BIM-based full-life-cycle data asset management method and system

Abstract

The invention discloses a BIM-based full life cycle data asset management method and system, and the method comprises the steps: carrying out the coding of a component in a BIM model, and generating a component attribute vector; generating a component state and data logic and structure alignment completion method through a state acquisition template driven by component attributes; constructing a state modeling structure embedded with component attribute information, inputting a state data stream based on a state prediction model to obtain a predicted risk score sequence, traversing time points with risk scores greater than a risk threshold in the risk score sequence, and setting the time points as suggested maintenance time; and a component asset package is designed and comprises a visual label set, and the visual label set is based on an asset scoring mechanism, and the component asset package is output for structured packaging and unified interface pushing. According to the method, a full-life-cycle data asset expression system is constructed, so that building BIM components are converted from information fragments to data assets capable of being managed continuously.

Description

Technical Field

[0001] This invention relates to the field of engineering management technology, specifically to a method and system for full lifecycle data asset management based on BIM. Background Technology

[0002] With the widespread application of Building Information Modeling (BIM) throughout the entire engineering construction process, BIM has become a data carrier and collaborative platform integrating design, construction, and operation. However, in actual projects, BIM models still suffer from serious data management gaps across different phases. While the component information constructed during the design phase includes static content such as geometry, classification, and attributes, the actual evolution of the components' state cannot be tracked during the construction and operation phases due to platform fragmentation, semantic inconsistencies, and changes in data granularity. The information structure is often reconstructed during construction and staticized during operation, causing the BIM system to degenerate into a graphical display tool rather than a dynamic management system. Especially during the operation phase, the maintenance behavior, usage risks, and health status of components cannot be expressed through the original BIM model, resulting in an invisible, unquantifiable, and unpredictable asset status. Furthermore, current systems generally lack the ability to model the lifecycle evolution of components, failing to organically integrate unstructured data such as maintenance records, sensor observations, and image inspections into the status system during actual operation, thus affecting asset health assessments and maintenance strategy formulation. In addition, the design, construction and operation and maintenance units are often separated in the project, and the responsibility for components and the consistency of data cannot be guaranteed. This makes the data flow of the entire building life cycle in a state of "usable but uncontrollable" from generation and flow to the sedimentation process.

[0003] Therefore, there is an urgent need to establish a lifecycle data structure oriented towards the component level, with multi-dimensional capabilities such as attribute recognition, status update, behavior recording, responsibility tracking, and scoring encapsulation, to truly connect the links between "data-model-behavior-management" and transform building BIM components from information fragments into "data asset units" for sustainable management. Summary of the Invention

[0004] The purpose of this invention is to provide a BIM-based full lifecycle data asset management method and system to solve the problems mentioned in the background art.

[0005] In one aspect, the present invention provides a BIM-based full lifecycle data asset management method, the method comprising the following steps:

[0006] The components in the BIM model are encoded according to a preset encoding method. The attribute information corresponding to the component is associated with the component code to generate a component attribute vector. A component status acquisition configuration vector is generated through attribute-driven status template. Based on the actual acquisition characteristics, a status dataset is recorded based on the status acquisition configuration vector. A structured status vector is obtained by analyzing the status dataset. The structured status vector is organized in chronological order, and the status data stream of the component is output. A component status prediction model is established. An attribute embedding vector is concatenated on the structured status vector. A risk scoring function that integrates a behavior mismatch guidance mechanism and a time smoothing mechanism is designed. The status data stream is input based on the status prediction model to obtain a predicted risk scoring sequence. A risk threshold is set. Time points in the risk scoring sequence where the risk score is greater than the risk threshold are set as suggested maintenance times. A component asset package is designed. The component asset package includes a set of visual tags. Based on the asset scoring mechanism, the visual tag set is output. The component asset package is structured and encapsulated and pushed through a unified interface.

[0007] Furthermore, the preset encoding method includes: integrating the feature fields of the component to generate a unique identifier through a combination numbering function, wherein the feature fields include: component type number, component three-dimensional center coordinates, and component installation time.

[0008] Furthermore, generating the component attribute vector includes: based on the component code, collecting the key attribute information of the component, the key attribute information including: component type number, component main material number, most recent maintenance time, and most recent operation type.

[0009] Furthermore, the step of generating a component state acquisition configuration vector through attribute-driven state templates includes: inputting key attributes of the component, looking up the key attributes in a predetermined template library, and outputting the state acquisition configuration vector corresponding to the component.

[0010] Further, the state data flow of the component is obtained by analyzing the state dataset, including: inputting the state dataset, calling the structure alignment function to map the original state vector to a unified structure, constructing a missing indication vector, introducing regular completion terms to control the weight of missing information in the model, and outputting a structured state vector.

[0011] Furthermore, the behavior mismatch guidance mechanism includes: constructing a flag vector from maintenance behavior to determine whether maintenance behavior has occurred; setting weights for times without maintenance to make the model predict low risk; and enhancing the prediction of high risk when maintenance is performed. The time smoothing mechanism includes: controlling the smoothness parameter and the weight of the time smoothing regularization term to limit state transitions.

[0012] Furthermore, the component asset package includes: the component code, the component attribute vector, the first risk value of the prediction period, the recommended maintenance time, the asset health score, the component responsibility chain identifier, and the set of visual tags; the component responsibility chain identifier is based on the responsibility chain code item, which is a combination of the design code, the construction code, and the operation and maintenance unit code.

[0013] Furthermore, the asset scoring mechanism includes: constructing an asset scoring function, introducing a non-linear lifespan incentive term, and improving the scoring of old components; the set of visual labels includes: the maintenance urgency, historical maintenance frequency indicators, labels and risk scores corresponding to the asset health score in the asset scoring function, and labels corresponding to the recommended maintenance time.

[0014] On the other hand, the present invention also provides a BIM-based full lifecycle data asset management system, the system comprising: an information encoding module, used to encode components in the BIM model according to a preset encoding method, associate the attribute information corresponding to the component with the component encoding, and generate a component attribute vector; a data flow acquisition and processing module, used to generate a component status acquisition configuration vector through attribute-driven status templates, record a status dataset based on the status acquisition configuration vector in combination with actual acquisition characteristics, analyze the status dataset to obtain a structured status vector, organize the structured status vector in chronological order, and output the status data flow of the component; a maintenance behavior prediction module, used to establish a component status prediction model, concatenate attribute embedding vectors on the structured status vector, design a risk scoring function that integrates a behavior mismatch guidance mechanism and a time smoothing mechanism, input the status data flow based on the status prediction model to obtain a predicted risk scoring sequence, set a risk threshold, and traverse the time points in the risk scoring sequence where the risk score is greater than the risk threshold as the suggested maintenance time; and a data asset management module, used to design a component asset package, the component asset package including a set of visual tags, the set of visual tags being structured and encapsulated based on an asset scoring mechanism, and a unified interface push for the output of the component asset package.

[0015] By adopting the above technical solutions, the transformation path of building BIM components from static information to dynamic assets has been realized.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a data modeling method based on a unique component number and attribute structure aggregation mechanism, realizing semantic unity and data chain maintenance of components from design to operation and maintenance stages, ensuring that components at any point in their lifecycle can be accurately identified and invoked; through the component attribute-driven state acquisition template generation logic and structural alignment completion method, component state information can be abstracted into standardized vectors and form a sustainable input time-series data stream, solving the problems of sparse, discontinuous, and inconsistent dimensions of observation data in engineering practice; further, a state modeling structure embedding component attribute information is constructed, and a loss function that integrates behavioral reinforcement terms and smoothing regularization terms is adopted to realize the ability to predict component operation risks, and the next maintenance time suggestion can be generated based on the prediction output; an encapsulation mechanism oriented towards asset expression is proposed, in which prediction results, behavior records, and responsibility chain information are uniformly mapped to asset health scores, while introducing nonlinear life incentive terms and behavior frequency penalty functions to form a component asset data unit with lifecycle continuity, responsibility traceability, and value expressibility. This data unit has a standardized interface structure and can be embedded into BIM platforms, operation and maintenance platforms, asset management systems, etc., to achieve cross-stage and cross-platform data asset sharing and management closed loop. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method steps of the present invention;

[0018] Figure 2 This is a schematic diagram of the system structure in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 This invention provides a method and system for full lifecycle data asset management based on BIM, the method comprising the following steps:

[0021] S1: Encode the components in the BIM model according to the preset encoding method, associate the attribute information corresponding to the component with the component encoding, and generate a component attribute vector.

[0022] Specifically, the preset encoding method includes: integrating the feature fields of the component to generate a unique identifier through a combination numbering function, wherein the feature fields include: component type number, component three-dimensional center coordinates, and component installation time.

[0023] Specifically, generating the component attribute vector includes: based on the component code, collecting the key attribute information of the component, the key attribute information including: component type number, component main material number, most recent maintenance time, and most recent operation type.

[0024] In one possible reality, step S1, component identity standardization modeling and cross-stage attribute organization, includes:

[0025] Structured fields are obtained from actual projects. Data sources include: BIM model files from the design phase, such as IFC format, which can be exported from modeling software with fields such as component type, geometric dimensions, spatial location, and floor to which they belong; installation records and construction logs provided by the construction unit, including information such as installation date, installation batch, and construction unit; and simplified maintenance records extracted from the operation and maintenance system, including the most recent maintenance time and maintenance type labels, such as "replacement" and "cleaning".

[0026] The following coding method is used to uniquely identify associated components throughout their lifecycle:

[0027] ;

[0028] in, Indicates the first The unique number of each component; This function combines multiple fields into a unique identifier. Assign component type numbers, for example: "fan coil unit" is 1, "water pump" is 2; This represents the coordinates of the three-dimensional geometric center of the component, in meters, and is derived from the design model. This indicates a floor operation, used to build stable spatial location indexes; The actual installation time of the component is extracted from the construction log and coded as an integer.

[0029] In one embodiment, the type number of the fan coil unit component is 1, the center coordinates are (12.7, 24.3, 3.2), and the installation time is June 28, 2023. Its number can be represented as: ϕ(1,12,24,3,20230628). This number can be used to index component information across stages and systems.

[0030] After establishing the component code, the key attribute information of the component is collected to form an attribute vector:

[0031] ;

[0032] in, Indicates the first The attribute information vector of each component; Number the main materials of the components; This indicates the time of the component's most recent maintenance, retrieved from the maintenance system. If no record is found, it should be left blank. This indicates the operation type of the most recent maintenance, mapped to an enumeration number.

[0033] The attribute information vectors are uniformly structured into a table indexed by their serial numbers, with each column being a standard field. For components that have not yet entered the operation and maintenance phase, and The field can be empty and will be marked as "unobservable" in subsequent steps.

[0034] In one embodiment, a fan coil unit has the following characteristics: type 1, material 2, installation date 20230628, most recent maintenance date 20240115, and maintenance type 1 (replacement). Its attribute vector is:

[0035] ;

[0036] The output of step S1 is: component number. : Used for stable referencing of component data in subsequent systems; attribute vector : A structured representation of the component's type, material, time, and basic operational status.

[0037] S2: Generate a component status acquisition configuration vector through an attribute-driven status template, combine it with actual acquisition features, record a status dataset based on the status acquisition configuration vector, analyze the status dataset to obtain a structured status vector, organize the structured status vector in chronological order, and output the status data stream of the component.

[0038] Specifically, the attribute-driven state template generation mechanism includes: inputting key attributes of the component, looking up the key attributes in a predetermined template library, and outputting the state acquisition configuration vector corresponding to the component.

[0039] Specifically, the state data flow of the component is obtained by analyzing the state dataset, including: inputting the state dataset, calling the structure alignment function to map the original state vector to a unified structure, constructing a missing indication vector, introducing regularization completion terms to control the weight of missing information in the model, and outputting a structured state vector.

[0040] In one possible reality, step S2 involves acquiring the component's operational status data stream and performing structured preprocessing, including:

[0041] Based on the component code generated in the previous step With attribute vectors This paper proposes an attribute-driven state template generation mechanism based on component attributes. As input, a predefined template library is invoked to determine the state dimensions to be collected for each component, such as temperature, current, and image scoring, as well as the suggested time granularity, and to construct a component state collection configuration vector. :

[0042] ;

[0043] in, Indicates the first The status acquisition configuration for each component includes the required status variable names, units, and suggested sampling frequency, etc. This is a template lookup function, and its implementation is based on the project rule base.

[0044] In one embodiment, fan coil unit ( ) + Metal Material → Collect temperature, current, and image damage scores every 30 days.

[0045] The actual collected operational status data exhibits the following characteristics: low periodic collection frequency, with inspection images only once per quarter; observable components are concentrated in key equipment, including HVAC, electrical, and water supply and drainage components; and inconsistent status dimension structure, with some only showing whether a fault has occurred, while others include multiple variables such as temperature, vibration, and current.

[0046] Based on the state acquisition configuration vector and actual acquisition features, observations at each time point are extracted from the operation and maintenance system, sensor platform, or inspection image system to form an irregular state dataset. This step includes designing a structure alignment function. All state observations are embedded into a unified state vector structure, and a regularization completion mechanism is added:

[0047] ;

[0048] in: For at a certain point in time Above, the first The structured state vector of each component; The original observation data (which may be partially missing or have irregular dimensions) (may be an irregular state dataset); For the structural alignment function, a dimensional expansion + marker filling strategy is adopted to complete the state information of different components into the same structure (e.g., missing items are filled with zeros and a missing mask is attached). The observation missing rate vector, also known as the structured state mask vector, has each dimension taking the value of... This indicates that the maintenance status was missing at that point in time; otherwise, it is... ; The weights for the complete items (typically ranging from 0.1 to 0.5) are used to emphasize the impact of observation completeness on subsequent state judgments.

[0049] Introducing a regularized completion mechanism not only preserves the sparsity information of the original observations but also provides important clues about missing dimensions for subsequent models, avoiding spurious states introduced by blind completion. After constructing the state vector at each time point, a state-time series matrix for that component is formed. :

[0050] ;

[0051] in, For the first The state data flow of each component; each Includes a fully aligned state vector; The number of observation points for this component is determined by the number of maintenance visits, inspection frequency, etc.

[0052] In one embodiment, if a fan coil unit Four status data points were collected throughout the year. The raw observations included temperature, current, and image scores. It is a three-dimensional vector; however, for a certain water pump component, only one "replacement" operation was recorded, so its This is a one-dimensional 0 / 1 type state. (Through...) After unified mapping, It will have a unified structure, such as a 6-dimensional vector, where 3 dimensions are actual observations and 3 dimensions are for missing information filling, while retaining the missing information.

[0053] The output of step S2 is: Component state time series matrix. Structured state mask vector and its regularization parameters It is used to identify effective observation dimensions during modeling.

[0054] S3: Establish a component state prediction model, concatenate attribute embedding vectors on the structured state vector, design a risk scoring function that integrates behavior mismatch guidance mechanism and time smoothing mechanism, input the state data stream based on the state prediction model to obtain the predicted risk score sequence, set a risk threshold, and set the time points in the risk score sequence where the risk score is greater than the risk threshold as the suggested maintenance time.

[0055] Specifically, the behavior mismatch guidance mechanism includes: constructing a flag vector from maintenance behavior to determine whether maintenance behavior has occurred; setting weights for times without maintenance to make the model predict low risk; and enhancing the prediction of high risk when maintenance is performed. The time smoothing mechanism includes: controlling the smoothness parameter and the weight of the time smoothing regularization term to limit state transitions.

[0056] In one possible reality, S3 component lifecycle state modeling and maintenance behavior prediction includes:

[0057] The state sequence matrix output by S2 and maintenance behavior mask sequence As the primary input, it is combined with the attribute vector of the component from step one. Establish a state evolution prediction model for BIM components.

[0058] The first part of this step involves constructing the "attribute-driven state input structure." This is the structured state vector output by S2. Based on this, the attribute embedding vector of the splicing components This is used to inject static features into the model, enabling it to model class differentiation. The state sequence input is defined as:

[0059] ;

[0060] in, It is the first Each component in time The complete input vector; It is a component attribute embedding vector, which is composed of component categories. and main materials Mapped from, with an embedding dimension of It is usually set to 4 to 8 dimensions; Indicates a splicing operation; Source This is for state observation after alignment.

[0061] The second part of this step introduces a health risk scoring function with a "maintenance behavior calibration item." To address issues such as significant differences in component behavior patterns and weak maintenance signals, a risk prediction loss function with bidirectional behavior reinforcement is constructed. Health prediction value. Indicates the first Each component in time The risk score, with a value range of [value missing]. An adaptive maintenance behavior alignment loss function is introduced:

[0062] ;

[0063] in, It is a predicted risk value; the higher the value, the more likely the component is to fail soon. , , used for setting behavior guidance weights; For the first Components in time A flag indicating whether maintenance is required (missing indicator vector); The enhancement factor for maintenance period identification is set to 3-5; the last term is a time smoothing regularization term. To control the weights, To control smoothness, a value of 0.5 is recommended.

[0064] The time smoothing regularization term guides the model to identify true "chronic degradation" signals rather than just responding to fluctuations by limiting the magnitude of state transitions.

[0065] The maintenance behavior alignment loss function explicitly models "behavior mismatch" as the source of error, encouraging the model to predict low risk when there is no maintenance and high risk when there is maintenance. At the same time, it prevents the risk curve from oscillating excessively through a smoothing term, which fits the reality of the construction industry where "hidden faults account for a large proportion and have little fluctuation".

[0066] The third part of this step provides predicted outputs and maintenance time suggestions. After training, the model can predict future... Predict the state at each time step and output a risk score sequence. We set risk thresholds. Once the predicted risk value exceeds this value, the component is considered to have entered the maintainable period, and the system suggests the earliest high-risk point as the target maintenance time.

[0067] ;

[0068] in, Indicates the next recommended maintenance time; This is the risk threshold; a value of 0.7 to 0.8 is recommended and can be adjusted. For component number One predicted time point.

[0069] The S3 output includes: a predicted risk sequence for the future state of the component. Next recommended maintenance time .

[0070] S4: Design a component asset package, which includes a set of visual tags. The set of visual tags is based on an asset scoring mechanism. The component asset package is then structured, encapsulated, and pushed through a unified interface.

[0071] Specifically, the component asset package includes: the component code, the component attribute vector, the first risk value of the prediction period, the recommended maintenance time, the asset health score, the component responsibility chain identifier, and the set of visual tags; the component responsibility chain identifier is based on the responsibility chain code item, which is a combination of the design code, the construction code, and the operation and maintenance unit code.

[0072] Specifically, the asset scoring mechanism includes: constructing an asset scoring function, introducing a non-linear lifespan incentive term, and improving the score of old components; the set of visual labels includes: the maintenance urgency index, historical maintenance frequency index, labels and risk scores corresponding to the asset health score in the asset scoring function, and labels corresponding to the recommended maintenance time.

[0073] In one possible reality, the encapsulation of S4 component asset packages and the generation of data asset interfaces specifically include:

[0074] Combined with the predicted risk sequence in S3 and recommended maintenance time Combined with the attribute vector in S1 Design a data asset encapsulation structure for the entire building lifecycle—the "component asset package"; Component asset package Defined as:

[0075] ;

[0076] in: Number the components; For component attribute vectors; This is the first risk value in the forecast period, reflecting the current operating status; The next recommended maintenance time output by the model; The health score for the component assets is calculated using the comprehensive mapping function designed in this step. It serves as a component responsibility chain identifier, used to link the design, construction, and operation and maintenance parties. A collection of cross-platform visual tags.

[0077] In practical implementation, terms such as "high risk" and "long-term uninsured" are used for highlighting on the BIM platform.

[0078] Asset rating mechanism This is used to map multidimensional state information into a comprehensive score that reflects the health of component assets and the risk of operating costs. This rating not only considers the risk prediction results of components, but also incorporates the urgency of maintenance judgment and the maintenance cost index, adapting to the decision-making needs of the construction industry where "failures are few, but once they occur, the costs are high."

[0079] The scoring function is as follows:

[0080] ;

[0081] in: To score the health of component assets, the numerical range is: ; The initial risk prediction score for the component in the future period; This indicates the deviation between the remaining days until the next maintenance and the set period (e.g., 30 days). This is a historical maintenance frequency indicator, representing the number of maintenance operations that occur per unit of time, based on... and Statistics on maintenance activities over the past year; The average remaining life of the current component type (e.g., 6 years for a fan, 4 years for a water pump) is obtained by referring to a table and is used to encourage components that have been operating stably for a long time to receive a score. This is an adjustable parameter, which in reality can be adjusted as follows: , , , .

[0082] The scoring function also introduces a nonlinear lifetime excitation term. This addresses the issue of generally low scores for all older components in traditional component evaluation. In BIM full lifecycle management, although some older components have been in operation for a long time, their true asset value remains high due to their good operation and structural stability, and they should be incentivized and expressed in the model.

[0083] Meanwhile, to address the issue of unclear responsibility positioning during multi-platform system integration, a responsibility chain coding item was defined in the asset package. This field is composed of the designer's code, the construction company's code, and the operation and maintenance unit's code, and is used for cross-platform tracking of stakeholders related to components. For example:

[0084] ;

[0085] in: Indicates the design unit code; Indicates the construction unit code; This indicates the code of the party responsible for operation and maintenance. All three originate from the component responsibility table maintained in the construction management platform, BIM delivery log, or owner platform; the introduction of this structure provides a foundation for subsequent asset budget management, risk attribution, and maintenance allocation, and is an important supporting field for the digital governance of building assets.

[0086] To enable the BIM system to visualize and access asset packages, we define an optional tag set. It is generated by the following rules: If Add the label "high risk"; if Add the tag "Upcoming Maintenance"; if Add the tag "Frequent Repairs"; if and Add the tag "Requires manual confirmation".

[0087] These tags will be used to highlight components in the BIM 3D model, helping maintenance personnel to quickly identify and locate them.

[0088] The final output component asset package It will be encapsulated into a structured data format, such as JSON or structured SQL records, and can be written to BIM platforms, smart operation and maintenance platforms, auditing systems, etc. through a unified interface to achieve a closed loop of lifecycle data asset management across systems.

[0089] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: they achieve semantic unification and data chain maintenance of components from design to operation and maintenance; through the component attribute-driven state acquisition template generation logic and structural alignment completion method, component state information can be abstracted into standardized vectors and form a time-series data stream for continuous input; they realize the ability to predict component operation risks; they generate a component asset data unit with lifecycle continuity, traceability of responsibility, and expressibility of value; they realize the transformation path of building BIM components from static information to dynamic assets, and construct a full lifecycle data asset expression system with complete structure, clear variables, and closed logic.

[0090] This application also provides a BIM-based full lifecycle data asset management system, such as... Figure 2As shown, the system includes: an information encoding module, used to encode components in the BIM model according to a preset encoding method, associate the attribute information corresponding to the component with the component encoding, and generate a component attribute vector; a data flow acquisition and processing module, used to generate a component status acquisition configuration vector through attribute-driven status templates, record a status dataset based on the status acquisition configuration vector in combination with actual acquisition characteristics, analyze the status dataset to obtain a structured status vector, organize the structured status vector in chronological order, and output the status data flow of the component; a maintenance behavior prediction module, used to establish a component status prediction model, concatenate attribute embedding vectors on the structured status vector, design a risk scoring function that integrates a behavior mismatch guidance mechanism and a time smoothing mechanism, input the status data flow based on the status prediction model to obtain a predicted risk scoring sequence, set a risk threshold, and traverse the risk scoring sequence to set the time points where the risk score is greater than the risk threshold as the suggested maintenance time; and a data asset management module, used to design a component asset package, the component asset package including a set of visual tags, the set of visual tags based on the asset scoring mechanism, and output the component asset package for structured encapsulation and unified interface push.

[0091] The BIM-based full lifecycle data asset management system provided in this embodiment can implement the steps of the BIM-based full lifecycle data asset management method described above, and achieve the same results.

[0092] 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 BIM-based full lifecycle data asset management method, characterized in that, The method includes: The components in the BIM model are encoded according to a preset encoding method, and the attribute information corresponding to the components is associated with the component codes to generate a component attribute vector. A component status acquisition configuration vector is generated by using an attribute-driven status template. Combined with actual acquisition features, a status dataset is recorded based on the status acquisition configuration vector. A structured status vector is derived from the status dataset. The structured status vector is organized in chronological order, and the status data stream of the component is output. A component state prediction model is established. An attribute embedding vector is concatenated on the structured state vector. A risk scoring function that integrates a behavior mismatch guidance mechanism and a time smoothing mechanism is designed. The state data stream is input based on the state prediction model to obtain a predicted risk score sequence. A risk threshold is set. The time points in the risk score sequence where the risk score is greater than the risk threshold are set as the suggested maintenance time. The design includes a component asset package, which comprises a set of visual tags. Based on an asset scoring mechanism, the visual tag set is used to output the component asset package for structured encapsulation and unified interface push.

2. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The preset encoding method includes: integrating the feature fields of the component to generate a unique identifier by combining the numbering function, wherein the feature fields include: component type number, component three-dimensional center coordinates, and component installation time.

3. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, Generating a component attribute vector includes: based on the component code, collecting the key attribute information of the component, including: component type number, component main material number, most recent maintenance time, and most recent operation type.

4. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The step of generating a component state acquisition configuration vector by using attribute-driven state templates includes: inputting key attributes of the component, looking up the key attributes in a predetermined template library, and outputting the corresponding state acquisition configuration vector of the component.

5. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The state data flow of the component is obtained by analyzing the state dataset, including: inputting the state dataset, calling the structure alignment function to map the original state vector to a unified structure, constructing a missing indication vector, introducing regularization completion terms to control the weight of missing information in the model, and outputting a structured state vector.

6. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The behavior mismatch guidance mechanism includes: constructing a flag vector from maintenance behavior to determine whether maintenance behavior has occurred; setting weights for times without maintenance to make the model predict low risk; and enhancing the prediction of high risk when maintenance is performed. The time smoothing mechanism includes: controlling the smoothness parameter and the weight of the time smoothing regularization term to limit state jumps.

7. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The component asset package includes: the component code, the component attribute vector, the first risk value of the prediction period, the recommended maintenance time, the asset health score, the component responsibility chain identifier, and the set of visual tags; the component responsibility chain identifier is based on the responsibility chain code item, which is a combination of the design code, the construction code, and the operation and maintenance unit code.

8. The BIM-based full lifecycle data asset management method according to claim 1, characterized in that, The asset scoring mechanism includes: constructing an asset scoring function, introducing a non-linear lifespan incentive term, and improving the scoring of old components; the set of visual labels includes: the repair urgency, historical maintenance frequency indicators, labels and risk scores corresponding to the asset health score in the asset scoring function, and labels corresponding to the recommended maintenance time.

9. A BIM-based full lifecycle data asset management system, characterized in that: The system includes: an information encoding module, used to encode components in the BIM model according to a preset encoding method, associate the attribute information corresponding to the component with the component encoding, and generate a component attribute vector; a data flow acquisition and processing module, used to generate a component status acquisition configuration vector through attribute-driven status templates, record a status dataset based on the status acquisition configuration vector in combination with actual acquisition characteristics, analyze the status dataset to obtain a structured status vector, organize the structured status vector in chronological order, and output the status data flow of the component; a maintenance behavior prediction module, used to establish a component status prediction model, concatenate attribute embedding vectors on the structured status vector, design a risk scoring function that integrates a behavior mismatch guidance mechanism and a time smoothing mechanism, input the status data flow based on the status prediction model to obtain a predicted risk scoring sequence, set a risk threshold, and traverse the risk scoring sequence to set the time points where the risk score is greater than the risk threshold as the suggested maintenance time; and a data asset management module, used to design a component asset package, the component asset package including a set of visual tags, the set of visual tags based on the asset scoring mechanism, and output the component asset package for structured encapsulation and unified interface push.

Images