined to form a cross-source differential field set and normalized according to the weight parameter field. The differential analysis unit writes the differential fields corresponding to the primary key fields of each asset unit into the differential result table, and attaches a source identifier combination field and a differential direction identifier field to each record. If a sub-vector of a certain source is missing or abnormal, the system writes the differential result of the corresponding record into the differential anomaly table and registers the differential anomaly type field and the generation time identifier field in the audit log.

[0086] Differential sorting is performed by the conflict identification service, defined as establishing a differential priority sequence based on weight fields and difference values ​​within the cross-source differential field set. Specifically, the conflict identification service reads the normalized cross-source differential field set and calculates the difference score field according to the overall consistency weight field. The system combines the difference score field with the data source identifier field and writes it into the differential sorting table, generating a sorting index field. The sorting index field is arranged from high to low according to the difference score field, identifying potential conflict areas. Further, the system extracts conflict candidate pairs from records where the difference score field exceeds a preset threshold. Each candidate pair is determined by the source identifier combination field and the asset unit primary key field, generating a conflict candidate index field.

[0087] The generation of conflict evidence pairs is handled by the differential analysis unit, defined as establishing evidence correspondences for high-priority differences in the sorting results. Specifically, the differential analysis unit retrieves the corresponding evidence chain fields from the evidence association repository and generates evidence pairing records based on the fragment unique identifier field and the orchestration session primary key. Each pairing record includes an evidence source identifier, a collection time field, a spatial boundary index, and a set of attribute tags. The differential analysis unit performs a consistency check on the pairing records, generating a conflict level field from the check results and writing it to the conflict evidence table. If the conflict level field exceeds a threshold, the system marks the pair of records as a high-risk conflict evidence pair and writes the identifier to the difference candidate set. The difference candidate set, as the output of this step, includes a cross-source difference field, a difference score field, a conflict level field, and an evidence pairing record index field, and is used by the graph node set generation process of S330.

[0088] S330. Perform graph node set generation, graph edge set generation, and conflict intensity field registration on the candidate set of differences to generate a consistent conflict graph structure.

[0089] In this embodiment, the input source for this step is the difference candidate set output by S320. The conflict graph generation is performed by the conflict modeling unit on a cloud analysis node. The conflict modeling unit is connected to the graph structure generation engine and the weight allocation service. Specifically, the conflict modeling unit first reads the asset unit primary key field, source identifier combination field, and conflict level field from the difference candidate set to construct a graph node set. Each graph node represents an asset unit primary key instance and carries its conflict level field and source quantity field; node attributes are recorded in the node attribute table, including a node identifier field, a node type field, and a node status field.

[0090] The graph edge set generation process is executed by the graph structure generation engine, defined as the set of edges that establish conflict relationships between nodes. Specifically, the graph structure generation engine reads the evidence pairing record index field and the cross-source difference field, and establishes edge connections when the source identifier combination field has an intersection. Each edge records a conflict type field, a difference direction identifier field, and a time weight field. The time weight field is calculated from the difference between the conflict occurrence time and the collection time field, and is used to characterize the freshness of the conflict. The engine further calculates the edge strength field, generates an initial conflict strength value based on the conflict level field and the difference score field, and performs edge weight normalization operations under the rules provided by the weight allocation service.

[0091] The conflict intensity field registration process is handled by the conflict modeling unit, which is defined as uniformly registering the intensity results in the graph node set and edge set and generating a conflict intensity field. Specifically, the conflict modeling unit aggregates the initial conflict intensity value field of nodes according to the asset unit primary key field, calculates the comprehensive conflict intensity field of nodes, and combines the normalized results of the edge weights to form the final conflict intensity field. The system writes the conflict intensity field into the conflict graph record table, and simultaneously registers the generation time identifier and source batch identifier fields of the conflict intensity field in the audit log. If the conflict intensity field is lower than the threshold, the system marks the corresponding node as "low conflict" in the node attribute table so that the subsequent S410 fusion constraint rule set loading process can reduce constraints. After completing the registration of the node and edge sets, the conflict modeling unit outputs a consistent conflict graph structure and writes it into the conflict graph repository.

[0092] The consistency conflict graph structure, as the output of this step, includes a set of graph nodes, a set of graph edges, a conflict intensity field, and a time weight field. The set of graph nodes and the set of graph edges form an integrated graph topology, providing input for the subsequent loading of the fusion constraint rule set in S410. The generation of this structure also supports backtracking across main steps. The evidence chain field from the preceding S230 and the conflict intensity field from this step are linked through the asset unit primary key field, forming a continuously updated, consistent clearance loop.

[0093] Step S400 includes at least steps S410-S430:

[0094] S410. Obtain the consistency conflict graph structure, perform fusion constraint rule set loading, evidence priority sequence generation, and fusion session primary key registration to obtain the fusion solution configuration structure.

[0095] In this embodiment, the input source for this step is the consistency conflict graph structure output by S330. The consistency conflict graph structure includes a set of graph nodes, a set of graph edges, a conflict intensity field, and a time weight field, and is bound to the asset unit primary key field. This step is run by the fusion orchestration unit on the cloud fusion computing node. The fusion orchestration unit is connected to the rule warehouse service, evidence index service, and session registration service to complete constraint loading, evidence sorting, and session primary key solidification before fusion solving. Specifically, the fusion orchestration unit reads the set of graph nodes and the set of graph edges in the consistency conflict graph structure, establishes a fusion candidate node index according to the asset unit primary key field, and extracts the conflict type field, differential direction identifier field, and time weight field from the graph edge set to form a fusion candidate edge index. The fusion candidate node index and the fusion candidate edge index serve as the context carrier for subsequent constraint loading and generation of evidence priority sequences. The relevant index records are written to the fusion orchestration log, which includes the orchestration session primary key, asset unit primary key field, node index summary field, and edge index summary field.

[0096] The loading process for the fusion constraint rule set is provided by the rule repository service and executed by the fusion orchestration unit. The fusion constraint rule set is defined as a set of rule entries adapted to the consistency conflict graph structure. This set of rule entries consists of several rule entries, each containing at least a primary key field, a rule version identifier field, an applicable conflict type field, an applicable data source identifier field set, a constraint field path, a constraint decision criterion field, a constraint priority field, and a constraint revocability flag field. The constraint field path points to the evidence field location associated with a node attribute field or edge attribute field in the consistency conflict graph structure. The constraint decision criterion field is composed of a criterion field, a boundary version identifier field, a rule version identifier field, a capability field version identifier field, and a terminology version identifier field, used to limit criterion consistency during the constraint execution phase. The fusion orchestration unit reads the set of rule entries whose rule version identifier field matches the primary key of the current orchestration session from the rule repository service, and filters out the constraint subsets corresponding to the conflict type field in the graph edge set according to the applicable conflict type field. When the conflict intensity field of the consistent conflict graph structure is in different ranges, the fusion orchestration unit adjusts the loading order of the constraint subsets according to the constraint priority field and writes the adjustment record into the constraint loading record table, which includes the rule entry primary key field, loading order field, and loading time identifier field. For rule entries whose constraint revocability flag field is in a revocable state, the fusion orchestration unit registers them as soft constraint entry fields; for rule entries whose constraint revocability flag field is in an irrevocable state, the fusion orchestration unit registers them as hard constraint entry fields. The hard constraint entry fields and soft constraint entry fields are jointly written into the fusion constraint rule set structure and form the fusion constraint rule set field in the fusion solver configuration structure. The fusion constraint rule set field serves as one of the input sources for the constraint satisfaction check processing in subsequent S420, and together with the graph node set and graph edge set read by S420, constitutes an executable constraint input pair.

[0097] The evidence priority sequence generation process is supported by the evidence index service and executed by the fusion orchestration unit. The evidence priority sequence is defined as the sorting result structure of multi-source evidence package records under the same asset unit's primary key field. The evidence package records are derived from the mapping index of the multi-source fragmented evidence package set structure output by S230 in the consistency conflict graph structure. Each evidence package record contains at least a fragment unique identifier field, a quality field, an evidence chain field, a timestamp field, a data source identifier field, and a source type identifier field. For each graph node set element in the consistency conflict graph structure, the fusion orchestration unit initiates an evidence retrieval request to the evidence index service. The evidence retrieval request carries the asset unit primary key field, the data source identifier field set, and the time weight field. The evidence index service returns an evidence list, where each evidence record contains a fragment unique identifier field and a quality field. The fusion orchestration unit extracts the integrity marker field, caliber consistency marker field, and time window marker field from the quality field, and the source identifier field, acquisition method identifier field, submission time identifier field, boundary version identifier field, rule version identifier field, and capability field version identifier field from the evidence chain field. The fusion orchestration unit combines these fields to generate the evidence scoring field. The construction of the evidence scoring field adopts a segmented mapping method. The segmented mapping rules are defined by the evidence scoring rule table, which at least includes the quality mapping rule field, the time weight mapping rule field, and the source credibility mapping rule field, and is associated with the rule version identifier field. The fusion orchestration unit sorts the evidence list according to the evidence scoring field to generate an evidence priority sequence. The evidence priority sequence includes at least the evidence sorting index field, the fragment unique identifier field, the data source identifier field, and the evidence scoring field, and writes this sequence into the evidence priority sequence table. The evidence priority sequence table is bound to the asset unit's primary key field, and its index is written into the evidence priority sequence field of the fusion solver configuration structure, so that S420 can read the evidence priority sequence field and select evidence input according to the sorting index field when generating the fusion result field. For evidence records whose evidence scoring field is in the low confidence range, the fusion orchestration unit writes the evidence downgrade marker field and records the downgrade reason summary field. The downgrade reason summary field is bound to the fragment unique identifier field and written to the audit log. The subsequent S430 exception event generation process will read the downgrade marker field and use it to generate review-type exception events.

[0098] The fusion session primary key registration process is supported by the session registration service and executed by the fusion orchestration unit. The fusion session primary key is defined as a unique identifier for the session at the fusion solver level, used for auditing the fusion link from S410 to S430, and establishes a traceable association with the primary key of the preceding orchestration session. The fusion orchestration unit reads the conflict graph batch identifier field, rule version identifier field, boundary version identifier field, and evidence priority sequence field index from the consistency conflict graph structure to generate a session registration input record. This record contains at least the orchestration session primary key, conflict graph batch identifier field, rule version identifier field, boundary version identifier field, evidence sequence batch identifier field, and registration time identifier field. The session registration service performs uniqueness verification on the session registration input record and returns the fusion session primary key. The fusion orchestration unit writes the fusion session primary key into the fusion session primary key registration table and registers the fusion session primary key, registration time identifier field, and associated orchestration session primary key in the audit log. Subsequently, the fusion orchestration unit encapsulates the fusion session primary key with the fusion constraint rule set field, evidence priority sequence field, graph node set index field, and graph edge set index field to obtain the fusion solver configuration structure and writes it into the fusion configuration repository. The fusion solution configuration structure is the output product of this step, and its output field is named "fusion solution configuration structure". After this step is completed, it is called as the input object of S420. S420 extracts the graph node set and graph edge set from the fusion solution configuration structure and performs constraint satisfaction check processing and conflict residual field calculation processing. At the same time, the fusion session primary key in the fusion solution configuration structure will be referenced by the result version chain registration processing of S430, forming a session-level interconnected association with the result version chain structure.

[0099] S420. Extract the graph node set and graph edge set from the fusion solution configuration structure, perform constraint satisfaction check processing, conflict residual field calculation processing, and fusion result field generation processing to generate a consistent fusion result set.

[0100] In this embodiment, the input source for this step is the fusion solution configuration structure output by S410. The fusion solution configuration structure includes at least a fusion session primary key, a fusion constraint rule set field, an evidence priority sequence field, a graph node set index field, and a graph edge set index field. This step is run by the fusion solution unit on a cloud-based fusion computing node. The fusion solution unit is connected to the constraint execution engine, the evidence reading service, and the residual calculation service, and is used to perform constraint checks, residual calculations, and output fusion result fields on the conflict graph. Specifically, the fusion solution unit extracts the graph node set index field and the graph edge set index field from the fusion solution configuration structure, and obtains the graph node set and graph edge set through index parsing. Simultaneously, the fusion solution unit reads the fusion constraint rule set field and constructs a constraint execution context locally. The constraint execution context includes at least a set of rule entry primary key fields, a set of hard constraint entry fields, a set of soft constraint entry fields, and a constraint determination caliber field. The fusion solving unit also reads the evidence sorting index field and fragment unique identifier field from the evidence priority sequence field, and sends a batch read request to the evidence reading service. The batch read request carries the fragment unique identifier field set, the asset unit primary key field set and the caliber field. The evidence reading service returns evidence data fragments with quality field and evidence chain field. The fusion solving unit writes the returned results into the solving input buffer. The solving input buffer is bound to the fusion session primary key and the cache write time identifier field is recorded in the audit log.

[0101] The constraint satisfaction check is supported by the constraint execution engine and initiated by the fusion solution unit. The constraint satisfaction check is defined as a process of verifying each rule entry of the candidate evidence combination corresponding to the primary key field of each asset unit. The verification objects include node attribute fields, edge attribute fields, and their associated evidence field paths. Specifically, the fusion solution unit generates candidate evidence combinations for each graph node set element. The generation of candidate evidence combinations is based on a progressive selection according to the evidence sorting index field. This progressive selection process is recorded in the candidate combination generation record table, which at least includes the fusion session primary key, asset unit primary key field, candidate combination number field, fragment unique identifier field sequence, and generation time identifier field. The constraint execution engine performs hard constraint entry field verification on the candidate evidence combination and outputs a hard constraint satisfaction flag field. This hard constraint satisfaction flag field serves as a feasibility gate field during the solution process. When the hard constraint satisfaction flag field is not satisfied, the fusion solution unit switches candidate evidence combinations and registers the failure reason summary field, the failure rule entry primary key field, and the candidate combination number field in the constraint failure record table. For candidate evidence combinations where the hard constraint satisfaction flag field is satisfied, the constraint execution engine continues to check the soft constraint entry fields and outputs the soft constraint deviation flag field and deviation index field. The fusion solution unit writes the deviation index field into the deviation details record table and binds it to the candidate combination number field, enabling subsequent conflict residual field calculation and processing to read the deviation index field and locate the source of the residual. The execution process of constraint satisfaction check has automatic triggering conditions under the primary key field of each asset unit. The triggering conditions are composed of a conflict intensity field, a time weight field, and an evidence score field. When the conflict intensity field is in the high range or the time weight field indicates a recent sampling record, the fusion solution unit increases the upper limit field of the candidate evidence combination's check count and writes the increase record into the solution strategy record table. When the conflict intensity field is in the low range, the fusion solution unit decreases the upper limit field of the check count and writes the decrease record into the solution strategy record table. The upper limit field of the check count, the candidate combination number field, and the failure reason summary field are all written to the audit log, thereby supporting the auditable traceability of the subsequent result version chain structure.

[0102] The conflict residual field calculation process is supported by the residual calculation service and initiated by the fusion solution unit. A conflict residual field is defined as a residual measurement record for a set of fields that still deviate after constraint satisfaction checks. The residual measurement record includes at least a residual type field, a residual source field path, a residual magnitude level field, an evidence source identifier field, and a time identifier field. Specifically, the fusion solution unit reads the deviation item index field and locates the corresponding rule entry primary key field and constraint field path. The residual calculation service performs difference calculations on the relevant fields of the candidate evidence combination under the same constraint determination caliber field, forming a residual detail field. The residual detail fields are grouped by the residual type field and written into the residual detail table, which is bound to the asset unit primary key field and the fusion session primary key. For geometric residuals, the residual source field path points to the geometric contour set or node geometric attribute field, and the residual magnitude level field is obtained through segmented threshold mapping and written into the geometric residual field. For temporal residuals, the residual source field path points to the timestamp field or version identifier field, and the residual magnitude level field is written into the temporal residual field. For semantic residuals, the residual source field path points to the semantic label field or attribute consistency field, and the residual magnitude level field is written into the semantic residual field. The fusion solution unit combines the geometric residual field, temporal residual field, and semantic residual field to generate conflicting residual fields, and writes the conflicting residual fields into the solution output record table. The solution output record table also records the hard constraint satisfaction mark field, the soft constraint deviation mark field, and the candidate combination number field, which facilitates the subsequent S430 anomaly event generation and processing to generate anomaly events according to the conflicting residual fields.

[0103] The generation of fusion result fields is completed by the fusion solving unit and driven by the constraint execution results. Fusion result fields are defined as fusion output records for asset units in terms of spatial boundaries, attribute labels, and time versions. Each fusion output record includes at least a fusion boundary field, a fusion attribute field, a fusion time identifier field, a fusion source evidence index field, and a fusion confidence flag field. Specifically, the fusion solving unit selects the primary evidence record from candidate evidence combinations where the hard constraint satisfaction flag field is satisfied, based on the evidence sorting index field. It also selects secondary evidence records based on the degree of deviation from the soft constraint deviation flag field. The fragment unique identifier fields of the primary and secondary evidence records are written into the fusion source evidence index field. The fusion solving unit generates fusion boundary fields using a field-level assembly method. This assembly process is recorded in the assembly record table, which records the assembly field path, the unique identifier field of the source evidence fragment, and the assembly time identifier field. The fusion solving unit generates fusion attribute fields using attribute mapping rules provided by the terminology dictionary entry set, performing a unified mapping and writing the mapping process to the attribute mapping log. The fusion solving unit generates the fusion time identifier field by aligning the submission time identifier field and the historical version identifier field in the evidence chain field, and the version alignment result is written to the version alignment record table. The fusion confidence flag field is obtained by mapping the evidence score field and the conflict residual field. The fusion solving unit writes this flag to the fusion output record and binds it to the asset unit's primary key field.

[0104] After completing constraint satisfaction checks, conflict residual field calculations, and fusion result field generation, the fusion solution unit aggregates the fusion output records of the primary key fields of all asset units under the same fusion session primary key, generates a consistent fusion result set, and writes it to the fusion result repository. The consistent fusion result set, as the output product of this step, has the output field name "Consistent Fusion Result Set," and is called as the input object of S430 after this step. S430 performs anomaly event generation processing, disposal strategy matching processing, and result version chain registration processing on the consistent fusion result set. Simultaneously, the conflict residual field and fusion source evidence index field in the consistent fusion result set will also be referenced by the result version chain registration processing, thereby establishing a traceable relationship between the fusion output and the evidence chain field and fragment unique identifier field.

[0105] S430. Perform abnormal event generation processing, handling strategy matching processing, and result version chain registration processing on the consistency fusion result set, generate the abnormal event set, handling strategy set, and result version chain structure, and generate the collection session configuration structure.

[0106] In this embodiment, the input source for this step is the consistent fusion result set output by the aforementioned S420. The consistent fusion result set includes at least the asset unit primary key field, fusion boundary field, fusion attribute field, fusion time identifier field, fusion source evidence index field, conflict residual field, and fusion confidence flag field. This step is run by the closed-loop processing unit on the cloud processing node. The closed-loop processing unit is connected to the event generation service, policy library service, version chain registration service, and collection orchestration service. It is used to convert the fusion output into anomaly events, generate processing strategies, complete the result version chain registration, and simultaneously write back the collection session configuration structure to enter the next round of collection. Specifically, the closed-loop processing unit reads the consistent fusion result set and establishes a processing index according to the asset unit primary key field. The processing index includes the fusion time identifier field and the fusion source evidence index field. The closed-loop processing unit writes the processing index into the processing session log and binds it to the fusion session primary key. The processing session log is used to record the process trajectory of anomaly generation, strategy matching, and version registration.

[0107] The abnormal event generation and processing is supported by an event generation service and initiated and executed by a closed-loop processing unit. An abnormal event is defined as a structured description of an event record that triggers review, supplementary collection, rule revision, or manual verification in the fusion results. The event record includes at least the following fields: abnormal event primary key field, abnormal type field, abnormal level field, abnormal location field, associated asset unit primary key field, set of associated fragment unique identifier fields, conflict residual field summary, trigger condition summary field, and generation time identifier field. Specifically, the closed-loop processing unit extracts the geometric residual field, temporal residual field, and semantic residual field from the conflict residual field, and performs segmented mapping in conjunction with the fusion confidence marker field to obtain the abnormal level field. The closed-loop processing unit also parses the primary evidence fragment unique identifier field and the secondary evidence fragment unique identifier field from the fusion source evidence index field to form a set of associated fragment unique identifier fields. The anomaly location field is generated by combining the fusion boundary field and the residual source field path. This field contains both spatial location fragments and field location fragments. Spatial location fragments can be directly read by the Geographic Information System (GIS) rendering engine, while field location fragments can be directly read by the audit playback module. The event generation service encapsulates these fields and generates an anomaly event primary key field. The closed-loop processing unit writes this primary key field into the anomaly event investigation set. This anomaly event investigation set is one of the output products of this step, with the output field named "Anomaly Event Investigation Set," and is associated with the subsequent processing strategy set and result version chain structure. The anomaly event generation process has automated triggering conditions, which are composed of the residual magnitude level field, fusion confidence flag field, and evidence downgrade flag field of the conflict residual field. When the evidence downgrade flag field is in a downgraded state or the residual magnitude level field is in a high range, the event generation service registers the anomaly event type field as a review or supplementary acquisition type and writes the triggering condition summary field into the audit log, enabling the subsequent write-back of the acquisition session configuration structure to directly reference the triggering condition summary field to generate supplementary acquisition task parameters.

[0108] The handling strategy matching process is supported by the strategy library service and initiated by the closed-loop handling unit. A handling strategy is defined as a set of actions to be performed on an abnormal event. This set of actions includes at least a strategy primary key field, a strategy type field, an action parameter field, an execution entity identifier field, an execution time window field, and a set of write-back fields. The strategy library service maintains a handling strategy rule table, which includes at least the mapping rules from the abnormality type field to the strategy type field, the mapping rules from the abnormality level field to the action parameter field, the mapping rules from the source identifier field to the execution entity identifier field, and the generation rules for the execution time window field. The closed-loop handling unit writes the abnormality type field, abnormality level field, abnormality location field, associated asset unit primary key field, and associated shard unique identifier field set from the abnormal event set into a strategy matching request record and initiates a matching request to the strategy library service. The strategy library service returns a list of strategy candidates, each containing a strategy type field and an action parameter field. The closed-loop processing unit performs executability checks on the strategy candidate list. The executability check reads the status of the acquisition method identifier field and the authorization credential index field in the evidence chain field, and also reads the status of the sampling time window field in the acquisition session configuration structure. When the executability check is in an unexecutable state, the closed-loop processing unit marks the strategy candidate record as suspended and writes it to the strategy suspension record table. The strategy suspension record table is bound to the exception event primary key field and written to the audit log. For strategy candidate records that are in an executable state after the executability check, the closed-loop processing unit registers the strategy primary key field and generates a processing strategy set. The processing strategy set is one of the output products of this step, and its output field is named "Processing Strategy Set". The action parameter field in the processing strategy set contains parameter fragments written back to the acquisition orchestration service. The parameter fragments at least include the asset unit primary key field, the sampling space range field, the sampling time window field, and the sampling layer identifier field, so that supplementary acquisition tasks or review tasks can be directly assembled when generating the acquisition session configuration structure later. The handling strategy matching process also supports access to the manual verification link. When the execution subject identifier field indicates the manual review queue, the closed-loop handling unit writes the anomaly location field and the fusion source evidence index field into the manual review task form, and writes the primary key field of the manual review task form into the write-back field set of the handling strategy set, so that the result version chain registration process can record the version nodes of the manual review input and output.

[0109] The result version chain registration process is supported by the version chain registration service and initiated by the closed-loop processing unit. The result version chain structure is defined as a chain-like record structure formed by serially registering the boundary version identifier field, rule version identifier field, capability field version identifier field, terminology version identifier field, conflict graph batch identifier field, fusion session primary key, fusion source evidence index field, and abnormal event primary key field during the cleanup process. The chain-like record structure includes at least a version node set, a version edge set, and a version audit field. The closed-loop processing unit extracts the fusion time identifier field and fusion source evidence index field from the consistency fusion result set, the abnormal event primary key field from the cleanup abnormal event set, the strategy primary key field and write-back field set from the processing strategy set, and extracts the fusion session primary key, rule version identifier field, and boundary version identifier field from the fusion solution configuration structure associated record to generate a version registration input package. The version chain registration service performs node deduplication verification on the version registration input package and generates a version node set and a version edge set. The closed-loop processing unit writes these into the result version chain structure and registers the version audit field, which includes at least a registration time identifier field, a registration subject identifier field, and an audit log index field. The result version chain structure is one of the output products of this step. Its output field is named "result version chain structure", and it is read by the subsequent display and analysis platform for result playback and audit verification.

[0110] After generating the anomaly event set, handling strategy set, and result version chain structure, the closed-loop handling unit also performs the acquisition session configuration structure generation process, assembling the write-back field set in the handling strategy set as the input for the next round of acquisition orchestration. The acquisition session configuration structure, as a write-back product in this step, includes the orchestration session primary key, asset unit primary key field set, sampling space range field, sampling time window field, three-layer acquisition orchestration structure field, and sampling evidence chain field. The asset unit primary key field set is obtained by aggregating the associated asset unit primary key fields from the anomaly event set; the sampling space range field is mapped from the spatial location fragments of the anomaly location field; the sampling time window field is generated by combining the fusion time identifier field and the execution time window field; and the three-layer acquisition orchestration structure field is mapped from the strategy type field and includes task parameters for the baseline anchor layer, verification sampling layer, and change capture layer. The closed-loop processing unit writes the generated acquisition session configuration structure into the acquisition session configuration repository and registers the write-back time identifier field, trigger condition summary field, and associated fusion session primary key in the audit log. This enables the next round of S210 to read the latest orchestration session primary key and task parameters under the same audit link when obtaining the acquisition session configuration structure. After the acquisition session configuration structure is generated at the end of this step, it is called as the input object for subsequent S210, thus forming a closed-loop link from S210 to S430 and back to S210. As an engineering example, in the continuous operation scenario of county-level inventory, the graph node corresponding to the primary key field of a certain asset unit shows that the geometric residual field is in the high range and the evidence degradation marker field is in the degradation state in the fusion result. The event generation service generates a supplementary collection anomaly event and writes it into the inventory anomaly event set; the strategy library service returns supplementary collection strategy candidate records. After the closed-loop processing unit passes the executability verification, it registers the processing strategy set and sets the sampling time window field to the sampling window of the next working day. At the same time, it writes the encrypted sampling task parameters of the change capture layer into the three-layer acquisition orchestration structure field; the version chain registration service registers the fusion session primary key, the anomaly event primary key field, the strategy primary key field and the fragment unique identifier field to form the result version chain structure; after receiving the written-back acquisition session configuration structure, the acquisition orchestration service triggers the acquisition and processing of remote sensing image data and lidar point cloud data of S210 in the next sampling window, thereby completing the anomaly-driven automated back-collection closed loop.

[0111] Example 2: Figure 2 This diagram illustrates a structural block diagram of a land resource asset inventory system based on multi-source data analysis according to an embodiment of the present invention. Figure 2 As shown, the structure may include:

[0112] The asset unit primary key table generation unit 01 is used to obtain the target domain boundary data, land asset object list, multi-source data source list, and asset unit division rules, perform terminology dictionary entry set registration and asset unit segmentation processing, generate an asset unit primary key table, and output the asset unit primary key table to the anchor point candidate list generation unit. Specifically, after receiving the target domain boundary data, land asset object list, multi-source data source list, and asset unit division rules from external system input, the asset unit primary key table generation unit performs boundary closure verification and boundary element missing verification on the target domain boundary data, object field integrity verification and object field duplication verification on the land asset object list, data source entry validity verification and data source entry duplication verification on the multi-source data source list, and rule field integrity verification and rule conflict verification on the asset unit division rules. During the verification process, the identified missing fields, duplicate fields, invalid entries, and other states are written into the registration records involved in the terminology dictionary entry set registration. The terminology dictionary entry set registration is used to uniformly write object names, field names, scope names, and identifier names from the target domain boundary data, land asset object list, multi-source data source list, and asset unit division rules into the terminology dictionary entry set, and uses the terminology dictionary entry set as input constraints for asset unit segmentation processing. The asset unit segmentation processing divides the target domain boundary data into asset units according to the asset unit division rules, and establishes a binding record between each asset unit and the asset object entries in the land asset object list. Simultaneously, an asset unit primary key is generated for each asset unit, forming an asset unit primary key table. The asset unit primary key table records the association between the asset unit primary key and the corresponding target domain boundary data range record, land asset object list entry record, and asset unit division rule version record. After the asset unit primary key table is generated, the asset unit primary key table generation unit provides the asset unit primary key table to the anchor point candidate list generation unit for use when extracting the asset unit primary key sequence. Simultaneously, the registration record formed by the terminology dictionary entry set registration is retained for the subsequent acquisition session configuration structure generation unit to use consistent terminology entries when generating the acquisition session configuration structure.

[0113] Anchor point candidate list generation unit 02 is used to extract the asset unit primary key sequence from the asset unit primary key table, perform multi-source data source capability field registration and anchor point candidate matching processing, generate an anchor point candidate list, and output the anchor point candidate list to the acquisition session configuration structure generation unit; specifically, after receiving the asset unit primary key table output from the asset unit primary key table generation unit, the anchor point candidate list generation unit reads the asset unit primary key from the asset unit primary key table and forms the asset unit primary key sequence, and at the same time reads the records of the clearing target domain boundary data range and the land asset object list entries registered in the asset unit primary key table, and establishes a table association between the records and the asset unit primary key. The multi-source data source capability field registration is performed by the anchor point candidate list generation unit, which reads and registers the available caliber fields of each data source entry in the multi-source data source list. These available caliber fields include at least coordinate projection caliber-related fields and timestamp caliber-related fields. The unit also registers the data source entry type fields corresponding to remote sensing image data and lidar point cloud data acquisition and processing, ground survey data acquisition and processing, and historical inventory data loading and processing. During registration, the anchor point candidate list generation unit performs coverage relationship verification between the multi-source data source list entries and the boundary data range records of the inventory target domain corresponding to the asset unit primary key sequence, and writes the verification results into the capability field record formed by the multi-source data source capability field registration. The anchor point candidate matching process involves the anchor point candidate list generation unit iterating through each asset unit primary key sequence and calling the capability field record to select data source entries that pass the coverage relationship verification, generating a set of anchor point candidate entries bound to each asset unit primary key, forming the anchor point candidate list. The anchor point candidate list registers the asset unit primary key, the matched data source entry record, and the corresponding coordinate projection caliber-related fields and timestamp caliber-related fields. After the anchor point candidate list is generated, the anchor point candidate list generation unit provides the anchor point candidate list to the acquisition session configuration structure generation unit, so that the acquisition session configuration structure generation unit can read it when performing baseline anchor point layer registration processing, verification sampling layer task orchestration processing, and change capture layer encrypted sampling task orchestration processing.

[0114] The acquisition session configuration structure generation unit 03 is used to perform baseline anchor layer registration processing, verification sampling layer task orchestration processing, and change capture layer encrypted sampling task orchestration processing on the anchor candidate list, generate an acquisition session configuration structure, and output the acquisition session configuration structure to the multi-source data packet generation unit. Specifically, after receiving the anchor candidate list output by the anchor candidate list generation unit, the acquisition session configuration structure generation unit reads the registered asset unit primary key, data source entry record, coordinate projection caliber related fields, and timestamp caliber related fields in the anchor candidate list, and performs caliber parallel registration for multiple data source entries under the same asset unit primary key, writing the caliber differences into the caliber field mapping record included in the acquisition session configuration structure. The baseline anchor layer registration processing is performed by the acquisition session configuration structure generation unit selecting the baseline anchor layer entry bound to the asset unit primary key in the anchor candidate list and registering the entry as a baseline anchor layer registration record; the baseline anchor layer registration record at least includes the asset unit primary key, the selected data source entry record, coordinate projection caliber related fields, and timestamp caliber related fields. The verification sampling layer task orchestration process is generated by the acquisition session configuration structure generation unit based on the combination relationship of multiple data source entries under the same asset unit primary key in the anchor candidate list. The task orchestration record registers the asset unit primary key, sampling object field, sampling caliber field, and sampling window field, and associates the sampling caliber field with the aforementioned caliber field mapping record. The change capture layer encrypted sampling task orchestration process is generated by the acquisition session configuration structure generation unit for data source entries marked as change capture layer in the anchor candidate list. The encrypted sampling task orchestration record registers the asset unit primary key, sampling object field, sampling caliber field, sampling window field, and encrypted sampling field, and associates the encrypted sampling field with the data source entry record. The acquisition session configuration structure consists of a baseline anchor point layer registration record, a verification sampling layer task arrangement record, a change capture layer encrypted sampling task arrangement record, and a caliber field mapping record. The acquisition session configuration structure generation unit provides the acquisition session configuration structure to the multi-source data packet generation unit, so that the multi-source data packet generation unit can trigger the corresponding acquisition task according to the acquisition session configuration structure when acquiring and processing remote sensing image data and lidar point cloud data, acquiring and processing ground survey data, and loading and processing historical inventory data.

[0115] The multi-source data packet generation unit 04 is used to receive the acquisition session configuration structure, perform remote sensing image data and lidar point cloud data acquisition and processing, ground survey data acquisition and processing, and historical inventory data loading and processing to obtain multi-source data packets, and output the multi-source data packets to the multi-source fragmented evidence packet set structure generation unit. Specifically, after receiving the acquisition session configuration structure output from the acquisition session configuration structure generation unit, the multi-source data packet generation unit reads the baseline anchor layer registration record, the verification sampling layer task arrangement record, and the change capture layer encrypted sampling task arrangement record registered in the acquisition session configuration structure, and initiates remote sensing image data and lidar point cloud data acquisition and processing according to the asset unit primary key and data source entry record in the record. The acquisition and processing includes loading data fragments within the corresponding time range according to the sampling window field and loading data entries with the corresponding caliber according to the sampling caliber field. At the same time, the coordinate projection caliber related fields and timestamp caliber related fields in the acquisition session configuration structure are written into the same entry record along with the acquired data. The ground survey data acquisition and processing is handled by the multi-source data packet generation unit, which calls the corresponding data source entries according to the verification sampling layer task arrangement record. The ground survey data is bound to the asset unit primary key and written into the multi-source data packet. Simultaneously, the timestamp caliber-related fields of the ground survey data are associated with the caliber field mapping record in the acquisition session configuration structure. The historical inventory data loading process is handled by the multi-source data packet generation unit, which loads historical inventory data according to the baseline anchor point layer registration record and binds the historical inventory data to the asset unit primary key. The coordinate projection caliber-related fields and timestamp caliber-related fields of the historical inventory data entries are written into the multi-source data packet. The multi-source data packet consists of remote sensing image data and lidar point cloud data entries, ground survey data entries, and historical inventory data entries. The multi-source data packet generation unit provides the multi-source data packet to the multi-source fragmented evidence package set structure generation unit, which then extracts the asset unit primary key, binds the fragments, and performs caliber unification processing.

[0116] The multi-source fragmented evidence package structure generation unit 05 is used to extract asset unit primary key binding fragments from the multi-source data package and perform coordinate projection caliber unification processing and timestamp caliber unification processing to generate a preprocessed fragment set. It then performs quality self-inspection field generation processing, evidence chain field registration processing, and fragment unique identifier registration processing on the preprocessed fragment set to generate a multi-source fragmented evidence package structure. This structure is then output to the consistency conflict graph structure generation unit. Specifically, after receiving the multi-source data package output from the multi-source data package generation unit, the multi-source fragmented evidence package structure generation unit traverses the remote sensing image data and lidar point cloud data entries, ground survey data entries, and historical inventory data entries within the multi-source data package according to the asset unit primary key. It then extracts fragments bound to the same asset unit primary key from each entry to form a fragment sequence to be processed. The unified processing of coordinate projection caliber involves the multi-source fragmented evidence package structure generation unit reading the coordinate projection caliber-related fields carried by the primary key bound fragments of each asset unit, and calling the caliber field mapping record registered in the acquisition session configuration structure to complete the unified writing of coordinate projection caliber; the unified processing of timestamp caliber involves the multi-source fragmented evidence package structure generation unit reading the timestamp caliber-related fields carried by the primary key bound fragments of each asset unit, and calling the timestamp field mapping record to complete the unified writing of timestamp caliber, thereby generating a preprocessed fragment set. The quality self-inspection field generation process involves the multi-source fragmented evidence package structure generation unit performing field integrity checks, caliber consistency checks, and entry readability checks on the preprocessed fragmented set, and writing the check results into the quality self-inspection field. The evidence chain field registration process involves the multi-source fragmented evidence package structure generation unit registering data source entry records, acquisition session configuration structure record reference relationships, and asset unit primary key binding relationships on the preprocessed fragmented set, and writing the registration results into the evidence chain field. The fragment unique identifier registration process involves the multi-source fragmented evidence package structure generation unit generating a fragment unique identifier for the preprocessed fragmented set based on the asset unit primary key, data source entry records, and sampling window field, and writing it into the fragment unique identifier registration record. The multi-source fragmented evidence package structure consists of a preprocessed fragmented set carrying the quality self-inspection field, evidence chain field, and fragment unique identifier. The multi-source fragmented evidence package structure generation unit provides the multi-source fragmented evidence package structure to the consistency conflict graph structure generation unit, allowing the consistency conflict graph structure generation unit to construct the asset unit consistency fingerprint vector and generate the consistency conflict graph structure.

[0117] The consistency conflict graph structure generation unit 06 is used to receive the multi-source fragmented evidence package set structure and perform geometric consistency sub-vector construction processing, temporal consistency sub-vector construction processing, and semantic consistency sub-vector construction processing to obtain the asset unit consistency fingerprint vector. It also extracts cross-source difference fields from the asset unit consistency fingerprint vector and performs difference sorting processing and conflict evidence pair generation processing to generate a difference candidate set. Furthermore, it performs graph node set generation processing, graph edge set generation processing, and conflict intensity field registration processing on the difference candidate set to generate a consistency conflict graph structure, and outputs the consistency conflict graph structure to the consistency fusion result set generation unit. Specifically, after receiving the multi-source fragmented evidence package set structure output from the multi-source fragmented evidence package set structure generation unit, the consistency conflict graph structure generation unit reads the preprocessed fragment set carrying the fragment unique identifier, evidence chain field, and quality self-check field according to the asset unit primary key. Fragments whose quality self-check field is marked as abnormal are written into the pending verification record and do not participate in this round of sub-vector construction. The geometric consistency sub-vector construction process involves the consistency conflict graph structure generation unit reading geometric element fields from the preprocessed fragment set under the primary key of the same asset unit and performing geometric element alignment, geometric element differencing, and geometric element aggregation to form a geometric consistency sub-vector. The temporal consistency sub-vector construction process involves the consistency conflict graph structure generation unit reading timestamp-related fields from the preprocessed fragment set under the primary key of the same asset unit and performing time order rearrangement, time window alignment, and time difference aggregation to form a temporal consistency sub-vector. The semantic consistency sub-vector construction process involves the consistency conflict graph structure generation unit reading the corresponding semantic element fields from the land asset object list entries from the preprocessed fragment set under the primary key of the same asset unit and performing semantic element merging, semantic element conflict identification, and semantic element difference aggregation to form a semantic consistency sub-vector. The asset unit consistency fingerprint vector is composed of the geometric consistency sub-vector, temporal consistency sub-vector, and semantic consistency sub-vector, and is bound to the asset unit primary key. The cross-source differential field is generated by the consistency conflict graph structure generation unit extracting the differential position field, differential direction field, and differential source fragment unique identifier field from the asset unit consistency fingerprint vector. Then, differential sorting processing is performed on the cross-source differential field to generate sorting records. At the same time, evidence pair records required for conflict evidence pair generation processing are established based on the differential source fragment unique identifier field and the evidence chain field, thereby generating a differential candidate set.The graph node set generation process involves the consistency conflict graph structure generation unit writing the asset unit primary key, shard unique identifier, and difference field entries from the difference candidate set into the graph node set. The graph edge set generation process involves the consistency conflict graph structure generation unit writing the association between conflict evidence pairs and difference field entries into the graph edge set. The conflict intensity field registration process involves the consistency conflict graph structure generation unit writing the sorted records formed by the difference sorting process into the conflict intensity field and associating them with the corresponding graph edge set entries, ultimately generating the consistency conflict graph structure. This consistency conflict graph structure is provided by the consistency conflict graph structure generation unit to the consistency fusion result set generation unit, allowing the consistency fusion result set generation unit to load the fusion constraint rule set and generate the fusion solution configuration structure.

[0118] The consistency fusion result set generation unit 07 is used to receive the consistency conflict graph structure and perform fusion constraint rule set loading processing, evidence priority sequence generation processing, and fusion session primary key registration processing to obtain a fusion solution configuration structure. It also extracts the graph node set and graph edge set from the fusion solution configuration structure and performs constraint satisfaction checking processing, conflict residual field calculation processing, and fusion result field generation processing to generate a consistency fusion result set. The consistency fusion result set is then output to the closed-loop handling and result version chain generation unit. Specifically, after receiving the consistency conflict graph structure output by the consistency conflict graph structure generation unit, the consistency fusion result set generation unit reads the graph node set, graph edge set, and conflict intensity field from the consistency conflict graph structure, and records the rule entries corresponding to the fusion constraint rule set loading processing according to the order of the conflict intensity field. The fusion constraint rule set loading process involves the consistent fusion result set generation unit establishing association records between rule entries and the graph node set and graph edge set, and writing the association records into the fusion solution configuration structure. The evidence priority sequence generation process involves the consistent fusion result set generation unit generating an evidence priority sequence based on the conflict intensity field, evidence chain field, and quality self-check field, and writing the evidence priority sequence into the fusion solution configuration structure. The fusion session primary key registration process involves the consistent fusion result set generation unit generating a fusion session primary key and establishing a binding record between the fusion session primary key and the fusion solution configuration structure. Subsequently, the consistent fusion result set generation unit extracts the graph node set and graph edge set from the fusion solution configuration structure, performs constraint satisfaction checks, and writes nodes and edges that do not meet the constraints into the input record of the conflict residual field calculation process. The conflict residual field calculation process generates conflict residual fields for nodes and edges that do not meet the constraints and registers their association with the corresponding asset unit primary key, fragment unique identifier, and evidence chain field. The fusion result field generation process performs value merging on candidate fields in the graph node set under the evidence priority sequence constraints and writes them into the fusion result field to form a consistent fusion result set. The consistent fusion result set consists of records associated with the fusion session primary key, fusion result field, conflict residual field, and evidence chain field. The consistent fusion result set generation unit provides the consistent fusion result set to the closed-loop processing and result version chain generation unit, which then generates the abnormal event set, processing strategy set, and result version chain structure and writes back the collection session configuration structure.

[0119] The closed-loop processing and result version chain generation unit 08 is used to receive the consistency fusion result set and perform abnormal event generation processing, processing strategy matching processing, and result version chain registration processing to generate an investigation abnormal event set, a processing strategy set, and a result version chain structure. It also generates and outputs a collection session configuration structure based on the investigation abnormal event set and the processing strategy set. Specifically, after receiving the consistency fusion result set output by the consistency fusion result set generation unit, the closed-loop processing and result version chain generation unit reads the fusion session primary key, fusion result field, conflict residual field, and evidence chain field association records from the consistency fusion result set. It then performs abnormal event generation processing on the conflict residual field to generate abnormal event entries bound to the asset unit primary key, writes the abnormal event entries into the investigation abnormal event set, and simultaneously establishes binding records between the abnormal event entries and the fragment unique identifier and evidence chain field association records. The handling strategy matching process involves the closed-loop handling and result version chain generation unit reading each abnormal event entry in the abnormal event set, and calling the handling strategy matching processing rule entries according to the conflict residual field and evidence chain field associated records registered in the abnormal event entries. This generates handling strategy entries and writes them into the handling strategy set. Each handling strategy entry includes a sampling caliber field, a sampling window field, and a data source entry record reference relationship field bound to the asset unit primary key. The caliber of these fields is kept consistent with the field calibers in the baseline anchor layer registration records, verification sampling layer task orchestration records, and change capture layer encrypted sampling task orchestration records in the acquisition session configuration structure. The result version chain registration process involves the closed-loop handling and result version chain generation unit establishing a continuous registration relationship between the fusion session primary key, the abnormal event set entry records, the handling strategy set entry records, and the registration records formed by the terminology dictionary entry set registration, as well as the caliber field mapping records, and writing this relationship into the result version chain structure. Subsequently, the closed-loop processing and result version chain generation unit generates a collection session configuration structure based on the abnormal event set and the processing strategy set. It writes the sampling caliber field, sampling window field, and data source entry record reference relationship field from the processing strategy entries into the new collection session configuration structure, and outputs the collection session configuration structure to the multi-source data packet generation unit. As the collection session configuration structure received by the multi-source data packet generation unit, it participates in the next round of remote sensing image data and lidar point cloud data collection and processing, ground survey data collection and processing, and historical data loading and processing. At the same time, it establishes an association record between the collection session configuration structure and the result version chain structure for subsequent rounds of abnormal event generation processing and result version chain registration processing to read.

Claims

1. A method for land resource asset inventory based on multi-source data analysis, characterized in that, include: Acquire the boundary data of the target domain, the list of land asset objects, the list of multi-source data sources and the asset unit division rules, perform asset unit segmentation, register the capability fields of multi-source data sources and perform anchor point candidate matching, and generate the collection session configuration structure. Based on the data collection session configuration structure, the system performs ground survey data collection and historical inventory data loading and processing to generate a multi-source fragmented evidence package structure. Based on the multi-source fragmented evidence package set structure, geometric consistency sub-vector construction, temporal consistency sub-vector construction and semantic consistency sub-vector construction are performed to generate a consistency conflict graph structure. Based on the consistency conflict graph structure, the fusion constraint rule set is loaded, the evidence priority sequence is generated, and the fusion session primary key registration is processed to generate the collection session configuration structure.

2. The method according to claim 1, characterized in that, The process of dividing asset units also includes: The asset unit segmentation process includes sorting candidate boundary sources by the parcel boundary priority field, performing boundary trimming or pruning actions based on the validity status of the boundary version identifier using the parcel boundary or grid skeleton, and performing void area threshold comparison and connectivity merging based on the suspected void repair rule field; and performing multi-source data source capability field registration processing.

3. The method according to claim 1, characterized in that, The process of registering capability fields from multiple data sources and handling anchor candidate matching also includes: The multi-source data source capability field registration process includes reading the data source identifier, source type identifier, coverage identifier, refresh interval field, acquisition method identifier, and interface stability identifier from the multi-source data source list and generating capability field records by combining them with historical fetch logs; performing anchor point candidate matching processing, which includes constructing an anchor point candidate matching request for the primary key field of each asset unit, filtering data source identifiers in the capability field records whose coverage identifier field and spatial range identifier field have an overlap relationship, generating candidate sorting by spatial resolution field priority order field, point cloud density level field priority order field, and interface stability level field priority order field, and performing time availability verification; and generating the acquisition session configuration structure.

4. The method according to claim 1, characterized in that, The process of collecting ground survey data and loading and processing historical inventory data also includes: The ground survey data acquisition and processing includes reading the sampling evidence chain field from the acquisition session configuration structure to generate an acquisition instruction package, sending it to the mobile acquisition terminal to start the acquisition session, acquiring the positioning trajectory record index, the field image index and the set of mandatory attribute fields, and performing form field integrity verification and basic legality verification; performing historical inventory data loading processing, which includes retrieving historical inventory result files by sampling spatial range field, filtering historical result batches by sampling time window field and historical version identifier field falling into the window, and performing file index loading; aggregating the acquisition results into multi-source data packets; and performing coordinate projection standard unification processing based on multi-source data packets, which includes reading coordinate reference labels from remote sensing image data fragments, lidar point cloud data fragments, ground survey data fragments and historical inventory data fragments and performing projection transformation to generate unified coordinate reference labels.

5. The method according to claim 1, characterized in that, The process of constructing geometrically consistent subvectors also includes: The geometric consistency sub-vector construction process includes reading the coordinate system identifier and projection parameter identifier in the aperture field, uniformly projecting geometric fragments from different sources to the standard coordinate frame, generating a standardized geometric contour set, and calculating the spatial overlap field, morphological offset field, and boundary fit field; and performing temporal consistency sub-vector construction process.

6. The method according to claim 1, characterized in that, The processes of constructing temporally consistent subvectors and semantically consistent subvectors also include: The temporal consistency sub-vector construction process includes extracting the collection time, submission time, and historical version identifier fields from the evidence chain field and converting them into a standard timeline format; aggregating time series fragments to calculate the sampling density field, time drift field, and rate of change field; and performing semantic consistency sub-vector construction process, which includes reading the source identifier, sampled evidence type, investigator identifier, and historical attribute labels from the evidence chain field; uniformly mapping them to a vocabulary standard set; establishing an attribute correspondence matrix based on the terminology dictionary entry set; merging synonym items; performing conflict labeling on contradictory items to generate a semantic difference marker field; and calculating the consistency distribution field and semantic confidence field.

7. The method according to claim 1, characterized in that, The process of loading the fusion constraint rule set also includes: The fusion constraint rule set loading process includes reading the set of rule entries that match the rule version identifier field from the rule repository service, filtering constraint subsets by the applicable conflict type field, adjusting the loading order of constraint subsets according to the conflict intensity field, and registering hard constraint entry fields and soft constraint entry fields.

8. The method according to claim 1, characterized in that, The process of generating the evidence priority sequence also includes: The evidence priority sequence generation process includes initiating evidence retrieval requests to the evidence index service for the graph node set elements, extracting the evidence score field from the quality field and the evidence chain field, and sorting the evidence list according to the evidence score field to generate an evidence priority sequence.

9. The method according to claim 1, characterized in that, The process of merging session primary key registration also includes: The fusion session primary key registration process includes reading the conflict graph batch identifier field, rule version identifier field, boundary version identifier field, and evidence priority sequence field index to generate session registration input records, performing uniqueness verification, and returning the fusion session primary key.

10. A land resource asset inventory system based on multi-source data analysis, characterized in that, include: The asset unit primary key table generation unit, anchor point candidate list generation unit, acquisition session configuration structure generation unit, multi-source data packet generation unit, multi-source fragmented evidence packet set structure generation unit, consistency conflict graph structure generation unit, consistency fusion result set generation unit, and closed-loop processing and result version chain generation unit are connected in sequence to implement the method described in any one of claims 1-9.

Images