A whole-cycle engineering information evolution management system

Abstract

The application discloses a kind of whole cycle engineering information evolution management systems, comprising: information collection module, for generating whole cycle engineering information collection;State identification module is used to extract the attribute state of different stage engineering object, and generates state evolution record;Evolution diagram module is used to establish state connection edge, and constructs engineering information evolution diagram;Characteristic evolution module is used to generate cross-stage evolution characteristic sequence;Connection correction module is used to construct source anchor area and destination anchor area, and generates evolution relationship correction result;Evolution tracing module is used to generate whole cycle engineering information evolution management result.The application realizes the continuous evolution, dynamic update and historical source accurate tracing of engineering information by constructing engineering information evolution diagram and combining the improvement EvolveGCN model and cross-stage connection relationship correction.

Description

Technical Field

[0001] This invention relates to the field of engineering information management technology, and in particular to a management system for the evolution of engineering information throughout its entire lifecycle. Background Technology

[0002] As the level of digitalization in engineering construction continues to increase, engineering projects generate a large amount of engineering information during the design, construction, and operation and maintenance phases. This includes information on engineering object attributes, design changes, construction status, quality acceptance, equipment status, and operation and maintenance updates. Existing engineering information management methods typically categorize and store this information according to project stage, professional category, or document version, and enable data retrieval through engineering object identifiers, document numbers, and time information. This ensures that different engineering stages form corresponding information records, providing a data foundation for engineering construction management and subsequent operation and maintenance.

[0003] However, engineering objects do not remain in a fixed state throughout their entire lifecycle. Design adjustments, construction changes, object splitting, equipment replacement, and maintenance updates all cause changes in the attribute state of engineering objects. Information about the same engineering object may form inheritance, replacement, differentiation, and invalidation relationships in adjacent stages. Existing technologies mainly focus on the storage and version updates of information at each stage, lacking a continuous description of the cross-stage state inheritance relationships and change directions of engineering objects. When an engineering object undergoes multiple state changes, it is prone to situations such as interrupted state inheritance edges, incorrect correspondence of evolution branches, and inaccurate association between historical states and current states.

[0004] Therefore, existing engineering information management technologies struggle to establish complete information evolution relationships based on the continuous state changes of engineering objects during the design, construction, and operation and maintenance phases. In particular, after cross-phase information is replaced, differentiated, or becomes invalid, it is impossible to accurately restore the interrupted state succession relationship and correct the evolution branches. This results in a lack of reliable tracking paths between the current effective state and historical sources, leading to problems such as discontinuous evolution process, difficulty in accurately tracing historical sources, and insufficient accuracy in restoring the state throughout the entire lifecycle.

[0005] Therefore, how to provide a management system for the evolution of engineering information throughout its entire lifecycle 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 a management system for the evolution of engineering information throughout its entire lifecycle. This invention fully utilizes engineering information evolution diagrams, the improved EvolveGCN model, and cross-stage state continuity analysis technology to associate the states of engineering objects in the design, construction, and operation and maintenance stages. It identifies information inheritance, replacement, differentiation, and failure processes, and corrects interrupted state continuity edges and evolution branches through source anchor areas and destination anchor areas. This achieves continuous evolution of engineering information, determination of current valid states, and traceability of historical sources. It has the advantages of accurate cross-stage information association, continuous evolution relationships, traceable historical sources, and high efficiency in engineering information management.

[0007] A management system for the full lifecycle evolution of engineering information according to an embodiment of the present invention includes:

[0008] The information collection module is used to collect engineering information during the design, construction and operation and maintenance phases, establish cross-phase correspondence based on engineering object identifiers, collect information of each engineering object in chronological order, and generate a full-cycle engineering information set.

[0009] The status recognition module is used to extract the attribute status of each engineering object at different stages based on the full-cycle engineering information set, compare the attribute status changes of adjacent stages, identify information inheritance, replacement, differentiation and failure, and generate status evolution records.

[0010] The evolution map module is used to establish state connection edges for engineering objects with consistent object identification and continuous time based on state evolution records, and connect the states of each engineering object in the stage sequence to construct an engineering information evolution map.

[0011] The feature evolution module is used to divide the engineering information evolution graph into a phase graph snapshot sequence according to the design, construction and operation and maintenance stages, and input it into the improved EvolveGCN model. The improved EvolveGCN model sets an evolution relationship encoding layer before the graph convolution layer, sets a phase inheritance constraint mechanism in the parameter evolution unit, and sets a historical source consistency correction layer before the output layer to generate cross-stage evolution feature sequences.

[0012] The acceptance correction module is used to identify the interruption position of the state acceptance edge and the position of the evolution branch based on the cross-stage evolution feature sequence, extract the state of continuous engineering objects along both sides of the interruption position, construct the source anchor area and the destination anchor area, determine the closed acceptance path based on the object identifier, the direction of state change and the state acceptance relationship, correct the interrupted state acceptance edge and evolution branch, and generate the evolution relationship correction result.

[0013] The evolution traceability module is used to update the engineering information evolution diagram based on the evolution relationship correction results, track the process of information inheritance, replacement, differentiation and failure along the state succession edge, determine the current effective state and historical source, and generate the full-cycle engineering information evolution management results.

[0014] Optionally, the project information includes project object identifier, project stage identifier, project object attribute information, design change information, construction status information, quality acceptance information, equipment status information, operation and maintenance update information, and information generation time.

[0015] Optionally, the step of establishing cross-stage correspondences based on project object identifiers, collecting information on each project object in chronological order, and generating a full-cycle project information set includes:

[0016] Read the engineering object identifiers from the engineering information in the design, construction and operation and maintenance phases, classify the engineering information in each phase according to the engineering object identifiers, and generate engineering object information groups;

[0017] Read the project stage identifiers from each project object information group, and establish object correspondences for project information that belongs to the same project object but is distributed across different project stages;

[0018] Based on the object correspondence, the information generation time corresponding to each project information is read, and the project information of the same project object is sorted according to the information generation time to generate an object time series;

[0019] Based on the time sequence of objects, the engineering information of the design phase, construction phase and operation and maintenance phase are connected in the order of the engineering phase to generate a cross-phase engineering object information sequence;

[0020] The information sequence of engineering objects across each stage is summarized, and the corresponding engineering object identifier, engineering stage identifier, and information generation time are associated to generate a full-cycle engineering information set.

[0021] Optionally, generating the state evolution record includes:

[0022] Read the project stage identifier and project object attribute information corresponding to each project object in the full life cycle project information set, extract the attribute status of each project object in the design, construction and operation and maintenance stages according to the project stage order, and generate a stage attribute status sequence;

[0023] Based on the stage attribute state sequence, the attribute states of adjacent project stages of the same project object are matched item by item, and attribute content retention, attribute content change, attribute content addition and attribute content disappearance are identified to generate attribute change records of adjacent stages.

[0024] Based on the attribute change records of adjacent stages, the property content is kept in the corresponding project object state as information inheritance, and the property content of the preceding property is replaced by the corresponding project object state as information replacement, and the inheritance state record and the replacement state record are generated.

[0025] Based on the attribute change records of adjacent stages, the branching relationship of multiple subsequent project object states corresponding to the same previous project object state is identified, and the termination relationship of the previous project object state having no corresponding attribute state in the subsequent project stage is identified, generating differentiation state records and failure state records.

[0026] Based on the project object identifier and the project stage sequence, the inherited state record, replacement state record, differentiation state record and failure state record are associated with the corresponding stage attribute state sequence to generate a state evolution record.

[0027] Optionally, the construction project information evolution diagram includes:

[0028] Read the project object identifier, project stage identifier, and information generation time from the state evolution record, arrange the states of project objects with the same project object identifier according to the project stage order and information generation time, and generate a candidate state succession sequence.

[0029] Based on the candidate state succession sequence, the project object identifier and information generation time of adjacent project object states are matched in turn, and the project object states with the same project object identifier and the subsequent state succeeding the previous state are determined as continuous succession state pairs.

[0030] Read the corresponding inheritance status record, replacement status record, differentiation status record and failure status record of the continuous acceptance status pair, set the status of each project object as a status node, and determine the acceptance direction and acceptance type between status nodes according to the corresponding information inheritance, replacement, differentiation and failure.

[0031] Based on the direction of acceptance, connect the corresponding state nodes of the continuous acceptance state pairs, establish state acceptance edges, and associate the project object identifier, project stage identifier, information generation time and acceptance type with the corresponding state nodes and state acceptance edges to generate a project object evolution sub-graph.

[0032] The evolution sub-graphs of each engineering object are summarized in the order of design, construction and operation and maintenance. The evolution sub-graphs of engineering objects with branch relationships are connected according to the differentiation state records. The termination state nodes are marked according to the failure state records to construct the engineering information evolution graph.

[0033] Optionally, generating the cross-stage evolution feature sequence includes:

[0034] Read the engineering stage identifier and information generation time of each state node in the engineering information evolution diagram, divide the state nodes and state connection edges according to the design, construction and operation and maintenance stages, and arrange them according to the stage order to generate a stage diagram snapshot sequence.

[0035] Input the snapshot sequence of the stage graph into the evolution relationship encoding layer, read the inheritance, replacement, differentiation and failure acceptance types corresponding to each state acceptance edge, associate and encode the acceptance type with the attribute state of the corresponding state node, and generate stage evolution relationship features;

[0036] The stage evolution relationship features are input into the graph convolutional layer, and the stage evolution relationship features corresponding to the attribute states of each state node and the adjacent state receiving edges are aggregated to generate the graph convolutional features of the state of the engineering object at each stage.

[0037] Input the graph convolution features and stage evolution relationship features of each stage into the stage bearing constraint parameter evolution unit. Based on the state bearing relationship corresponding to inheritance, replacement, differentiation and failure, determine the parameter retention, parameter update, parameter branch transmission and parameter termination transmission states respectively. Update the graph convolution parameters in the stage order to generate cross-stage parameter evolution results.

[0038] The graph convolution features of the corresponding stage are updated based on the cross-stage parameter evolution results. The updated graph convolution features are input into the historical source consistency correction layer. The engineering object identifier, engineering stage identifier and acceptance type are matched in reverse along the state acceptance edge to correct the state node features with inconsistent historical sources and generate the corrected evolution features.

[0039] The corrected evolution characteristics of each state node are summarized according to the engineering stage sequence, and the evolution states of the corresponding state receiving edges are associated to generate a cross-stage evolution characteristic sequence.

[0040] Optionally, the generated evolutionary relationship correction result includes:

[0041] Read the corrected evolution features of each state node and the evolution state of the state connection edge in the cross-stage evolution feature sequence, detect the continuity of the connection between adjacent state nodes according to the engineering stage sequence, determine the position where the state connection edge is missing as the state connection edge interruption position, and determine the position where the same preceding state node connects multiple subsequent state nodes as the evolution branch position.

[0042] Using the interruption position of the state-bearing edge as the boundary, continuously extract the states of engineering objects with the bearing relationship in the preceding engineering stage and the subsequent engineering stage respectively. The preceding continuous engineering object states form the source anchor area, and the subsequent continuous engineering object states form the destination anchor area.

[0043] Read the project object identifier, project stage identifier, attribute status and acceptance type corresponding to the status of each project object in the source anchor area and the destination anchor area, and determine the candidate acceptance relationship between the end status of the source anchor area and the beginning status of the destination anchor area according to the project stage order.

[0044] Based on the candidate acceptance relationship, the engineering object identifiers of the source anchor area and the destination anchor area are matched, and the state change direction is determined according to the inheritance, replacement, differentiation and failure changes of adjacent attribute states. The candidate acceptance relationship corresponding to the object identifier and with continuous state change direction is determined as a closed acceptance path.

[0045] Based on the closed connection path, restore the state connection edge corresponding to the interrupted position of the state connection edge, and check the preceding state node and each subsequent state node corresponding to the position of the evolution branch according to the differentiation state record, and adjust the evolution branches that are inconsistent with the differentiation state record.

[0046] The restored state acceptance edge, the adjusted evolution branch, and the corresponding acceptance type are associated with the state nodes in the engineering information evolution diagram to generate the evolution relationship correction result.

[0047] Optionally, generating the full-cycle engineering information evolution management results includes:

[0048] Read the restored state acceptance edge, adjusted evolution branch and corresponding acceptance type from the evolution relationship correction result, write the corresponding relationship into the engineering information evolution diagram, and generate the corrected engineering information evolution diagram;

[0049] Based on the correction engineering information evolution diagram, starting from the current stage state node of each engineering object, the corresponding inheritance, replacement, differentiation and failure relationships are traced in reverse order of engineering stages along the state connection edge to generate a historical state tracing sequence.

[0050] Based on the historical state tracking sequence, identify the state retention, state replacement, state branching and state termination processes of each engineering object in the continuous engineering stage, and associate them with the corresponding engineering object identifier, engineering stage identifier and information generation time to generate the engineering information evolution trajectory.

[0051] Based on the evolution trajectory of engineering information, the state of the end engineering object that has not been replaced by the subsequent one and has not been marked as invalid is determined as the current valid state, and the historical source corresponding to the current valid state is determined in reverse along the corresponding state-bearing edge;

[0052] It summarizes the current effective status, historical source, and evolution trajectory of each project object, establishes query relationships according to the project object identifier, and generates full-cycle project information evolution management results.

[0053] The beneficial effects of this invention are:

[0054] This invention establishes cross-stage correspondences of engineering information in the design, construction, and operation and maintenance phases by identifying engineering object identifiers. It also identifies information inheritance, replacement, differentiation, and failure based on changes in the attribute status of engineering objects. This transforms the discrete engineering information traditionally stored separately according to stages and files into an engineering information evolution diagram with continuous state inheritance relationships. This allows information generated by the same engineering object in different life cycle stages to maintain sequential correlation, improving the continuity of the organization of engineering information throughout the entire life cycle and the accuracy of cross-stage correspondence.

[0055] This invention processes the snapshot sequence of the stage graph by improving the EvolveGCN model, uses the evolution relationship encoding layer to represent the different state succession relationships, maintains the evolutionary association of the state of the engineering object between consecutive stages through the stage succession constraint mechanism, and uses the historical source consistency correction layer to correct the historical sources across stages. In this way, it extracts cross-stage evolution feature sequences that can reflect the state changes of the engineering object and the evolution process of the state succession edge, reduces the information succession confusion caused by multiple design changes, construction adjustments and operation and maintenance updates, and improves the accuracy of identifying the evolution relationship of engineering information.

[0056] This invention constructs source and destination anchor areas for interrupted state acceptance edges and evolution branches. It determines closed acceptance paths through object identification, state change direction, and state acceptance relationships. It corrects interrupted state acceptance edges and evolution branches, and tracks the process of information inheritance, replacement, differentiation, and failure based on the corrected engineering information evolution diagram. This enables the determination of the current valid state and the tracing of historical sources, allowing the continuous reconstruction of the historical evolution process of engineering objects and improving the reliability of full-cycle engineering information tracing and the efficiency of engineering information management. Attached Figure Description

[0057] 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:

[0058] Figure 1 This is a schematic diagram of the structure of a management system for the full lifecycle engineering information evolution proposed in this invention;

[0059] Figure 2 This is a schematic diagram of the structure of the improved EvolveGCN model, a management system for the full lifecycle engineering information evolution proposed in this invention. Detailed Implementation

[0060] 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.

[0061] refer to Figures 1-2A management system for the full lifecycle evolution of engineering information, comprising:

[0062] The information collection module is used to collect engineering information during the design, construction and operation and maintenance phases, establish cross-phase correspondence based on engineering object identifiers, collect information of each engineering object in chronological order, and generate a full-cycle engineering information set.

[0063] The status recognition module is used to extract the attribute status of each engineering object at different stages based on the full-cycle engineering information set, compare the attribute status changes of adjacent stages, identify information inheritance, replacement, differentiation and failure, and generate status evolution records.

[0064] The evolution map module is used to establish state connection edges for engineering objects with consistent object identification and continuous time based on state evolution records, and connect the states of each engineering object in the stage sequence to construct an engineering information evolution map.

[0065] The feature evolution module is used to divide the engineering information evolution graph into a phase graph snapshot sequence according to the design, construction and operation and maintenance stages, and input it into the improved EvolveGCN model. The improved EvolveGCN model sets an evolution relationship encoding layer before the graph convolution layer, sets a phase inheritance constraint mechanism in the parameter evolution unit, and sets a historical source consistency correction layer before the output layer to generate cross-stage evolution feature sequences.

[0066] The acceptance correction module is used to identify the interruption position of the state acceptance edge and the position of the evolution branch based on the cross-stage evolution feature sequence, extract the state of continuous engineering objects along both sides of the interruption position, construct the source anchor area and the destination anchor area, determine the closed acceptance path based on the object identifier, the direction of state change and the state acceptance relationship, correct the interrupted state acceptance edge and evolution branch, and generate the evolution relationship correction result.

[0067] The evolution traceability module is used to update the engineering information evolution diagram based on the evolution relationship correction results, track the process of information inheritance, replacement, differentiation and failure along the state succession edge, determine the current effective state and historical source, and generate the full-cycle engineering information evolution management results.

[0068] In this embodiment, the project information includes project object identifier, project stage identifier, project object attribute information, design change information, construction status information, quality acceptance information, equipment status information, operation and maintenance update information, and information generation time.

[0069] In this embodiment, the step of establishing cross-stage correspondences based on engineering object identifiers, collecting information on each engineering object in chronological order, and generating a full-cycle engineering information set includes:

[0070] Read the project object identifiers from the project information in the design, construction, and operation and maintenance phases, classify the project information for each phase according to the project object identifiers, and generate project object information groups. Specifically, generating project object information groups involves:

[0071] Read the project object identifiers corresponding to the project information at each stage, standardize the format of the project object identifiers, identify project information with the same project object identifiers, and classify them into the corresponding design stage information, construction stage information, and operation and maintenance stage information according to the project stage identifiers. Then, summarize the project information at each stage that belongs to the same project object identifier to generate the project object information group corresponding to each project object.

[0072] Read the project stage identifiers from each project object information group, and establish object correspondences for project information that belongs to the same project object but is distributed across different project stages;

[0073] Based on the object correspondence, the information generation time corresponding to each project information is read, and the project information of the same project object is sorted according to the information generation time to generate an object time series;

[0074] Based on the time sequence of objects, the engineering information of the design phase, construction phase and operation and maintenance phase are connected in the order of the engineering phase to generate a cross-phase engineering object information sequence;

[0075] The information sequence of engineering objects across each stage is summarized, and the corresponding engineering object identifier, engineering stage identifier, and information generation time are associated to generate a full-cycle engineering information set.

[0076] In this embodiment, the generation of the state evolution record includes:

[0077] Read the project stage identifier and project object attribute information corresponding to each project object in the full life cycle project information set, extract the attribute status of each project object in the design, construction and operation and maintenance stages according to the project stage order, and generate a stage attribute status sequence;

[0078] Based on the stage attribute state sequence, the attribute states of adjacent project stages of the same project object are matched item by item, and attribute content retention, attribute content change, attribute content addition and attribute content disappearance are identified to generate attribute change records of adjacent stages.

[0079] Based on the attribute change records of adjacent stages, the property content is kept in the corresponding project object state as information inheritance, and the property content of the preceding property is replaced by the corresponding project object state as information replacement, and the inheritance state record and the replacement state record are generated.

[0080] Based on the attribute change records of adjacent stages, the branching relationship of multiple subsequent project object states corresponding to the same previous project object state is identified, and the termination relationship of the previous project object state having no corresponding attribute state in the subsequent project stage is identified, generating differentiation state records and failure state records.

[0081] Based on the project object identifier and the project stage sequence, the inherited state record, replacement state record, differentiation state record and failure state record are associated with the corresponding stage attribute state sequence to generate a state evolution record.

[0082] In this embodiment, the construction project information evolution diagram includes:

[0083] Read the project object identifier, project stage identifier, and information generation time from the state evolution record, arrange the states of project objects with the same project object identifier according to the project stage order and information generation time, and generate a candidate state succession sequence.

[0084] Based on the candidate state succession sequence, the project object identifier and information generation time of adjacent project object states are matched in turn, and the project object states with the same project object identifier and the subsequent state succeeding the previous state are determined as continuous succession state pairs.

[0085] Read the corresponding inheritance status record, replacement status record, differentiation status record and failure status record of the continuous acceptance status pair, set the status of each project object as a status node, and determine the acceptance direction and acceptance type between status nodes according to the corresponding information inheritance, replacement, differentiation and failure.

[0086] Based on the direction of acceptance, connect the corresponding state nodes of the continuous acceptance states, establish state acceptance edges, and associate the project object identifier, project stage identifier, information generation time, and acceptance type with the corresponding state nodes and state acceptance edges to generate a project object evolution subgraph. Specifically, the generation of the project object evolution subgraph is as follows:

[0087] Using the states of each engineering object in a continuous state pair as state nodes, connect the corresponding state nodes according to the direction of acceptance from the previous state to the next state, write the engineering object identifier and engineering stage identifier into the state nodes, use the information generation time as the temporal attribute of the state nodes, write the acceptance type into the corresponding state acceptance edge, and mark the relationship category of each state acceptance edge according to inheritance, replacement, differentiation and failure, summarize the state nodes and state acceptance edges corresponding to the same engineering object, and generate an engineering object evolution subgraph.

[0088] The evolution sub-graphs of each engineering object are summarized according to the design, construction, and operation and maintenance phases. The evolution sub-graphs of engineering objects with branching relationships are connected based on the differentiation state records. Termination state nodes are marked based on the failure state records to construct an engineering information evolution graph. Specifically, the construction of the engineering information evolution graph involves:

[0089] Arrange the evolution sub-graphs of each engineering object in the order of design, construction, and operation and maintenance phases. Connect the state nodes with continuous inheritance relationships in adjacent phases according to the corresponding state inheritance edges. Read the previous engineering object state and the corresponding multiple subsequent engineering object states in the differentiated state records. Use the previous state node as the branch starting point and connect it to each subsequent state node to form an evolution branch. Read the failed engineering object state in the failure state record, mark the corresponding state node as the termination state node, and stop the establishment of state inheritance edges for the subsequent engineering phases for that state node. Summarize the evolution sub-graphs, evolution branches, and termination state nodes of each engineering object to construct the engineering information evolution graph.

[0090] In this embodiment, generating the cross-stage evolution feature sequence includes:

[0091] Read the engineering stage identifier and information generation time of each state node in the engineering information evolution diagram, divide the state nodes and state connection edges according to the design, construction and operation and maintenance stages, and arrange them according to the stage order to generate a stage diagram snapshot sequence.

[0092] The snapshot sequence of the stage graph is input into the evolution relationship encoding layer. The inheritance, replacement, differentiation, and failure acceptance types corresponding to the edges of each state are read. The acceptance types are associated with the attribute states of the corresponding state nodes and encoded to generate stage evolution relationship features. Specifically, the generation of stage evolution relationship features is as follows:

[0093] Read the inheritance type of each state-accepting edge in the phase diagram snapshot sequence and the attribute status of the corresponding preceding and subsequent state nodes. Map inheritance, replacement, differentiation, and failure to different relation category identifiers. Compare the retention, modification, addition, and disappearance of corresponding attribute items of preceding and subsequent state nodes to generate attribute change descriptions. Combine the relation category identifiers, attribute change descriptions, project phase identifiers, and state acceptance directions, and write them into the phase diagram snapshot according to the corresponding state-accepting edges to generate phase evolution relation features.

[0094] The stage evolution relationship features are input into the graph convolutional layer, and the stage evolution relationship features corresponding to the attribute states of each state node and the adjacent state receiving edges are aggregated to generate the graph convolutional features of the state of the engineering object at each stage.

[0095] The graph convolution features and stage evolution relationship features of each stage are input into the stage inheritance constraint parameter evolution unit. Based on the state inheritance relationships corresponding to inheritance, replacement, differentiation, and failure, the parameter retention, parameter update, parameter branching, and parameter termination states are determined respectively. The graph convolution parameters are updated according to the stage sequence to generate cross-stage parameter evolution results. Specifically, the generation of cross-stage parameter evolution results is as follows:

[0096] The system reads the graph convolution features of the current stage, the graph convolution parameters of the previous stage, and the stage evolution relationship features. Based on the inheritance type corresponding to the state inheritance edge, it determines the parameter transfer state. When the inheritance type is inheritance, it retains the graph convolution parameters of the corresponding state node from the previous stage and transfers them to the current stage. When the inheritance type is replacement, it updates the transferred graph convolution parameters based on the graph convolution features of the corresponding state node in the current stage. When the inheritance type is differentiation, it transfers the graph convolution parameters of the corresponding state node from the previous stage to each branch state node and updates them based on the graph convolution features of each branch. When the inheritance type is failure, it terminates the transfer of the corresponding graph convolution parameters to the next stage. Parameter transfer and updating are executed sequentially according to the design, construction, and operation and maintenance stages, generating cross-stage parameter evolution results.

[0097] The graph convolution features of the corresponding stage are updated based on the cross-stage parameter evolution results. The updated graph convolution features are input into the historical source consistency correction layer. The engineering object identifier, engineering stage identifier and acceptance type are matched in reverse along the state acceptance edge to correct the state node features with inconsistent historical sources and generate the corrected evolution features.

[0098] The corrected evolution characteristics of each state node are summarized according to the engineering stage sequence, and the evolution states of the corresponding state receiving edges are associated to generate a cross-stage evolution characteristic sequence.

[0099] This invention improves the structure of the traditional EvolveGCN model by adding an evolutionary relationship encoding layer before the graph convolutional layer. This layer associates the inheritance, replacement, differentiation, and failure succession types formed by engineering objects during the design, construction, and operation and maintenance phases with the attribute states of corresponding state nodes, enabling the model to obtain clear evolutionary semantics between state nodes before performing graph convolutional feature extraction. Compared to traditional EvolveGCN, which mainly relies on dynamic graph structure changes for feature propagation, this structure can distinguish the nature of engineering information changes expressed by different state succession edges, enhancing the model's ability to represent cross-stage engineering object state changes and state succession relationships.

[0100] To address the issue of different evolutionary patterns among various inheritance relationships during cross-stage information transfer in engineering projects, this invention establishes a stage inheritance constraint mechanism within the parameter evolution unit. Based on inheritance, replacement, differentiation, and failure, it performs parameter retention, parameter update, parameter branching, and parameter termination, respectively. By ensuring that the cross-stage evolution process of graph convolution parameters corresponds to the actual state inheritance process of the engineering object, it reduces the propagation of failed states, the excessive influence of the original state on replaced states, and the information aliasing between differentiated states. This allows the model to continuously retain effective historical information and respond promptly to changes in the state of the engineering object, improving the continuity and accuracy of cross-stage evolution feature extraction.

[0101] Furthermore, this invention sets up a historical source consistency correction layer before the output layer. It performs reverse matching of the engineering object identifier, engineering stage identifier, and acceptance type along the state acceptance edge, verifies the historical source of the state node features generated after cross-stage parameter evolution, and corrects state node features with inconsistent historical sources. This forms a continuous processing procedure of "evolutionary relationship encoding—acceptance constraint parameter evolution—historical source reverse correction," enabling the improved EvolveGCN model not only to describe the dynamic changes of engineering information with each stage but also to maintain the evolutionary consistency between the current engineering object state and its historical source. This improves the accuracy of identifying state acceptance edge interruptions and evolutionary branches, providing reliable cross-stage evolutionary features for determining closed acceptance paths and tracing the historical source of engineering information.

[0102] In this embodiment, the generation of evolutionary relationship correction results includes:

[0103] Read the corrected evolution features of each state node and the evolution state of the state connection edge in the cross-stage evolution feature sequence, detect the continuity of the connection between adjacent state nodes according to the engineering stage sequence, determine the position where the state connection edge is missing as the state connection edge interruption position, and determine the position where the same preceding state node connects multiple subsequent state nodes as the evolution branch position.

[0104] Using the break point of the state continuation edge as the boundary, continuously extract the states of engineering objects with continuation relationships in both the preceding and subsequent engineering stages. Form source anchor regions for the preceding continuous engineering object states and destination anchor regions for the subsequent continuous engineering object states. Specifically, forming source anchor regions for the preceding continuous engineering object states and destination anchor regions for the subsequent continuous engineering object states is as follows:

[0105] Starting from the preceding state node corresponding to the interruption position of the state acceptance edge, read the state of the project object level by level in reverse order of the project stage along the state acceptance edge. When the object identifiers of adjacent project object states correspond and the state change direction is continuous, the corresponding project object states are sequentially included in the source anchor area until the first project stage is reached or a state node with a discontinuous acceptance relationship appears. Starting from the following state node corresponding to the interruption position of the state acceptance edge, read the state of the project object level by level in order of the project stage along the state acceptance edge. When the object identifiers of adjacent project object states correspond and the state change direction is continuous, the corresponding project object states are sequentially included in the destination anchor area until the last project stage is reached or a state node with a discontinuous acceptance relationship appears. Arrange the included project object states according to the order of the project stages to generate the source anchor area and the destination anchor area.

[0106] Read the project object identifier, project stage identifier, attribute status and acceptance type corresponding to the status of each project object in the source anchor area and the destination anchor area, and determine the candidate acceptance relationship between the end status of the source anchor area and the beginning status of the destination anchor area according to the project stage order.

[0107] Based on the candidate acceptance relationship, the engineering object identifiers of the source anchor area and the destination anchor area are matched, and the state change direction is determined according to the inheritance, replacement, differentiation and failure changes of adjacent attribute states. The candidate acceptance relationship corresponding to the object identifier and with continuous state change direction is determined as a closed acceptance path.

[0108] Based on the closed connection path, restore the state connection edge corresponding to the interrupted position of the state connection edge, and check the preceding state node and each subsequent state node corresponding to the position of the evolution branch according to the differentiation state record, and adjust the evolution branches that are inconsistent with the differentiation state record.

[0109] The restored state acceptance edge, the adjusted evolution branch, and the corresponding acceptance type are associated with the state nodes in the engineering information evolution diagram to generate the evolution relationship correction result.

[0110] In this embodiment, generating the full-cycle engineering information evolution management result includes:

[0111] Read the restored state acceptance edge, adjusted evolution branch and corresponding acceptance type from the evolution relationship correction result, write the corresponding relationship into the engineering information evolution diagram, and generate the corrected engineering information evolution diagram;

[0112] Based on the correction engineering information evolution diagram, starting from the current stage state node of each engineering object, the corresponding inheritance, replacement, differentiation and failure relationships are traced in reverse order of engineering stages along the state connection edge to generate a historical state tracing sequence.

[0113] Based on the historical state tracking sequence, identify the state retention, state replacement, state branching and state termination processes of each engineering object in the continuous engineering stage, and associate them with the corresponding engineering object identifier, engineering stage identifier and information generation time to generate the engineering information evolution trajectory.

[0114] Based on the evolution trajectory of engineering information, the state of the end engineering object that has not been replaced by the subsequent one and has not been marked as invalid is determined as the current valid state, and the historical source corresponding to the current valid state is determined in reverse along the corresponding state-bearing edge;

[0115] It summarizes the current effective status, historical source, and evolution trajectory of each project object, establishes query relationships according to the project object identifier, and generates full-cycle project information evolution management results.

[0116] Example 1: In a comprehensive engineering information management scenario entering the operation and maintenance phase, the system receives a total of 42,860 engineering information entries generated across the design, construction, and operation and maintenance phases, including 12,640 entries from the design phase, 18,120 from the construction phase, and 12,100 from the operation and maintenance phase. There are a total of 4,362 engineering objects, including 1,486 structural components, 812 electromechanical equipment, 936 water supply, drainage, and fire protection pipelines, 574 ventilation objects, and 554 power distribution objects. The original information contains instances of object number adjustments, design attribute modifications, construction replacements, object splitting, and operation and maintenance shutdowns. After manual pre-verification, 96 status transition interruptions, 41 evolution branches, and 26 erroneous transition candidates were identified to verify the method of this invention.

[0117] After the project information enters the system, it is first aggregated across stages based on the project object identifier. Taking the beam object in the internal location "Area A - Structural Unit 12" as an example, the design stage corresponds to identifier JG-012, with an initial cross-section of 300×600mm, which was later changed to 300×650mm; the construction stage still corresponds to JG-012, and the cross-section in the acceptance record is 300×650mm; in the operation and maintenance stage, the object's status is recorded as "in service". The system connects these four records according to the order of information generation, forming an object time series of length 4. After processing all 42,860 pieces of information, 4,362 sets of project object information sequences are formed, of which 4,178 sets can be directly established with cross-stage correspondence, and 184 sets have number changes or object splits, requiring subsequent status identification.

[0118] The system then extracts the attribute states of each engineering object. It compares the attribute changes of adjacent stages for each of the 4362 sequences, resulting in 13874 adjacent state pairs. Among these, 9216 pairs are attributed to information inheritance, 2384 to information replacement, 1216 to information differentiation, and 1058 to information failure. Taking equipment object SB-087 as an example, its rated air volume was 28000 m³ / h and power 18.5 kW during the design stage. During the construction stage, due to equipment procurement adjustments, this changed to 30000 m³ / h and 20.0 kW. The system identifies this state pair as a replacement. During the operation and maintenance stage, the main body of the equipment remains unchanged, only the filter components are replaced, and the main attributes remain consistent; the corresponding state is identified as inheritance. For pipeline object GX-116, it was a single DN150 main pipe during the design stage, and during the construction stage, it was split into two branch pipes, DN100 and DN80. The system records one preceding state corresponding to two subsequent states, identifying this as differentiation.

[0119] Based on the aforementioned state evolution records, the system constructs an engineering information evolution diagram. A total of 14,692 state nodes and 13,874 state connection edges are generated for 4,362 engineering objects, including 9,216 inheritance edges, 2,384 replacement edges, 1,216 differentiation edges, and 1,058 failure edges. For states with the same engineering object identifier and continuous information time, connection edges are directly established; for records where the object number changes, they are first retained as candidate nodes. For example, object SB-154 was numbered SB-154 in the design phase and changed to ME-154 after construction detailing, but the installation location, equipment category, and main attributes are continuous; the system sets the two states as candidate connection relationships. After the diagram is completed, 4,362, 4,487, and 4,410 valid state nodes are formed in the design, construction, and operation and maintenance phases, respectively.

[0120] The system divides the engineering information evolution diagram into three sets of snapshots according to stages and inputs them into the improved EvolveGCN model. The training data consists of 18,600 labeled cross-stage state samples, including 11,120 inheritance samples, 3,120 replacement samples, 2,440 differentiation samples, and 1,920 failure samples, divided into 13,020 training samples, 2,790 validation samples, and 2,790 test samples. In the training samples, one input is "JG-012, 300×600mm→300×650mm", labeled as replacement; another input is "GX-116, DN150 main pipe→DN100 branch pipe+DN80 branch pipe", labeled as differentiation; and another input is "SB-203, in service→removal", labeled as failure.

[0121] During the construction of training samples, the system uniformly retains the project object identifier, project stage identifier, information generation order, location code, main attributes, and succession label for each state pair. Of the 128-dimensional relationship features, project object attributes account for 64 dimensions, stage and time sequence for 16 dimensions, succession relationship code for 32 dimensions, and location and object category for 16 dimensions. In the validation set, 120 sets of samples with varying numbers were randomly selected. After manual verification, 113 sets were able to recover the true preceding source, used to check the model's adaptability to scenarios where the numbering changes but the attributes remain continuous.

[0122] After processing by the evolutionary relation encoding layer, each state node forms a 128-dimensional relation feature. The graph convolutional layer outputs 128-dimensional state features. The stage-based constraint mechanism performs parameter retention, update, branch propagation, and termination propagation according to inheritance, replacement, differentiation, and failure, respectively. The test accuracy was 87.1% at the 20th round of training, 92.9% at the 40th round, 95.6% at the 60th round, and reached 96.7% at the 80th round. Ultimately, 2698 out of 2790 test samples were correctly identified.

[0123] During the historical source consistency correction process, the system verifies the object identifier, stage sequence, and acceptance type along the state acceptance edge. Of the 26 erroneous acceptance candidates in the original diagram, 23 were eliminated at this stage. For example, ME-154 is similar to equipment states in two design stages, but the installation unit of one object is different. After reverse tracing, a continuous source chain cannot be formed with the construction state, and this acceptance edge is deleted. After processing, 14,692 cross-stage evolution feature sequences are generated, including 14,231 high-confidence continuous states and 461 states to be corrected.

[0124] The system continues to identify state continuity interruption locations and evolution branch locations from the cross-stage evolution feature sequence. 94 out of 96 pre-set interruptions were identified, and 40 out of 41 pre-set branches were identified. For each interruption location, the continuous engineering object state was extracted forward and backward. The source anchor area contained an average of 3.6 state nodes, and the destination anchor area contained an average of 3.3 state nodes. Taking a certain equipment object as an example, the continuous record before the interruption was "Initial Design → Design Replacement → Construction and Installation," and after the interruption, it was "Operation and Maintenance Inspection → Operation and Maintenance Replacement." The system forms a closed continuity path based on object identifier mapping, attribute change direction, and inheritance relationship, reconnecting the construction and installation state with the operation and maintenance inspection state. A total of 118 candidate closed paths were formed for all interruption locations, and 91 were ultimately confirmed, of which 90 were consistent with the manual verification results.

[0125] After completing the acceptance correction, the system updated the engineering information evolution diagram, restored 90 status acceptance edges, adjusted 39 evolution branches, and re-tracked the inheritance, replacement, differentiation, and failure processes. The system randomly selected 300 current objects from the operation and maintenance phase for historical backtracking, including 176 ordinary inheritance objects, 58 objects that underwent replacement, 39 objects that underwent differentiation, and 27 objects containing failure histories. Ultimately, 292 objects were able to accurately reconstruct their design origin, construction acceptance status, and current operation and maintenance status, with a historical origin tracing accuracy rate of 97.3%; the current valid status of 296 objects was correctly determined, with an accuracy rate of 98.7%.

[0126] In the comparative experiment, using the same batch of 42,860 engineering information records, the method was compared with the traditional method of searching by stage documents, object numbers, and version records. The traditional method identified 71 out of 96 state continuity interruptions, a success rate of 74.0%, while the present invention identified 94, a success rate of 97.9%. The traditional method correctly recovered 64 interruption relationships, a recovery rate of 66.7%, while the present invention correctly recovered 90, a recovery rate of 93.8%. The traditional method correctly identified 27 out of 41 evolutionary branches, an accuracy rate of 65.9%, while the present invention correctly identified 39, an accuracy rate of 95.1%. The traditional method's accuracy rate for tracing historical sources was 75.7%, while the present invention's was 97.3%. The traditional method's accuracy rate for identifying the current valid state was 80.0%, while the present invention's was 98.7%. The average tracing time for a single complex object was 68.4 seconds for the traditional method, and 8.1 seconds for the present invention.

[0127] 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. A management system for the full lifecycle evolution of engineering information, characterized in that, include: The information collection module is used to collect engineering information during the design, construction and operation and maintenance phases, establish cross-phase correspondence based on engineering object identifiers, collect information of each engineering object in chronological order, and generate a full-cycle engineering information set. The status recognition module is used to extract the attribute status of each engineering object at different stages based on the full-cycle engineering information set, compare the attribute status changes of adjacent stages, identify information inheritance, replacement, differentiation and failure, and generate status evolution records. The evolution map module is used to establish state connection edges for engineering objects with consistent object identification and continuous time based on state evolution records, and connect the states of each engineering object in the stage sequence to construct an engineering information evolution map. The feature evolution module is used to divide the engineering information evolution graph into a phase graph snapshot sequence according to the design, construction and operation and maintenance stages, and input it into the improved EvolveGCN model. The improved EvolveGCN model sets an evolution relationship encoding layer before the graph convolution layer, sets a phase inheritance constraint mechanism in the parameter evolution unit, and sets a historical source consistency correction layer before the output layer to generate cross-stage evolution feature sequences. The acceptance correction module is used to identify the interruption position of the state acceptance edge and the position of the evolution branch based on the cross-stage evolution feature sequence, extract the state of continuous engineering objects along both sides of the interruption position, construct the source anchor area and the destination anchor area, determine the closed acceptance path based on the object identifier, the direction of state change and the state acceptance relationship, correct the interrupted state acceptance edge and evolution branch, and generate the evolution relationship correction result. The evolution traceability module is used to update the engineering information evolution diagram based on the evolution relationship correction results, track the process of information inheritance, replacement, differentiation and failure along the state succession edge, determine the current effective state and historical source, and generate the full-cycle engineering information evolution management results.

2. The management system for the full-cycle engineering information evolution according to claim 1, characterized in that, The project information includes project object identifier, project stage identifier, project object attribute information, design change information, construction status information, quality acceptance information, equipment status information, operation and maintenance update information, and information generation time.

3. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The process involves establishing cross-stage correspondences based on project object identifiers, collecting information on each project object in chronological order, and generating a full-cycle project information set, including: Read the engineering object identifiers from the engineering information in the design, construction and operation and maintenance phases, classify the engineering information in each phase according to the engineering object identifiers, and generate engineering object information groups; Read the project stage identifiers from each project object information group, and establish object correspondences for project information that belongs to the same project object but is distributed across different project stages; Based on the object correspondence, the information generation time corresponding to each project information is read, and the project information of the same project object is sorted according to the information generation time to generate an object time series; Based on the time sequence of objects, the engineering information of the design phase, construction phase and operation and maintenance phase are connected in the order of the engineering phase to generate a cross-phase engineering object information sequence; The information sequence of engineering objects across each stage is summarized, and the corresponding engineering object identifier, engineering stage identifier, and information generation time are associated to generate a full-cycle engineering information set.

4. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The generation of the state evolution record includes: Read the project stage identifier and project object attribute information corresponding to each project object in the full life cycle project information set, extract the attribute status of each project object in the design, construction and operation and maintenance stages according to the project stage order, and generate a stage attribute status sequence; Based on the stage attribute state sequence, the attribute states of adjacent project stages of the same project object are matched item by item, and attribute content retention, attribute content change, attribute content addition and attribute content disappearance are identified to generate attribute change records of adjacent stages. Based on the attribute change records of adjacent stages, the property content is kept in the corresponding project object state as information inheritance, and the property content of the preceding property is replaced by the corresponding project object state as information replacement, and the inheritance state record and the replacement state record are generated. Based on the attribute change records of adjacent stages, the branching relationship of multiple subsequent project object states corresponding to the same previous project object state is identified, and the termination relationship of the previous project object state having no corresponding attribute state in the subsequent project stage is identified, generating differentiation state records and failure state records. Based on the project object identifier and the project stage sequence, the inherited state record, replacement state record, differentiation state record and failure state record are associated with the corresponding stage attribute state sequence to generate a state evolution record.

5. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The construction project information evolution diagram includes: Read the project object identifier, project stage identifier, and information generation time from the state evolution record, arrange the states of project objects with the same project object identifier according to the project stage order and information generation time, and generate a candidate state succession sequence. Based on the candidate state succession sequence, the project object identifier and information generation time of adjacent project object states are matched in turn, and the project object states with the same project object identifier and the subsequent state succeeding the previous state are determined as continuous succession state pairs. Read the corresponding inheritance status record, replacement status record, differentiation status record and failure status record of the continuous acceptance status pair, set the status of each project object as a status node, and determine the acceptance direction and acceptance type between status nodes according to the corresponding information inheritance, replacement, differentiation and failure. Based on the direction of acceptance, connect the corresponding state nodes of the continuous acceptance state pairs, establish state acceptance edges, and associate the project object identifier, project stage identifier, information generation time and acceptance type with the corresponding state nodes and state acceptance edges to generate a project object evolution sub-graph. The evolution sub-graphs of each engineering object are summarized in the order of design, construction and operation and maintenance. The evolution sub-graphs of engineering objects with branch relationships are connected according to the differentiation state records. The termination state nodes are marked according to the failure state records to construct the engineering information evolution graph.

6. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The generation of cross-stage evolution feature sequences includes: Read the engineering stage identifier and information generation time of each state node in the engineering information evolution diagram, divide the state nodes and state connection edges according to the design, construction and operation and maintenance stages, and arrange them according to the stage order to generate a stage diagram snapshot sequence. Input the snapshot sequence of the stage graph into the evolution relationship encoding layer, read the inheritance, replacement, differentiation and failure acceptance types corresponding to each state acceptance edge, associate and encode the acceptance type with the attribute state of the corresponding state node, and generate stage evolution relationship features; The stage evolution relationship features are input into the graph convolutional layer, and the stage evolution relationship features corresponding to the attribute states of each state node and the adjacent state receiving edges are aggregated to generate the graph convolutional features of the state of the engineering object at each stage. Input the graph convolution features and stage evolution relationship features of each stage into the stage bearing constraint parameter evolution unit. Based on the state bearing relationship corresponding to inheritance, replacement, differentiation and failure, determine the parameter retention, parameter update, parameter branch transmission and parameter termination transmission states respectively. Update the graph convolution parameters in the stage order to generate cross-stage parameter evolution results. The graph convolution features of the corresponding stage are updated based on the cross-stage parameter evolution results. The updated graph convolution features are input into the historical source consistency correction layer. The engineering object identifier, engineering stage identifier and acceptance type are matched in reverse along the state acceptance edge to correct the state node features with inconsistent historical sources and generate the corrected evolution features. The corrected evolution characteristics of each state node are summarized according to the engineering stage sequence, and the evolution states of the corresponding state receiving edges are associated to generate a cross-stage evolution characteristic sequence.

7. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The generated evolution relationship correction results include: Read the corrected evolution features of each state node and the evolution state of the state connection edge in the cross-stage evolution feature sequence, detect the continuity of the connection between adjacent state nodes according to the engineering stage sequence, determine the position where the state connection edge is missing as the state connection edge interruption position, and determine the position where the same preceding state node connects multiple subsequent state nodes as the evolution branch position. Using the interruption position of the state-bearing edge as the boundary, continuously extract the states of engineering objects with the bearing relationship in the preceding engineering stage and the subsequent engineering stage respectively. The preceding continuous engineering object states form the source anchor area, and the subsequent continuous engineering object states form the destination anchor area. Read the project object identifier, project stage identifier, attribute status and acceptance type corresponding to the status of each project object in the source anchor area and the destination anchor area, and determine the candidate acceptance relationship between the end status of the source anchor area and the beginning status of the destination anchor area according to the project stage order. Based on the candidate acceptance relationship, the engineering object identifiers of the source anchor area and the destination anchor area are matched, and the state change direction is determined according to the inheritance, replacement, differentiation and failure changes of adjacent attribute states. The candidate acceptance relationship corresponding to the object identifier and with continuous state change direction is determined as a closed acceptance path. Based on the closed connection path, restore the state connection edge corresponding to the interrupted position of the state connection edge, and check the preceding state node and each subsequent state node corresponding to the position of the evolution branch according to the differentiation state record, and adjust the evolution branches that are inconsistent with the differentiation state record. The restored state acceptance edge, the adjusted evolution branch, and the corresponding acceptance type are associated with the state nodes in the engineering information evolution diagram to generate the evolution relationship correction result.

8. The management system for the full lifecycle engineering information evolution according to claim 1, characterized in that, The generation of full-cycle engineering information evolution management results includes: Read the restored state acceptance edge, adjusted evolution branch and corresponding acceptance type from the evolution relationship correction result, write the corresponding relationship into the engineering information evolution diagram, and generate the corrected engineering information evolution diagram; Based on the correction engineering information evolution diagram, starting from the current stage state node of each engineering object, the corresponding inheritance, replacement, differentiation and failure relationships are traced in reverse order of engineering stages along the state connection edge to generate a historical state tracing sequence. Based on the historical state tracking sequence, identify the state retention, state replacement, state branching and state termination processes of each engineering object in the continuous engineering stage, and associate them with the corresponding engineering object identifier, engineering stage identifier and information generation time to generate the engineering information evolution trajectory. Based on the evolution trajectory of engineering information, the state of the end engineering object that has not been replaced by the subsequent one and has not been marked as invalid is determined as the current valid state, and the historical source corresponding to the current valid state is determined in reverse along the corresponding state-bearing edge; It summarizes the current effective status, historical source, and evolution trajectory of each project object, establishes query relationships according to the project object identifier, and generates full-cycle project information evolution management results.

Images