Automatic Update Method for Engineering GIS Based on Multi-Source Data Fusion

Abstract

This invention discloses an automatic update method for engineering GIS based on multi-source data fusion, comprising: acquiring multi-source data, performing coordinate unification and time alignment, and performing noise removal and format standardization processing; constructing a historical state mirror, differentiating it with the current multi-source data, and forming candidate change objects and change types; constructing a construction state evolution chain, inverting theoretically existing geometric connection states, and screening candidate update objects; constructing an evidence spatiotemporal superposition, forming an evidence band structure, and determining geometric attribute topology candidate update actions; generating reversible topology patches, constructing a patch sandbox, and performing consistency verification and rollback recoverability verification; submitting the patch to complete the GIS update, generating version and rollback records, and blocking the output of audit results if the update fails. This invention achieves accurate, secure, and automatic updates of engineering GIS data through construction time-series constraint inversion, multi-source evidence spatiotemporal superposition, and reversible topology patch sandbox verification.

Description

Technical Field

[0001] This invention relates to the field of geographic information system technology, and in particular to an automatic update method for engineering GIS based on multi-source data fusion. Background Technology

[0002] Engineering GIS is a crucial technology for spatial representation, attribute management, and connectivity maintenance of engineering objects such as roads, pipelines, bridges, tunnels, park facilities, and municipal infrastructure. It has wide applications in engineering construction, operation and maintenance management, and digital delivery. With the continuous accumulation of remote sensing imagery, laser point clouds, BIM models, IoT sensor data, construction plans, construction logs, and historical GIS data, automatically updating engineering GIS using multi-source data has become an important development direction for improving update efficiency and reducing manual maintenance costs. Existing technologies typically update the geometric information, attribute information, or topological relationships of engineering objects by preprocessing, detecting changes, and writing the results to a database, thereby meeting the requirements of data timeliness and consistency in engineering scenarios.

[0003] Most existing technologies focus on identifying changes in engineering objects using image difference analysis, point cloud comparison, BIM comparison, or sensor data. They primarily rely on observation results to determine whether to perform an update, lacking joint constraints on construction plans, construction logs, component dependencies, and historical update status. This makes it difficult to correlate detected changes with the actual construction process. When engineering objects are in a phased construction, partial relocation, temporary connection, or demolition transition state, relying solely on observational differences can easily misjudge noise disturbances, isolated field changes, or phased anomalies as genuine updates, resulting in inaccurate selection of update objects, unreasonable update timing, and errors in update type identification.

[0004] Existing technologies, when faced with inconsistent or insufficiently continuous evidence from multiple sources, typically lack further screening mechanisms for update actions. This can easily trigger high-risk updates when evidence is insufficient, especially when adjustments to connectivity are involved. Such updates may cause local topology breaks, duplicate connections, adjacency conflicts, or abnormal states of associated objects. Most existing update methods rely on direct database writing, lacking measures for sandboxing and rollback replays of patches before formal updates. If the update result is unreasonable, it is difficult to promptly revert to the pre-update state.

[0005] Therefore, how to provide an automatic update method for engineering GIS based on multi-source data fusion is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose an automatic update method for engineering GIS based on multi-source data fusion. This invention comprehensively utilizes construction time-series constraint inversion, spatiotemporal overlay of multi-source evidence, and reversible topology patch sandbox verification technology to perform hierarchical judgment and secure update processing of changes in engineering objects. It realizes constraint control and verification management of the automatic update process of engineering GIS data, and has the advantages of high accuracy of update decision, low false update rate, strong topological relationship stability, and rollback of update results.

[0007] The automatic update method for engineering GIS based on multi-source data fusion according to embodiments of the present invention includes: Acquire multi-source data of the engineering area, perform standardization processing on the multi-source data, and obtain standardized multi-source data; Based on standardized multi-source data, a historical state image of the engineering object is constructed. The current multi-source data and the historical state image are then differentially analyzed on an object-by-object basis to obtain differential segments. These differential segments are then classified and merged to form candidate change objects. For each candidate change object, a construction state evolution chain is constructed. Based on the construction state evolution chain, the theoretical existence state, theoretical geometric state, and theoretical connection state of the candidate change object at the current moment are inverted, and candidate update objects that meet the construction time sequence constraints are selected. For candidate update objects, a multi-source evidence spatiotemporal overlay is constructed, and evidence from different data sources is mapped to a unified spatial grid and time slice to form evidence units. Based on the source type and spatiotemporal overlap of each evidence unit, an evidence overlay distribution is generated to determine the candidate update actions corresponding to geometric update, attribute update and topology update. For candidate update actions, generate reversible topology patches, construct a reversible topology patch sandbox, load the reversible topology patches into the reversible topology patch sandbox, and perform topology consistency verification, engineering rule verification, related object influence propagation verification, and rollback recoverability verification based on the loaded sandbox state; When a reversible topology patch passes verification, it is submitted to the engineering GIS database to complete automatic updates, generating corresponding version records and rollback records. When it fails verification, updates are blocked and a pending review result is output.

[0008] Optionally, the multi-source data includes remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction plan data, construction log data, and historical GIS data.

[0009] Optionally, the standardization process performed on the multi-source data to obtain standardized multi-source data includes: Coordinate system unification is applied to multi-source data, transforming data from different coordinate references to the same target coordinate system. Time alignment is performed on the acquisition time, recording time, or generation time corresponding to various types of data to form time-series data under a unified time benchmark. Outliers, noisy data, and duplicate data in multi-source data are removed. Data formats from different data sources are converted to form a unified data organization format. Engineering object identifiers, time identifiers, and status identifiers in text data are extracted in a structured manner. Missing fields are marked with null values ​​or filled in with correlations to obtain standardized multi-source data.

[0010] Optionally, classifying and merging the difference segments to form candidate change objects includes: Based on standardized multi-source data extraction, the geometric information, attribute information and connection relationship information of each engineering object in the engineering area at historical time are extracted. According to the engineering object identifier, historical GIS data, historical BIM model data, historical remote sensing image data, historical laser point cloud data and historical IoT sensor data are aligned at the object level to construct the historical state mirror of each engineering object. The historical state mirror includes the historical geometric mirror of the object, the historical attribute mirror of the object and the historical connection mirror of the object. Extract the current geometric information, current attribute information, and current connection relationship information corresponding to each engineering object from the multi-source data at the current moment. Compare the current state of the engineering object with the historical state mirror image for each object to generate geometric difference fragments, attribute difference fragments, and connection difference fragments. Based on geometric difference segments, a partitioned occupancy change sequence of engineering objects is constructed. The historical spatial occupancy range of the same engineering object is divided into an internal stable region, a boundary transition region, and an external adjacent region. The occupancy change of multi-source data in the internal stable region, the boundary transition region, and the external adjacent region at the current moment is statistically analyzed to obtain the geometric difference degree. The attribute change chain of the project object is constructed based on the attribute difference fragment. The attribute change field at the current moment is matched with the attribute registration order, attribute association field and attribute source record in the historical state mirror item by item to identify the main attribute change, derived attribute change and isolated attribute change and obtain the attribute difference degree. The connection migration trajectory of the engineering object is constructed based on the connection difference segment. The connection relationship change at the current moment is compared with the connection direction, connection position and connection order in the historical state mirror. The connection establishment, connection disconnection and connection transfer are identified to obtain the connection difference degree. Based on the geometric difference degree, attribute difference degree and connection difference degree, the spatially adjacent and temporally continuous difference segments are merged to form candidate change objects and their corresponding geometric change, attribute change and topological change types.

[0011] Optionally, the filtering process for obtaining candidate update objects that satisfy construction timing constraints includes: For candidate change objects, extract the construction plan data, construction log data, component dependencies in the BIM model, IoT sensor data and historical GIS data corresponding to the candidate change objects, and associate them according to a unified object identifier and a unified time sequence to form a construction constraint information set. Based on the construction constraint information set, a construction state evolution chain for candidate change objects is constructed. The construction state evolution chain includes the states of no construction, preparation for construction, construction in progress, construction completion, and acceptance effective. State transition conditions are written between adjacent states. State dependency relationships between candidate change objects and preceding components, related components on the same floor, and related continuation components are established. The process records in the construction log, the state change records in the IoT sensor data, and the update records in the historical GIS data are mapped to the corresponding state segments. Based on the construction state evolution chain, the target construction stage to which the candidate change object belongs at the current moment is determined. When the state records corresponding to multiple data sources are inconsistent, the duration corresponding to each state record is statistically analyzed, and compared according to the predefined state priority order in the construction state evolution chain. Among the state records that meet the minimum duration threshold, the state that is consistent with the state priority order is selected as the current stage state. Based on the target construction stage, the theoretical existence state, theoretical geometric state, and theoretical connection state of the candidate change objects at the current moment are inverted. The theoretical existence state includes not appearing, appearing, partially appearing, waiting to be demolished, and demolished. The theoretical geometric state includes maintaining historical geometry, local extension, local reduction, location migration, and forming a new boundary. The theoretical connection state includes maintaining the original connection, establishing a connection, disconnecting the connection, migrating the connection, and forming a temporary connection. The theoretical existence state, theoretical geometric state, and theoretical connectivity state are compared with the difference results corresponding to the candidate change objects. Candidate change objects that match all three are retained as candidate update objects, and the remaining candidate change objects are marked as objects to be constrained and verified.

[0012] Optionally, determining the candidate update actions corresponding to geometry update, attribute update, and topology update includes: Update evidence is extracted from remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction log data, and historical GIS data corresponding to candidate update objects. The evidence is then linked and organized according to a unified object identifier, a unified spatial location, and a unified time sequence to form an original evidence set. A spatiotemporal superposition of multi-source evidence is constructed based on the original evidence set. This multi-source evidence spatiotemporal superposition is composed of a combination of spatial hierarchical units and temporal segmentation units, wherein: The spatial layering unit is divided into an internal core layer, a boundary transition layer, and an external influence layer according to the historical spatial boundary of the candidate update object, the boundary position, and the outward adjacent range. The time segmentation unit is divided into pre-stage time period, intra-stage time period and post-stage time period according to the stage switching time of the candidate update object in the construction state evolution chain. Each piece of evidence in the original evidence set is written into the corresponding spatial stratification unit and time segmentation unit, forming an evidence occupancy structure that simultaneously represents the spatial stratification and time stage. For multi-source evidence spatiotemporal superposition, the evidence written in each spatial layer unit and time segment unit is superimposed and organized, and the distribution position of each piece of evidence in different spatial layers, the persistence in different time segments, and the extension between adjacent spatial layers and adjacent time segments are recorded to form the evidence superposition trajectory corresponding to the candidate update object. Based on the evidence superposition trajectory, geometric update evidence bands, attribute update evidence bands, and topological update evidence bands are constructed for candidate update objects respectively; Based on geometric update evidence bands, attribute update evidence bands, and topology update evidence bands, the geometric update actions, attribute update actions, and topology update actions corresponding to the candidate update objects are identified respectively.

[0013] Optionally, the step of performing topology consistency verification, engineering rule verification, related object influence propagation verification, and rollback recoverability verification based on the loaded sandbox state includes: For candidate update actions, read the pre-update state of the candidate update object in the historical state mirror and the post-update state in the multi-source data at the current moment, and extract the corresponding geometric state, attribute state and connection relationship state to form a pre-update state set and a post-update state set. A reversible topology patch is generated based on the state set before and after the update. The reversible topology patch includes geometric changes, attribute changes, and topology changes. A rollback record is written for each change. A reversible topology patch sandbox is constructed, comprising a base snapshot area, a patch application area, and a rollback / replay area, wherein: The basic snapshot area stores a local subgraph snapshot of the candidate update objects and their associated objects before the update. The patch application area generates a virtual update state by applying reversible topology patches without writing to the project GIS database; The rollback and replay area performs a rollback operation on the virtual update state and restores it to the local subgraph snapshot. The local subgraph snapshot is centered on the candidate update object and extracts the associated objects and associated connections with a preset association depth, including cross-layer associated elements related to the candidate update action. The reversible topology patch is loaded into the patch application area. After generating the virtual update state, consistency verification is performed. Geometric conflict verification and adjacency relationship verification are performed on associated objects involving spatial adjacency. Influence propagation verification is performed on the virtual update state to identify whether the connection changes of candidate update objects cause changes in the connected domains of associated objects or breakage of association relationships. In the rollback replay area, the virtual update state is rolled back and replayed item by item according to the rollback record. The rolled-back state is compared and verified with the local subgraph snapshot in the basic snapshot area to obtain the rollback recoverability verification result. When both the consistency verification result and the rollback recoverability verification result meet the preset commit conditions, the reversible topology patch is determined as a committable patch; otherwise, the reversible topology patch is determined as an uncommittable patch.

[0014] Optionally, the step of submitting the reversible topology patch to the engineering GIS database for automatic updating upon successful verification includes: Upon receiving a confirmed committable patch, the geometric change, attribute change, and topological change in the reversible topology patch are mapped to the corresponding object records, attribute fields, and connection relationship tables in the engineering GIS database, respectively, and a patch commit unit is generated. Based on the patch submission unit, update operations are performed in the engineering GIS database. Changes are written item by item in the order of attribute update, geometric update and topology update. During the topology update process, the connection records of related objects that have a connection relationship with the candidate update object are updated synchronously. After the update is completed, a version record and a rollback record are generated. The version record includes the update object identifier, update time, and update content identifier. The rollback record includes the geometric data, attribute data, and connection relationship data corresponding to the state before the update. The version record and the rollback record are then stored together. When a reversible topology patch fails verification, the update write operation is terminated and an update record to be reviewed is generated. The update record to be reviewed includes the candidate update object identifier, the type of failure to pass verification, and the corresponding differential result.

[0015] The beneficial effects of this invention are: Compared with existing automatic update technologies for engineering GIS, this invention first introduces a construction time-series constraint inversion mechanism at the update trigger level. It jointly associates construction plans, construction logs, BIM component dependencies, IoT sensor status, and historical GIS status, so that the judgment of changes in engineering objects no longer depends solely on single observation results, but can constrain and verify the theoretical existence state, theoretical geometric state, and theoretical connection state of objects at the current moment in conjunction with the engineering implementation process. This effectively reduces misjudgments caused by phased construction, temporary state changes, local abnormal collection, or isolated field changes, and improves the accuracy of candidate update object identification and the rationality of update timing determination.

[0016] The invention constructs a spatiotemporal overlay of multi-source evidence, organizing update evidence from different sources according to a unified spatial hierarchy and temporal segmentation. This forms evidence overlay trajectories and evidence bands oriented towards geometric updates, attribute updates, and topological updates. This ensures that the determination of update actions is based on the continuous distribution of multi-source evidence, rather than on a single data source or discrete differences. This effectively enhances the collaborative utilization capability among multi-source heterogeneous data, strengthens the ability to identify real changes, and reduces the probability of triggering high-risk updates when evidence is insufficient or discontinuous. As a result, it improves the reliability and stability of automatic updates in engineering GIS.

[0017] This invention constructs a reversible topology patch sandbox before formally writing to the engineering GIS database. Through a basic snapshot area, patch application area, and rollback replay area, candidate update actions are pre-tested and rolled back for verification. This allows the update results to complete checks on connection integrity, connection direction, spatial adjacency, and impact propagation range before submission, and can verify the recovery capability after update failure. It effectively avoids topology breaks, duplicate connections, abnormal association relationships, and unrecoverable problems caused by direct database writing, improves the security, topology consistency, and version controllability of automatic updates of engineering GIS, and realizes accurate, stable, and rollbackable updates of engineering GIS data under multi-source data fusion conditions. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the automatic update method for engineering GIS based on multi-source data fusion proposed in this invention; Figure 2 This diagram illustrates the generation of reversible topology patches and the construction of a reversible topology patch sandbox in the automatic update method for engineering GIS based on multi-source data fusion proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figure 1 and Figure 2 An automatic update method for engineering GIS based on multi-source data fusion includes: Acquire multi-source data of the engineering area, perform standardization processing on the multi-source data, and obtain standardized multi-source data; Based on standardized multi-source data, a historical state image of the engineering object is constructed. The current multi-source data and the historical state image are then differentially analyzed on an object-by-object basis to obtain differential segments. These differential segments are then classified and merged to form candidate change objects. For each candidate change object, a construction state evolution chain is constructed. Based on the construction state evolution chain, the theoretical existence state, theoretical geometric state, and theoretical connection state of the candidate change object at the current moment are inverted, and candidate update objects that meet the construction time sequence constraints are selected. For candidate update objects, a multi-source evidence spatiotemporal overlay is constructed, and evidence from different data sources is mapped to a unified spatial grid and time slice to form evidence units. Based on the source type and spatiotemporal overlap of each evidence unit, an evidence overlay distribution is generated to determine the candidate update actions corresponding to geometric update, attribute update and topology update. For candidate update actions, generate reversible topology patches, construct a reversible topology patch sandbox, load the reversible topology patches into the reversible topology patch sandbox, and perform topology consistency verification, engineering rule verification, related object influence propagation verification, and rollback recoverability verification based on the loaded sandbox state; When a reversible topology patch passes verification, it is submitted to the engineering GIS database to complete automatic updates, generating corresponding version records and rollback records. When it fails verification, updates are blocked and a pending review result is output.

[0021] In this embodiment, the multi-source data includes remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction plan data, construction log data, and historical GIS data.

[0022] In this embodiment, the standardization process performed on the multi-source data to obtain standardized multi-source data includes: Coordinate system unification is applied to multi-source data, transforming data from different coordinate references to the same target coordinate system. Time alignment is performed on the acquisition time, recording time, or generation time corresponding to various types of data to form time-series data under a unified time benchmark. Outliers, noisy data, and duplicate data in multi-source data are removed. Data formats from different data sources are converted to form a unified data organization format. Engineering object identifiers, time identifiers, and status identifiers in text data are extracted in a structured manner. Missing fields are marked with null values ​​or filled in with correlations to obtain standardized multi-source data.

[0023] In this embodiment, the step of classifying and merging the difference segments to form candidate change objects includes: Based on standardized multi-source data extraction, geometric, attribute, and connectivity information of each engineering object within the engineering area at historical moments is obtained. Historical GIS data, historical BIM model data, historical remote sensing image data, historical laser point cloud data, and historical IoT sensor data are aligned at the object level according to the engineering object identifier, constructing a historical state mirror of each engineering object. This historical state mirror includes a historical geometric mirror, a historical attribute mirror, and a historical connectivity mirror of the object. Object history geometric mirror is used to represent the spatial occupancy, location and shape of an engineering object at a historical moment; object history attribute mirror is used to represent the attribute registration content and status content of an engineering object at a historical moment; object history connection mirror is used to represent the connection direction, connection position and connection number between an engineering object and adjacent engineering objects at a historical moment. Extract the current geometric information, current attribute information, and current connectivity information corresponding to each engineering object from the multi-source data at the current moment. Compare the current state of the engineering object with its historical state mirror image object by object to generate geometric difference fragments, attribute difference fragments, and connectivity difference fragments, among which: Among them, geometric difference fragments are used to characterize the deviation area between the current space occupancy and the historical space occupancy, attribute difference fragments are used to characterize the changed fields between the current attribute registration content and the historical attribute registration content, and connection difference fragments are used to characterize the new connection, missing connection or connection position offset between the current connection relationship and the historical connection relationship. Based on geometric difference segments, a partitioned occupancy change sequence of engineering objects is constructed. The historical spatial occupancy range of the same engineering object is divided into an internal stable region, a boundary transition region, and an external adjacent region. The occupancy changes of multi-source data in the internal stable region, boundary transition region, and external adjacent region at the current time are statistically analyzed to obtain the geometric difference degree. The division of the internal stable region, boundary transition region, and external adjacent region is as follows: The spatial occupancy range of the engineering object in the historical state mirror is used as the reference area. The area within the reference area that is within a preset distance from the boundary is removed as the internal stable area. The ring-shaped area formed by extending the boundary of the reference area inward and outward by preset distances is used as the boundary transition area. The area formed by extending the boundary of the reference area outward by a preset adjacency distance is used as the external adjacency area. The preset distance and preset adjacency distance are set according to the type of engineering object and the spatial accuracy requirements. The geometric difference is determined by the location of the occupancy change, the direction of the occupancy change, and the continuous range of the occupancy change in each zone, in order to distinguish local noise disturbances from the actual geometric changes of the engineering object; The attribute change chain of the project object is constructed based on the attribute difference fragment. The attribute change field at the current moment is matched with the attribute registration order, attribute association field and attribute source record in the historical state mirror item by item to identify the main attribute change, derived attribute change and isolated attribute change and obtain the attribute difference degree. The attribute difference degree is determined by the number of main attribute changes, the number of consecutively changed fields and the degree of change consistent with the historical attribute association relationship, so as to distinguish between the real attribute update of the project object and the abnormal modification of isolated fields. The connection migration trajectory of the engineering object is constructed based on the connection difference segment. The connection relationship change at the current moment is compared with the connection direction, connection position and connection order in the historical state mirror. The connection establishment, connection disconnection and connection transfer are identified to obtain the connection difference degree. Based on the geometric difference degree, attribute difference degree and connection difference degree, the spatially adjacent and temporally continuous difference segments are merged to form candidate change objects and their corresponding geometric change, attribute change and topological change types.

[0024] In this embodiment, the process of filtering to obtain candidate update objects that satisfy the construction sequence constraints includes: For candidate change objects, extract the construction plan data, construction log data, component dependencies in the BIM model, IoT sensor data and historical GIS data corresponding to the candidate change objects, and associate them according to a unified object identifier and a unified time sequence to form a construction constraint information set. Based on the construction constraint information set, a construction state evolution chain for candidate change objects is constructed. The construction state evolution chain includes the states of no construction, preparation for construction, construction in progress, construction completion, and acceptance effective. State transition conditions are written between adjacent states. State dependency relationships between candidate change objects and preceding components, related components on the same floor, and related continuation components are established. The process records in the construction log, the state change records in the IoT sensor data, and the update records in the historical GIS data are mapped to the corresponding state segments. Based on the construction state evolution chain, the target construction stage to which the candidate change object belongs at the current moment is determined. When the state records corresponding to multiple data sources are inconsistent, the duration corresponding to each state record is statistically analyzed, and compared according to the predefined state priority order in the construction state evolution chain. Among the state records that meet the minimum duration threshold, the state that is consistent with the state priority order is selected as the current stage state. Based on the theoretical existence, theoretical geometric, and theoretical connection states of candidate change objects at the current moment during the target construction phase, the theoretical existence states include: not yet appeared, already appeared, partially appeared, awaiting demolition, and already demolished; the theoretical geometric states include: maintaining historical geometry, local extension, local reduction, location migration, and forming a new boundary; and the theoretical connection states include: maintaining the original connection, establishing a connection, disconnecting a connection, migrating a connection, and forming a temporary connection. Specifically, the theoretical existence, theoretical geometric, and theoretical connection states of candidate change objects at the current moment during the target construction phase are as follows: Based on the stage position of the target construction stage in the construction state evolution chain, determine whether the candidate change object has reached the stage that allows its appearance or demolition. When the target construction stage corresponds to the component generation stage or the construction completion stage, the theoretical existence state is determined to have appeared or partially appeared. When the target construction stage corresponds to the demolition stage or the demolition completion stage, the theoretical existence state is determined to be to be demolished or demolished. Otherwise, it is determined to have not appeared or to remain in the original state. Based on the construction process type corresponding to the target construction stage and the component geometric change information in the BIM model, the geometric change mode of the candidate change object is determined. When the target construction stage corresponds to the laying, expansion or reinforcement process, the theoretical geometric state is determined as local extension or the formation of a new boundary. When the target construction stage corresponds to the demolition or reduction process, the theoretical geometric state is determined as local reduction. When the target construction stage corresponds to the relocation process, the theoretical geometric state is determined as location migration. Otherwise, it is determined as maintaining the historical geometry. Based on the connection operation information and component dependencies corresponding to the target construction stage, determine the connection relationship change mode of the candidate change object. When the target construction stage corresponds to the connection establishment process, the theoretical connection state is determined to be the establishment of connection or the formation of temporary connection. When the target construction stage corresponds to the demolition or disconnection process, the theoretical connection state is determined to be the disconnection connection. When the target construction stage corresponds to the relocation or switching process, the theoretical connection state is determined to be the migration connection. Otherwise, the original connection is maintained. The theoretical existence state, theoretical geometric state, and theoretical connectivity state are compared with the difference results corresponding to the candidate change objects. Candidate change objects that match all three are retained as candidate update objects, and the remaining candidate change objects are marked as objects to be constrained and verified.

[0025] In this embodiment, determining the candidate update actions corresponding to geometric update, attribute update, and topology update includes: Update evidence is extracted from remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction log data, and historical GIS data corresponding to candidate update objects. The evidence is then linked and organized according to a unified object identifier, a unified spatial location, and a unified time sequence to form an original evidence set. A spatiotemporal superposition of multi-source evidence is constructed based on the original evidence set. This multi-source evidence spatiotemporal superposition is composed of a combination of spatial hierarchical units and temporal segmentation units, wherein: The spatial layering unit is divided into an internal core layer, a boundary transition layer, and an external influence layer according to the historical spatial boundary of the candidate update object, the boundary position, and the outward adjacent range. The time segmentation unit is divided into pre-stage time period, intra-stage time period and post-stage time period according to the stage switching time of the candidate update object in the construction state evolution chain. Each piece of evidence in the original evidence set is written into the corresponding spatial stratification unit and time segmentation unit, forming an evidence occupancy structure that simultaneously represents the spatial stratification and time stage. For the spatiotemporal superposition of multi-source evidence, the evidence written in each spatial layer unit and time segment unit is superimposed and organized. The distribution position of each piece of evidence in different spatial layers, the persistence in different time segments, and the extension between adjacent spatial layers and adjacent time segments are recorded to form the evidence superposition trajectory corresponding to the candidate update object. The evidence superposition trajectory is used to characterize the spatial expansion process of the same update evidence from the internal core layer to the boundary transition layer or the external influence layer, as well as the temporal continuation process across adjacent construction stages. Based on the evidence superposition trajectory, geometric update evidence bands, attribute update evidence bands, and topological update evidence bands are constructed for candidate update objects. The geometric update evidence band is formed by geometric evidence that appears continuously in the internal core layer and the boundary transition layer. The attribute update evidence band is formed by attribute evidence that continuously corresponds to the candidate update object in the same time segment. The topological update evidence band is formed by connection evidence that appears continuously in the boundary transition layer and the external influence layer and corresponds to the change in connection relationship. Based on geometric update evidence bands, attribute update evidence bands, and topology update evidence bands, the geometric update actions, attribute update actions, and topology update actions corresponding to the candidate update objects are identified respectively.

[0026] In this embodiment, the process of performing topology consistency verification, engineering rule verification, related object influence propagation verification, and rollback recoverability verification based on the loaded sandbox state includes: For candidate update actions, read the pre-update state of the candidate update object in the historical state mirror and the post-update state in the multi-source data at the current moment, and extract the corresponding geometric state, attribute state and connection relationship state to form a pre-update state set and a post-update state set. A reversible topology patch is generated based on the state set before and after the update. The reversible topology patch includes geometric change, attribute change and topology change. A rollback record is written for each change. The rollback record includes the value before the update, the value after the update, the object identifier and the location identifier. A reversible topology patch sandbox is constructed, comprising a base snapshot area, a patch application area, and a rollback / replay area, wherein: The basic snapshot area stores a local subgraph snapshot of the candidate update objects and their associated objects before the update. The patch application area generates a virtual update state by applying reversible topology patches without writing to the project GIS database; The rollback and replay area performs a rollback operation on the virtual update state and restores it to the local subgraph snapshot. The local subgraph snapshot is centered on the candidate update object and extracts the associated objects and associated connections with a preset association depth. It includes cross-layer associated elements related to the candidate update action. The preset association depth is 2 layers of connection relationship depth. When the candidate update action is a topology update, the preset association depth is expanded to 3 layers of connection relationship depth. The reversible topology patch sandbox is used to build an independent virtual update environment before the formal update of the engineering GIS database. It performs pre-rehearsal verification and rollback rehearsal verification of candidate update actions. Through the synergistic effect of the basic snapshot area, patch application area and rollback rehearsal area, it realizes full-process control of pre-update state preservation, update process simulation and post-update state recovery. By generating virtual update state and performing consistency verification in the patch application area, it identifies abnormal connection relationships, spatial conflicts and connectivity destruction problems in advance. The rollback rehearsal area verifies the recoverability of the update process, ensuring that the update operation has complete rollback capability after execution. It avoids topology pollution and irreversible errors caused by directly writing to the engineering GIS database, improves the security, stability and controllability of the automatic update process of engineering GIS, reduces the impact of erroneous updates on the overall topology structure, and improves the reliability of update results under multi-source data fusion conditions. The reversible topology patch is loaded into the patch application area. After generating a virtual update state, consistency verification is performed. Geometric conflict and adjacency checks are performed on associated objects involving spatial adjacency. Influence propagation checks are performed on the virtual update state to identify whether connection changes of candidate update objects have caused changes in the connected components of associated objects or broken association relationships. Specifically, the consistency verification performed after generating the virtual update state is as follows: Based on the virtual update state, the integrity of the connection relationship between the candidate update object and the associated object is checked to detect whether there are missing connections, duplicate connections or offset connection positions, and the connection direction and the number of connections are compared for consistency. Based on the virtual update state, a geometric consistency check is performed on the candidate update object and its spatial adjacent objects to determine whether the updated geometric range overlaps, intersects, or has abnormal spacing with the adjacent objects, and to check whether the adjacency relationship remains continuous. Based on the virtual update state, propagation analysis is performed on the connection relationship changes of candidate update objects. The connectivity relationship of associated objects is tracked along the preset association depth range to identify whether it causes changes in the connected domain or breaks in the association relationship. When no abnormal propagation results are detected, the consistency verification is deemed to pass; otherwise, the consistency verification is deemed to fail. In the rollback replay area, the virtual update state is rolled back and replayed item by item according to the rollback record. The rolled-back state is compared and verified with the local subgraph snapshot in the base snapshot area to obtain the rollback recoverability verification result. When both the consistency verification result and the rollback recoverability verification result meet the preset commit conditions, the reversible topology patch is determined to be a committable patch; otherwise, the reversible topology patch is determined to be an uncommittable patch. Specifically, the rollback replay of the virtual update state in the rollback replay area is performed item by item according to the rollback record as follows: Based on the updated object identifier and field identifier recorded in the rollback record, the geometric change, attribute change and connection relationship change in the virtual update state are located one by one, and the corresponding change is restored to the original value recorded in the state before the update. After rolling back the geometric and attribute changes, the connection relationships in the virtual update state are rolled back one by one based on the connection relationship recovery information in the rollback record. The original connection direction, connection position and connection number between the candidate update object and the associated object are restored, and the connection record of the associated object is updated synchronously. After all changes have been rolled back, the rolled-back state is compared with the local subgraph snapshot in the base snapshot area on an object-by-object and connection-by-connection basis according to the alignment of object identifier and connection relationship. If the geometric state, attribute state and connection relationship are consistent with the local subgraph snapshot, the rollback is considered successful; otherwise, the rollback is considered to have failed.

[0027] In this embodiment, the step of submitting the reversible topology patch to the engineering GIS database for automatic updating upon successful verification includes: Upon receiving a confirmed committable patch, the geometric change, attribute change, and topological change in the reversible topology patch are mapped to the corresponding object records, attribute fields, and connection relationship tables in the engineering GIS database, respectively, and a patch commit unit is generated. Based on the patch submission unit, update operations are performed in the engineering GIS database. Changes are written item by item in the order of attribute update, geometric update and topology update. During the topology update process, the connection records of related objects that have a connection relationship with the candidate update object are updated synchronously. After the update is completed, a version record and a rollback record are generated. The version record includes the update object identifier, update time, and update content identifier. The rollback record includes the geometric data, attribute data, and connection relationship data corresponding to the state before the update. The version record and the rollback record are then stored together. When a reversible topology patch fails verification, the update write operation is terminated and an update record to be reviewed is generated. The update record to be reviewed includes the candidate update object identifier, the type of failure to pass verification, and the corresponding differential result.

[0028] Example 1: To verify the feasibility of this invention in practice, it was applied to an integrated management project for the construction and operation of an underground municipal pipeline network, covering an area of ​​approximately 12 square kilometers. This project encompasses various engineering objects, including water supply, drainage, electricity, and communications, involving approximately 21,000 pipelines and 12,000 inspection wells. During the parallel construction and operation phases in this area, traditional engineering GIS updates primarily relied on periodic surveying and manual verification, supplemented by remote sensing imagery and point cloud data. In actual operation, typical problems gradually emerged: First, during the construction phase, point clouds or imagery reflected local changes before the construction plan was completed, leading to premature system updates and broken connections; second, there were time differences between multi-source data, for example, IoT sensor data reflected status changes before the attribute ledger was updated, resulting in erroneous attribute modifications; third, the update process directly wrote to the database without a pre-verification mechanism, requiring manual rollback of each data entry after a topology error occurred, resulting in low efficiency.

[0029] The method of this invention was applied in this scenario, and a trial run was conducted in the area for 20 consecutive days. The system first accessed UAV remote sensing imagery (0.15m resolution, updated once daily), mobile laser point cloud data (point density approximately 800 points / m², updated every 3 days), BIM model data (dynamically updated during construction), IoT sensor data (sampling period 10 minutes), construction plan and log data (updated daily), and historical GIS data. The system performed unified coordinate transformation and time alignment on the multi-source data and constructed a historical state mirror of the engineering objects. The system identified candidate changes through differential processing. For example, in a section of power pipeline, a geometric change was detected within an area of ​​approximately 85 meters. However, in the construction state evolution chain, the corresponding process for this pipeline was still in the "construction in progress" stage, not yet in the "construction completed" state. Therefore, this change was not selected as a candidate for update, avoiding premature updates.

[0030] For candidate update objects selected through construction constraints, the system further constructs a multi-source evidence spatiotemporal overlay, spatially dividing image changes, point cloud changes, and IoT traffic changes into an internal core layer, a boundary transition layer, and an external adjacent layer, and temporally dividing them into time periods corresponding to adjacent construction stages. For example, during the expansion of a certain inspection well, image and point cloud changes were detected in both the boundary transition layer and the internal core layer for two consecutive days, and IoT traffic changes appeared simultaneously, forming a continuous evidence band. Based on this, the system determined it to be a valid update action. On the other hand, in another case where only a single image change occurred without support from other data sources, it was identified as noise disturbance and did not enter the update process.

[0031] Before executing the update, the system builds a reversible topology patch and verifies it in a reversible topology patch sandbox. Taking a gas pipeline reconnection as an example, after a pre-run of the update in the sandbox, it was found that the connection relationship would form a duplicate connection path. The system determined that this did not meet the topology consistency condition, blocked the update, and generated a pending audit record. The update was executed again after the construction log was updated and confirmed, ensuring the correctness of the topology relationship.

[0032] Table 1 Comparison of Automatic Update Effects of Engineering GIS

[0033] As shown in Table 1, the total number of updates increased from 8420 to 8650 after introducing the method of this invention. This indicates that the system, while maintaining stability, did not suppress normal update behavior; on the contrary, it improved the coverage of updates with the support of multi-source data. More importantly, the number of correct updates increased from 7015 to 8290, while the number of incorrect updates significantly decreased from 1405 to 360, resulting in an overall update accuracy rate of 95.84% from 83.31%. This invention, through construction time-series constraint inversion and multi-source evidence overlay mechanism, effectively reduces misjudgments caused by data noise, time asynchrony, or phased construction status, making update decisions more consistent with engineering realities.

[0034] Regarding topology security, the number of topology errors was 312 before implementation, but dropped to 58 after implementation, a significant decrease. The average number of rollbacks decreased from 18 per day to 4, and the average rollback processing time was shortened from 12.5 minutes to 3.8 minutes. This indicates that by introducing a reversible topology patch sandbox mechanism, sufficient topology consistency verification was completed before the update, and a large number of potential errors were intercepted before writing to the database, reducing subsequent repair costs and improving the stability and controllability of system operation.

[0035] Analysis of the sources of erroneous updates shows that premature updates caused by misjudgments during the construction phase decreased from 420 to 62, and erroneous updates triggered by a single data source decreased from 610 to 95. This indicates that the present invention effectively integrates construction process constraints and a multi-source evidence consistency judgment mechanism in its update triggering logic. By avoiding updates directly driven by a single data source or incomplete evidence, the system can maintain high judgment accuracy in complex engineering environments, improving the reliability and engineering applicability of automatic updates to engineering GIS.

[0036] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An engineering GIS automatic updating method based on multi-source data fusion, characterized in that, include: Acquire multi-source data of the engineering area, perform standardization processing on the multi-source data, and obtain standardized multi-source data; Based on standardized multi-source data, a historical state image of the engineering object is constructed. The current multi-source data and the historical state image are then differentially analyzed on an object-by-object basis to obtain differential segments. These differential segments are then classified and merged to form candidate change objects. For each candidate change object, a construction state evolution chain is constructed. Based on the construction state evolution chain, the theoretical existence state, theoretical geometric state, and theoretical connection state of the candidate change object at the current moment are inverted, and candidate update objects that meet the construction time sequence constraints are selected. For candidate update objects, a multi-source evidence spatiotemporal overlay is constructed, and evidence from different data sources is mapped to a unified spatial grid and time slice to form evidence units. Based on the source type and spatiotemporal overlap of each evidence unit, an evidence overlay distribution is generated to determine the candidate update actions corresponding to geometric update, attribute update and topology update. For candidate update actions, generate reversible topology patches, construct a reversible topology patch sandbox, load the reversible topology patches into the reversible topology patch sandbox, and perform topology consistency verification, engineering rule verification, related object influence propagation verification, and rollback recoverability verification based on the loaded sandbox state; When a reversible topology patch passes verification, it is submitted to the engineering GIS database to complete automatic updates, generating corresponding version records and rollback records. When it fails verification, updates are blocked and a pending review result is output.

2. The engineering GIS automatic updating method based on multi-source data fusion according to claim 1, characterized in that, The multi-source data includes remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction plan data, construction log data, and historical GIS data.

3. The engineering GIS automatic updating method based on multi-source data fusion according to claim 1, characterized in that, The standardization process performed on the multi-source data to obtain standardized multi-source data includes: Coordinate system unification is applied to multi-source data, transforming data from different coordinate references to the same target coordinate system. Time alignment is performed on the acquisition time, recording time, or generation time corresponding to various types of data to form time-series data under a unified time benchmark. Outliers, noisy data, and duplicate data in multi-source data are removed. Data formats from different data sources are converted to form a unified data organization format. Engineering object identifiers, time identifiers, and status identifiers in text data are extracted in a structured manner. Missing fields are marked with null values ​​or filled in with correlations to obtain standardized multi-source data.

4. The automatic update method for engineering GIS based on multi-source data fusion according to claim 1, characterized in that, The process of classifying and merging the difference segments to form candidate change objects includes: Based on standardized multi-source data extraction, the geometric information, attribute information and connection relationship information of each engineering object in the engineering area at historical time are extracted. According to the engineering object identifier, historical GIS data, historical BIM model data, historical remote sensing image data, historical laser point cloud data and historical IoT sensor data are aligned at the object level to construct the historical state mirror of each engineering object. The historical state mirror includes the historical geometric mirror of the object, the historical attribute mirror of the object and the historical connection mirror of the object. Extract the current geometric information, current attribute information, and current connection relationship information corresponding to each engineering object from the multi-source data at the current moment. Compare the current state of the engineering object with the historical state mirror image for each object to generate geometric difference fragments, attribute difference fragments, and connection difference fragments. Based on geometric difference segments, a partitioned occupancy change sequence of engineering objects is constructed. The historical spatial occupancy range of the same engineering object is divided into an internal stable region, a boundary transition region, and an external adjacent region. The occupancy change of multi-source data in the internal stable region, the boundary transition region, and the external adjacent region at the current moment is statistically analyzed to obtain the geometric difference degree. The attribute change chain of the project object is constructed based on the attribute difference fragment. The attribute change field at the current moment is matched with the attribute registration order, attribute association field and attribute source record in the historical state mirror item by item to identify the main attribute change, derived attribute change and isolated attribute change and obtain the attribute difference degree. The connection migration trajectory of the engineering object is constructed based on the connection difference segment. The connection relationship change at the current moment is compared with the connection direction, connection position and connection order in the historical state mirror. The connection establishment, connection disconnection and connection transfer are identified to obtain the connection difference degree. Based on the geometric difference degree, attribute difference degree and connection difference degree, the spatially adjacent and temporally continuous difference segments are merged to form candidate change objects and their corresponding geometric change, attribute change and topological change types.

5. The automatic update method for engineering GIS based on multi-source data fusion according to claim 1, characterized in that, The screening process yields candidate update objects that satisfy the construction timing constraints, including: For candidate change objects, extract the construction plan data, construction log data, component dependencies in the BIM model, IoT sensor data and historical GIS data corresponding to the candidate change objects, and associate them according to a unified object identifier and a unified time sequence to form a construction constraint information set. Based on the construction constraint information set, a construction state evolution chain for candidate change objects is constructed. The construction state evolution chain includes the states of no construction, preparation for construction, construction in progress, construction completion, and acceptance effective. State transition conditions are written between adjacent states. State dependency relationships between candidate change objects and preceding components, related components on the same floor, and related continuation components are established. The process records in the construction log, the state change records in the IoT sensor data, and the update records in the historical GIS data are mapped to the corresponding state segments. Based on the construction state evolution chain, the target construction stage to which the candidate change object belongs at the current moment is determined. When the state records corresponding to multiple data sources are inconsistent, the duration corresponding to each state record is statistically analyzed, and compared according to the predefined state priority order in the construction state evolution chain. Among the state records that meet the minimum duration threshold, the state that is consistent with the state priority order is selected as the current stage state. Based on the target construction stage, the theoretical existence state, theoretical geometric state, and theoretical connection state of the candidate change objects at the current moment are inverted. The theoretical existence state includes not appearing, appearing, partially appearing, waiting to be demolished, and demolished. The theoretical geometric state includes maintaining historical geometry, local extension, local reduction, location migration, and forming a new boundary. The theoretical connection state includes maintaining the original connection, establishing a connection, disconnecting the connection, migrating the connection, and forming a temporary connection. The theoretical existence state, theoretical geometric state, and theoretical connectivity state are compared with the difference results corresponding to the candidate change objects. Candidate change objects that match all three are retained as candidate update objects, and the remaining candidate change objects are marked as objects to be constrained and verified.

6. The automatic update method for engineering GIS based on multi-source data fusion according to claim 1, characterized in that, The process of determining the candidate update actions corresponding to geometric update, attribute update, and topology update includes: Update evidence is extracted from remote sensing image data, laser point cloud data, BIM model data, IoT sensor data, construction log data, and historical GIS data corresponding to candidate update objects. The evidence is then linked and organized according to a unified object identifier, a unified spatial location, and a unified time sequence to form an original evidence set. A spatiotemporal superposition of multi-source evidence is constructed based on the original evidence set. This multi-source evidence spatiotemporal superposition is composed of a combination of spatial hierarchical units and temporal segmentation units, wherein: The spatial layering unit is divided into an internal core layer, a boundary transition layer, and an external influence layer according to the historical spatial boundary of the candidate update object, the boundary position, and the outward adjacent range. The time segmentation unit is divided into pre-stage time period, intra-stage time period and post-stage time period according to the stage switching time of the candidate update object in the construction state evolution chain. Each piece of evidence in the original evidence set is written into the corresponding spatial stratification unit and time segmentation unit, forming an evidence occupancy structure that simultaneously represents the spatial stratification and time stage. For multi-source evidence spatiotemporal superposition, the evidence written in each spatial layer unit and time segment unit is superimposed and organized, and the distribution position of each piece of evidence in different spatial layers, the persistence in different time segments, and the extension between adjacent spatial layers and adjacent time segments are recorded to form the evidence superposition trajectory corresponding to the candidate update object. Based on the evidence superposition trajectory, geometric update evidence bands, attribute update evidence bands, and topological update evidence bands are constructed for candidate update objects respectively; Based on geometric update evidence bands, attribute update evidence bands, and topology update evidence bands, the geometric update actions, attribute update actions, and topology update actions corresponding to the candidate update objects are identified respectively.

7. The automatic update method for engineering GIS based on multi-source data fusion according to claim 1, characterized in that, The process of performing topology consistency verification, engineering rule verification, related object impact propagation verification, and rollback recoverability verification based on the loaded sandbox state includes: For candidate update actions, read the pre-update state of the candidate update object in the historical state mirror and the post-update state in the multi-source data at the current moment, and extract the corresponding geometric state, attribute state and connection relationship state to form a pre-update state set and a post-update state set. A reversible topology patch is generated based on the state set before and after the update. The reversible topology patch includes geometric changes, attribute changes, and topology changes. A rollback record is written for each change. A reversible topology patch sandbox is constructed, comprising a base snapshot area, a patch application area, and a rollback / replay area, wherein: The basic snapshot area stores a local subgraph snapshot of the candidate update objects and their associated objects before the update. The patch application area generates a virtual update state by applying reversible topology patches without writing to the project GIS database; The rollback and replay area performs a rollback operation on the virtual update state and restores it to the local subgraph snapshot. The local subgraph snapshot is centered on the candidate update object and extracts the associated objects and associated connections with a preset association depth, including cross-layer associated elements related to the candidate update action. The reversible topology patch is loaded into the patch application area. After generating the virtual update state, consistency verification is performed. Geometric conflict verification and adjacency relationship verification are performed on associated objects involving spatial adjacency. Influence propagation verification is performed on the virtual update state to identify whether the connection changes of candidate update objects cause changes in the connected domains of associated objects or breakage of association relationships. In the rollback replay area, the virtual update state is rolled back and replayed item by item according to the rollback record. The rolled-back state is compared and verified with the local subgraph snapshot in the basic snapshot area to obtain the rollback recoverability verification result. When both the consistency verification result and the rollback recoverability verification result meet the preset commit conditions, the reversible topology patch is determined as a committable patch; otherwise, the reversible topology patch is determined as an uncommittable patch.

8. The automatic update method for engineering GIS based on multi-source data fusion according to claim 1, characterized in that, The automatic update process, which involves submitting the reversible topology patch to the engineering GIS database upon successful verification, includes: Upon receiving a confirmed committable patch, the geometric change, attribute change, and topological change in the reversible topology patch are mapped to the corresponding object records, attribute fields, and connection relationship tables in the engineering GIS database, respectively, and a patch commit unit is generated. Based on the patch submission unit, update operations are performed in the engineering GIS database. Changes are written item by item in the order of attribute update, geometric update and topology update. During the topology update process, the connection records of related objects that have a connection relationship with the candidate update object are updated synchronously. After the update is completed, a version record and a rollback record are generated. The version record includes the update object identifier, update time, and update content identifier. The rollback record includes the geometric data, attribute data, and connection relationship data corresponding to the state before the update. The version record and the rollback record are then stored together. When a reversible topology patch fails verification, the update write operation is terminated and an update record to be reviewed is generated. The update record to be reviewed includes the candidate update object identifier, the type of failure to pass verification, and the corresponding differential result.

Images