Modeling method for component credibility in station house pipe gallery information model

Abstract

The invention discloses a modeling method for component credibility in a station house pipe gallery information model, which comprises the following steps of: generating error mode characteristics according to a construction process section of a boundary component, analyzing a conflict mode among pre-stored multi-source data by utilizing the error mode characteristics, and calculating single credibility of the boundary component; constructing a process block-structure block double-layer graph comprising a process block layer and a structure block layer, and respectively associating the boundary components to process block nodes and structure block nodes in the process block-structure block double-layer graph; and configuring an anisotropic propagation weight according to the error mode characteristics, executing spatial propagation of the credibility of the single body on the process block-structure block double-layer graph, and generating a spatial credibility field. According to the method, the construction logic and the structural boundary are explicitly coded, so that the problems that errors cross physical boundaries and are diffused and multi-source data conflicts are difficult to explain in a traditional method are solved, and accurate evaluation on the credibility of the component in the complex boundary area is realized.

Description

Technical Field

[0001] This invention belongs to the field of building information modeling, and in particular, it is a modeling method for the credibility of components in a station building and utility tunnel information model. Background Technology

[0002] With the advancement of large-scale integrated transportation hub construction, the area where station buildings and integrated utility tunnels intersect has become a critical zone for engineering quality control due to its complex spatial logic and high-density electromechanical pipelines. In the digital delivery and operation and maintenance of such projects, the accuracy of BIM (Building Information Modeling) component information directly affects subsequent operation and maintenance safety and renovation and expansion decisions. Therefore, accurate modeling and evaluation of the credibility of component data in BIM models has significant research and application value.

[0003] Current BIM quality checks primarily rely on rule-based geometric collision detection or static attribute compliance checks. In verification involving multi-source data (such as design models, point cloud scans, and construction records), existing technologies typically employ weighted averaging based on Gaussian noise assumptions or Kalman filtering for data fusion. In spatial analysis, mainstream methods largely construct topological relationships based on a single geometric adjacency graph, assuming that physically close components have similar attribute states, and accordingly use isotropic smoothing algorithms (such as Markov random fields or Laplace smoothing) to spatially infer errors or confidence levels.

[0004] However, existing technologies struggle to address the non-uniform error distribution caused by both construction process logic and physical structural boundaries when dealing with complex station-pipe gallery interface projects. Specifically, existing solutions suffer from the following technical problems: a single geometric adjacency graph ignores the decisive influence of physical barriers and process associations on error propagation, leading to errors often propagating incorrectly across settlement joints or failing to effectively aggregate within the same pouring batch, resulting in false positives or false negatives in risk assessments; furthermore, traditional multi-source data fusion lacks the ability to explain the causal causes of conflicts, and when faced with inconsistencies between design drawings and on-site point clouds, it cannot distinguish between systematic construction deviations (such as overall formwork offset) and random measurement noise, resulting in the final calculated reliability values ​​lacking physical meaning and failing to guide precise rectification and decision-making. Summary of the Invention

[0005] The purpose of this invention is to provide a modeling method for the credibility of components in a station building utility tunnel information model, so as to solve the above-mentioned problems existing in the prior art.

[0006] The technical solution, a modeling method for component credibility in the station building utility tunnel information model, includes:

[0007] Obtain and generate error pattern features based on the construction sequence profile of the boundary component, use the error pattern features to analyze the conflict patterns between pre-stored multi-source data, and calculate the individual credibility of the boundary component.

[0008] Construct a two-layer process block-structure block graph that includes a process block layer and a structure block layer, and associate the boundary components with the process block nodes and structure block nodes in the two-layer process block-structure block graph respectively;

[0009] Based on the error mode characteristics, anisotropic propagation weights are configured, and spatial propagation of individual credibility is performed on the process block-structure block bilayer graph to generate a spatial credibility field.

[0010] Beneficial effects: By explicitly encoding construction logic and structural boundaries, this invention solves the problems of error propagation across physical boundaries and difficulty in interpreting conflicts between multiple sources of data in traditional methods, and achieves accurate assessment of the reliability of components in complex boundary areas. Attached Figure Description

[0011] Figure 1 A flowchart illustrating the steps of a method for modeling the credibility of components in a station building utility tunnel information model, as provided in this application embodiment.

[0012] Figure 2 A flowchart illustrating the steps for constructing a process block-structure block dual-layer diagram comprising a process block layer and a structure block layer, as provided in this embodiment of the application.

[0013] Figure 3 A flowchart illustrating the steps for configuring anisotropic propagation weights based on error mode characteristics, as provided in the embodiments of this application.

[0014] Figure 4 A flowchart illustrating the steps for calculating the individual reliability of boundary components in an embodiment of this application. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms include and have, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0017] like Figure 1 As shown, a method for modeling the reliability of components in a station building utility tunnel information model includes the following steps:

[0018] Error pattern features are obtained and generated based on the construction process profile of the boundary components. The error pattern features are used to analyze the conflict patterns between pre-stored multi-source data and calculate the individual credibility of the boundary components.

[0019] Alternatively, the process involves reading multi-source heterogeneous data, preprocessing it to obtain preprocessed multi-source heterogeneous data, acquiring the construction process profile of the boundary components, generating error pattern features based on the construction process profile, and using the error pattern features to analyze the conflict patterns between the preprocessed multi-source heterogeneous data to calculate the individual credibility of the boundary components.

[0020] In this embodiment, multi-source heterogeneous data is read, including the basic BIM model from the design phase, the original set of engineering documents, point cloud data or scan-to-BIM results, and manual measurement and inspection records from the operation and maintenance phase. The basic BIM model from the design phase includes geometric and attribute information such as floor slabs, shear walls, shafts, and utility tunnel structures. The original set of engineering documents covers construction drawings, as-built drawings, change orders, and technical negotiation records. Point cloud data reflects the actual on-site geometric state. The system unifies coordinates and standardizes units, ensuring all data sources are in the same three-dimensional coordinate system, and identifies boundary components located in the connection area between the station building and the utility tunnel, such as openings, shafts, and inspection holes, based on spatial location. Based on this, the discrete construction records are structurally reconstructed according to the error pattern characteristics generated from the construction process profiles of the boundary components. The construction process profile refers to a digital description of the key physical processes experienced by each boundary component throughout its entire construction lifecycle. Specifically, the system integrates construction process data such as construction logs, pouring records, and work orders to clearly identify the pouring section, formwork system type, and work team responsible for each boundary component. By querying a pre-set error knowledge base, the reconstructed construction process profile is matched with typical process features in the knowledge base. If a match is successful, the boundary component is assigned a corresponding error pattern feature. For example, for a component with an opening that has undergone post-excavation construction and is located in a densely reinforced area, it will be assigned a specific error pattern feature. This feature indicates that this process combination is prone to producing large planar position deviations, but relatively small vertical deviations.

[0021] For example, using error pattern features to analyze conflict patterns among multi-source data and calculate the individual reliability of boundary components transforms prior knowledge of the physical world into a mathematical measure of reliability. Conflict patterns refer to geometric or attribute differences exhibited by different data sources when describing the same component. Specifically, the system quantifies the differences in geometric location, dimensional parameters, and document descriptions among design models, as-built drawings, point clouds, and manual measurement data. Based on the aforementioned generated error pattern features, these differences are explained causally. For instance, in the post-drilling opening pattern, if a significant difference in planar position is observed between the design model and the point cloud data, the system will, based on the pattern's indication, determine that the planar information of the design model is invalid and assign a higher reliability weight to the point cloud data. This ensures that the final aggregated individual reliability truly reflects the physical reliability of the component in its current state.

[0022] Construct a process block-structure block bilayer graph that includes a process block layer and a structure block layer, and associate the boundary components with the process block nodes and structure block nodes in the process block-structure block bilayer graph, respectively.

[0023] Specifically, a two-layer graph of process blocks and structural blocks is constructed to establish a topological network that can simultaneously express construction logic relationships and physical structural boundaries. The process block layer consists of several process block nodes, each representing a physical region with the same construction attributes. Specifically, the system aggregates regions with the same pouring batch, using the same formwork system, or managed by the same construction team, mapping them to process block nodes. This aggregation method implicitly assumes the homogeneity of errors, meaning that components within the same process block often have similar error characteristics. The structural block layer consists of several structural block nodes, each representing a spatial region separated by physical structural boundaries. The system divides the physical space based on structural boundary information such as structural joints, expansion joints, and settlement joints, generating structural block nodes. These structural boundaries often physically disrupt the continuity of the structure, thus playing a role in blocking or attenuating errors in the error propagation model. In the connection relationships of the two-layer graph, each specific boundary component is regarded as a basic node in the graph. The system establishes a topological mapping relationship between the boundary component and its corresponding process block node and structural block node. This means that each boundary component node has two upward connecting edges, pointing to its respective process context and structural context, respectively. The two-layer topology provides multi-dimensional path support for subsequent credibility propagation.

[0024] Based on the error mode characteristics, anisotropic propagation weights are configured, and spatial propagation of individual credibility is performed on the process block-structure block bilayer graph to generate a spatial credibility field.

[0025] In this embodiment, the configuration of anisotropic propagation weights is crucial for achieving accurate error inference. The anisotropic propagation weights are not uniform constants but dynamically set variables based on edge type and error pattern characteristics. The system assigns a first basic weight to propagation paths within the same process block node and a second basic weight to propagation paths crossing structural boundary information. To reflect physical laws, the first basic weight is set higher than the second basic weight. For example, the first basic weight can be set to 1.0, representing that errors propagate easily within the same process block; while the second basic weight can be set to 0.1, representing that errors are difficult to propagate across settlement joints. Furthermore, the system analyzes the sensitive dimensions associated with error pattern characteristics and increases the components of the anisotropic propagation weights on these sensitive dimensions. For example, in the error pattern of overall template upward movement, vertical positional deviation is the main feature; therefore, the system increases the propagation weight on the vertical dimension while keeping the planar dimension weight unchanged. This ensures that the propagation of credibility has a clear direction and specificity.

[0026] During the propagation execution phase, the system initializes the node state values ​​of each node in the process block-structure block bilayer graph with individual confidence levels. Using anisotropic propagation weights, the system iteratively updates the node state values ​​between adjacent nodes in the process block-structure block bilayer graph. This simulates the diffusion and balance of error or confidence information in the network. The system continuously calculates the maximum change in node state values ​​between adjacent iterations. When the maximum change is less than a preset convergence threshold, such as 0.01, the propagation process is considered converged. The final node state values ​​are mapped back to the physical space, generating a spatial confidence field. This field data provides a quantified confidence index for each location within the boundary area between the station building and the utility tunnel, which can be directly used to guide subsequent high-risk operation management.

[0027] In one possible implementation, generating error pattern features includes:

[0028] The construction process data and structural segmentation information are acquired and integrated to reconstruct the construction process profile of the junction components. The construction process profile records the pouring section to which the junction components belong, the formwork system type, and the construction team attributes.

[0029] In other words, the construction process data and structural segmentation information are acquired and integrated, and based on the integration results, a construction process profile is reconstructed, including the pouring section to which the boundary components belong, the formwork system type, and the construction team attributes.

[0030] In this embodiment, the reconstructed construction process profile serves as a bridge connecting discrete construction records with specific physical components. For example, construction process data includes original documents such as construction logs, concrete pouring records, formwork installation plans, and work orders. Structural segmentation information includes the post-cast strip division, flow section division, and structural zoning diagram in the design drawings. The system associates discrete records with specific boundary components by parsing the timestamps, spatial coordinate ranges, and component numbers in the aforementioned data.

[0031] Specifically, the reconstructed construction sequence profile is a structured data object. For a specific boundary component, such as the opening numbered JD05, its sequence profile may record the following attributes: the pouring section is the second flow section, the pouring time is May 20, 2024; the formwork system type is a loose-fitting wooden formwork system; and the construction team attribute is Carpentry Team Three. These attribute combinations constitute the unique construction characteristics of the component, determining its potential error tendency.

[0032] The system queries a pre-defined error knowledge base and matches the construction process profile with typical process features in the error knowledge base.

[0033] In this embodiment, a pre-defined error knowledge base stores expert experience and historical data statistics in the form of a structured table. Each entry in the knowledge base contains at least the following fields: mode_id, used to uniquely identify an error mode; pattern_proc, a typical process feature describing the process combination conditions that trigger the mode; pattern_conflict, a typical conflict feature describing the expected range and direction of geometric deviation under the mode; and source_adjust, a source reliability adjustment parameter defining the correction coefficient for each data source. In a specific example, the knowledge base may contain the following typical entries: Mode 1 corresponds to a post-drilled opening located in a densely reinforced area, and its typical process features include whether it is post-drilled and a high reinforcement density level; Mode 2 corresponds to the overall upward movement of the formwork, and its typical process features include that the formwork system is a large steel formwork and whether it is post-drilled. During system operation, the process profile attributes of the reconstructed JD05 component are used as query conditions and compared with the pattern_proc field of the typical process feature in the knowledge base. If the attributes of JD05 match the features of Mode 1, the match is considered successful.

[0034] Based on the matching results, the corresponding error mode features are determined for the boundary components. The error mode features are defined with typical geometric deviation trends under the process combination.

[0035] Specifically, once a match is successful, the system extracts the corresponding detailed information from the knowledge base and encapsulates it into the error pattern characteristics of the component. The error pattern characteristics include not only the pattern number but also a quantitative description of typical geometric deviation trends. Taking Pattern 1 as an example, its error pattern characteristics might define the following typical geometric deviation trends: planar position deviation ranges from 30 mm to 80 mm, with the deviation direction mostly along the shorter side of the floor slab; vertical position deviation ranges from less than 15 mm. This means that if a 50 mm planar deviation is observed in the component, the system will tend to consider this to be a construction error consistent with expectations, rather than a measurement error. Conversely, if a vertical deviation of 40 mm is observed, it does not conform to the pattern characteristics and may indicate the existence of other unknown anomalies.

[0036] like Figure 4 As shown, in an exemplary embodiment, calculating the individual confidence level of the boundary member includes:

[0037] Quantify the differences in geometric location, size parameters, and document descriptions of multi-source data to generate conflict patterns.

[0038] In this embodiment, the conflict mode is a structured description of observed data inconsistencies. Specifically, the system extracts the geometric parameters of the same boundary component from different data sources. For example, for a square opening, the center coordinate P in the design model is extracted. design and size S design And the center coordinates P in the point cloud data cloud and size S cloud Calculate the deviation vector between the two. For example, calculate the plane deviation Δ. plan equals P cloud plane components minus P design The modulus of the planar component; calculate the vertical deviation Δ. vert equals P cloud The vertical component minus P design The system calculates the absolute value of the vertical component. Simultaneously, it compares discrepancies in the document description. The resulting conflict pattern is a vector containing multidimensional deviation values. For example, the conflict pattern of a component might be represented as: plane deviation Δ plan Equal to 50 millimeters, vertical deviation Δ vert It equals 5 millimeters, and the documentation is consistent.

[0039] Based on the conflict pattern, obtain the source reliability adjustment parameters corresponding to the error pattern features in the preset error knowledge base.

[0040] In a preferred implementation, the source reliability adjustment parameters corresponding to the error pattern features in a preset error knowledge base are obtained, including:

[0041] The error pattern features are analyzed to extract the corresponding typical conflict features; the conflict similarity between the conflict pattern and the typical conflict features is calculated; in response to the conflict similarity being greater than the preset matching threshold, the error pattern features are determined to be successfully matched, and the source reliability adjustment parameters associated with the error pattern features are extracted.

[0042] In this embodiment, similarity calculation is crucial for determining whether an error pattern is applicable. For example, the system reads typical conflict features from the error pattern characteristics. Assuming the current component is labeled as pattern 1, its typical planar deviation mean Δ plan_typ The typical vertical deviation is Δ, which is 55 mm. vert_typ The value is 10 mm. The conflict similarity Sim is calculated using a Gaussian similarity formula based on an exponential function. total The specific formula is:

[0043] Sim total =w plan *exp(-|Δ plan_obs -Δ plan_typ | / σ plan ) + w vert * exp(-|Δ vert_obs -Δ vert_typ | / σ vert );

[0044] Sim total The total similarity; w plan w vert Δ represents the weighting coefficients for the planar and vertical dimensions. plan_obs Δ vert_obs The observed planar and vertical deviation values; Δ plan_typ Δ vert_typ σ is the mean of typical deviations in the knowledge base. plan σ vert The scale parameter is used for similarity calculation.

[0045] In a specific numerical case, let w plan It is 0.7, w vert σ is 0.3 plan It is 20 mm, σ vert It is 10 millimeters. Observation value Δ plan_obs It is 50 mm, Δ vert_obs The value is 5 mm. The result is approximately 0.726. Assuming the preset matching threshold is 0.7, since 0.726 is greater than 0.7, the system determines that the match is successful and extracts the source reliability adjustment parameters associated with this pattern from the knowledge base, such as the design source plane position reduction coefficient α. design The point cloud source plane position lifting factor β is 0.4. cloud It is 0.5.

[0046] The source reliability weights of each data source in the multi-source data are asymmetrically corrected using source reliability adjustment parameters, and the individual credibility is calculated based on the corrected source reliability weights.

[0047] In this embodiment, asymmetric correction refers to adjusting the weights of different data sources in different directions according to the causal logic of the error. The system updates the original source reliability weights based on the source reliability adjustment parameters. Specifically:

[0048] R design_new = R design_old * ( 1 -α design );

[0049] R cloud_new = R cloud_old + β cloud * (1 -R cloud_old );

[0050] Where R design_new R cloud_new For the improved reliability of the design source and point cloud source; R design_old R cloud_old The original reliability before correction; α design β is the reduction factor for the design source; cloud This represents the enhancement factor for the point cloud source. The design source is directly reduced using a multiplication factor, while the point cloud source is enhanced using an increment factor.

[0051] Continuing with the numerical examples above, let the reliability of the original design source be R. design_old The reliability R of the original point cloud source is 0.8. cloud_old The value is 0.6. Substitute the parameters to calculate: R design_new R is 0.48; cloud_new The value is 0.8. It can be seen that after confirming the occurrence of post-construction errors, the reliability of the design drawings decreases, while the reliability of the point cloud data reflecting the current situation improves. The individual reliability of the component is calculated by weighting the geometric data from each source using the corrected weights.

[0052] In a further embodiment, calculating the individual reliability of the boundary member further includes:

[0053] Determine the risk level associated with error pattern characteristics, and perform time decay calculation on the credibility of the individual based on the risk level and the time interval since the last assessment.

[0054] Alternatively, the error pattern characteristics are analyzed to extract the risk level; based on the risk level and the time interval since the last assessment, the individual credibility is calculated using time decay to obtain the individual credibility after the time decay calculation.

[0055] In this embodiment, time decay calculations are used to simulate the cumulative effect of changes in the engineering site environment and uncertainties during the period without re-measurement. The system analyzes the error mode characteristics of each boundary component and extracts preset risk level parameters. Optionally, the risk level is divided into three levels: high risk, medium risk, and low risk, each corresponding to a different time decay coefficient λ. dim For example, for high-risk components involving post-excavation construction or soft foundation areas, due to the greater possibility of unstable physical states, the system uses its corresponding attenuation coefficient λ. high Set to a relatively large value, such as 0.01; for medium-risk components in conventional cast-in-place areas, the attenuation coefficient λ medium The attenuation coefficient λ is set to 0.005; while for prefabricated, reliably connected, low-risk components, the attenuation coefficient λ is... low The value is set to 0.002. For example, the time decay calculation uses an exponential decay model. Let t be the time of the last credibility assessment for a certain dimension. last The current evaluation time is T. now Then the time interval Δt between the two days equal to T now Subtract t last The number of days. Let this dimension be in t. last The reliability of the time point is C. dim_last Then in T now Credibility C of a moment after time decay dim_time for:

[0056] C dim_time = C dim_last * exp( -λ dim *Δt days );

[0057] Where C dim_time C represents the decayed confidence level at the current moment. dim_last The credibility level at the time of the previous assessment; λ dim This is the attenuation coefficient for this dimension, and its value depends on the risk level; Δt days This represents the number of days since the last assessment.

[0058] In a possible numerical calculation case, assume that the vertical safety margin confidence level C of a high-risk boundary component at its last re-measurement 180 days ago is... dim_last The value is 0.9. Since it belongs to the high-risk level, λ is taken as λ. dim The value is 0.01. Substituting it into the formula, the result is 0.1485. It can be seen that for high-risk components that have not been retested for a long time, their reliability will decrease over time, thus triggering mandatory retesting requirements in subsequent operational decisions.

[0059] In some alternative implementations, the time decay function can also take the form of linear decay or piecewise stepped decay to adapt to different engineering management needs. Furthermore, the system can also incorporate environmental impact factors, such as automatically and temporarily increasing the decay coefficient during the rainy season or freeze-thaw cycles, to reflect the accelerated impact of the environment on the stability of the components.

[0060] In response to a retest event for a boundary component, the individual unit confidence level after execution time decay calculation is incrementally updated based on the measurement accuracy attribute of the retest event.

[0061] In this embodiment, incremental updates are used to restore or improve the reliability of components using the latest measured data. When the system receives remeasurement event data for a specific interface component, it parses the measurement accuracy attribute of the event. The measurement accuracy attribute is usually determined by the type of remeasurement equipment; for example, precision measurements using a total station have millimeter-level accuracy, while coarse measurements using a measuring tape have centimeter-level accuracy. The system calculates the remeasurement quality factor q based on the measurement accuracy. quality For example, the rule can be set as follows: if the retest accuracy acc mm If q is less than or equal to 1 millimeter, then quality The value is 1.0; if the remeasurement accuracy is greater than 1 mm and less than or equal to 5 mm, then q quality The value is 0.8; if the remeasurement accuracy is greater than 5 mm, then q quality The value is 0.6. The incremental update formula can be expressed as:

[0062] C dim_new = max( C dim_time C reset + γ reset * q quality );

[0063] Where C dim_new To assess the reliability of the updated version; C dim_time The baseline confidence level after attenuation; C reset Reset the baseline value for retesting, i.e., the lower limit of the baseline confidence level after retesting, for example, 0.7; γ reset The increase can be, for example, 0.3; q quality This is a quality factor based on measurement accuracy.

[0064] Continuing with the previous example, suppose the component was remeasured with an accuracy of 2 mm on the 181st day. At this point, the confidence level C after attenuation... dim_time Approximately 0.1485. Due to the accuracy of 2 millimeters, the corresponding q... qualityThe value is 0.8. Substituting this into the update formula, the final result is 0.94. This shows that high-precision retesting can instantly raise the confidence level of a component from an extremely low state to a high confidence level, realizing the model's real-time response to real physical events.

[0065] like Figure 2 As shown, according to one aspect of this application, a process block-structure block two-layer diagram comprising a process block layer and a structure block layer is constructed, including:

[0066] Acquire and aggregate regions with the same construction attributes to generate process block nodes. The construction attributes include pouring batch, formwork system and construction team.

[0067] In this embodiment, the generation of process block nodes is a process of logical clustering of the physical space. The system traverses the components of the entire boundary area and extracts the construction attributes of each component. The construction attributes include not only the pouring batch in the time dimension, but also the formwork system in the process dimension and the construction team in the organizational dimension. The system aggregates the set of components with the same three-dimensional attribute tuples (pouring batch ID, formwork system ID, team ID) into a single process block node. For example, for the east side area of ​​the second floor of the station building, although it contains dozens of specific beam and slab components, if they all belong to the 20240520 pouring batch, all use the aluminum formwork system, and are all constructed by Zhang San's team, then the system creates a process block node (Node) representing this area in the graph structure. Proc_A This simplifies computational complexity while preserving the core characteristic of error homology.

[0068] The physical space is divided based on the pre-stored structural boundary information of structural joints, expansion joints, and settlement joints, and structural block nodes are generated.

[0069] Specifically, the generation of structural block nodes is a process of physically dividing physical space. The system reads structural design information from the BIM model and identifies all structural joints, expansion joints, and settlement joints. These joints constitute physical discontinuities. Based on these boundary lines, the system divides the entire building space into several independent closed regions, generating a structural block node for each region. For example, a utility tunnel structure typically has an expansion joint every 30 meters. The system then divides the utility tunnel into structural block nodes accordingly. Struct_1 Node Struct_2 In a graph structure, structural block nodes represent the physical constraints governing error propagation.

[0070] Establish the topological mapping relationship between the boundary components and their respective process block nodes and structural block nodes, and generate a process block-structural block two-layer graph.

[0071] In this embodiment, the system constructs the final graph topology. For each specific boundary component node... Comp_iThe system creates an edge pointing to the node of its respective process block. Proc_i and an edge pointing to its constituent structural block node. Struct_i This makes each component node a connecting hub between the process logic layer and the physical structure layer, forming a complete process block-structure block two-layer diagram.

[0072] like Figure 3 As shown, in a further embodiment, configuring anisotropic propagation weights based on error mode characteristics includes:

[0073] Assign a first basic weight to the propagation path within the same process block node, and assign a second basic weight to the propagation path that crosses structural boundary information. The first basic weight is higher than the second basic weight.

[0074] In this embodiment, the allocation of basic weights reflects the physical resistance to error propagation. The system identifies two different types of propagation paths in the two-layer graph: one type connects different components under the same process block node, and the other type connects adjacent structural block nodes but crosses structural boundaries. The system assigns a higher first basic weight w to the former. base_proc For example, 1.0 means that the error propagates almost without loss within the same process system; a lower second basic weight w is assigned to the latter. baseboundary For example, 0.1 means that the error is difficult to penetrate physical gaps.

[0075] In some preferred embodiments, the system can also employ a unidirectional blocking strategy for critical structural boundaries such as settlement joints. Specifically, the system only allows confidence levels to propagate from the main structure side, where settlement control accuracy is higher, to the auxiliary structure side, while forcibly setting the weight of back propagation to 0. This asymmetric boundary treatment effectively prevents noise from low-precision areas from contaminating the model in high-precision areas.

[0076] Based on the first and second basic weights, the sensitive dimension of the error pattern feature association is analyzed, and the component of the anisotropic propagation weight on the sensitive dimension is increased.

[0077] Specifically, the weight configuration further considers the directionality of the error patterns. The system examines the error pattern characteristics of the components involved in the current propagation path. If the pattern characteristic indicates a specific sensitive dimension, such as the vertical dimension corresponding to the template upward movement pattern, a dimension adjustment coefficient k is introduced. dim The final propagation weight w e_d The calculation formula is:

[0078] w e_d = w base * k dim * k boundary ;

[0079] Where w e_d w represents the final propagation weight of edge e in dimension d. base The base weights are based on path type (intra-process / cross-boundary); k dim k is the dimension adjustment coefficient based on the error pattern. boundary This refers to the structural boundary blocking coefficient or boundary adjustment coefficient. For example, in the vertical dimension, if k... dim A value of 1.5 enhances the propagation capability of this dimension, allowing vertical errors to affect surrounding components from a greater distance.

[0080] In one embodiment of this application, spatial propagation of individual credibility is performed on a process block-structure block bilayer graph to generate a spatial credibility field, including:

[0081] Initialize the node state values ​​of each node in the process block-structure block bilayer graph with individual confidence; use anisotropic propagation weights to iteratively update the node state values ​​between adjacent nodes in the process block-structure block bilayer graph; calculate the maximum change in the node state values ​​updated between adjacent iterations; and in response to the maximum change being less than a preset convergence threshold, map the final node state values ​​to a spatial confidence field.

[0082] In this embodiment, spatial propagation is an iterative convergence process. The system assigns the individual confidence level to the corresponding component node in the process block-structure block bi-level graph. For example, in each iteration k, the new state value C of node i... i_k The state value C of its neighbor node j in the previous round j_k-1 Weighted composition. The formula can be expressed as:

[0083] C i_k =θ* C i_initial + (1 -θ) *Σ j∈N(i) ( w ji_norm * C j_k-1 );

[0084] Where C i_k Let C be the state value of node i in the k-th iteration; i_initial Let be the initial individual confidence level of node i; θ is the retention coefficient, representing the degree of confidence in its own observations; w ji_norm C represents the normalized propagation weights from node j to node i. j_k-1 Let $\frac{j}{i}$ be the state value of neighbor node $j$ in the previous iteration; $N(i)$ be the set of neighbor nodes of node $i$. After each iteration, the system calculates the maximum change in the state values ​​of all nodes in the entire graph, Δ$. max =max i |C i_k - C i_k-1 |。 When Δ maxLess than the preset convergence threshold ε conv When the value is 0.01 (e.g., 0.01), or when the number of iterations reaches the upper limit (e.g., 20 times), the iteration stops. The final converged node state value not only contains the observation information of the component itself, but also integrates the inferred information from the process association and structural association neighborhood, forming a continuous spatial credibility field.

[0085] In one possible embodiment, the individual confidence level is a multi-dimensional vector containing geometric location, size parameters, and safety margin dimensions; the method further includes:

[0086] Obtain a set of job types for boundary components and identify key credibility dimensions associated with specific job types within the set of job types.

[0087] In this embodiment, the system receives a set of operation types (Operation_set) to be executed, such as drilling reinforcement, slotting and pipe laying, and scaffold dismantling. The system has a built-in mapping rule table between operation types and reliability dimensions. For a specific operation type, the system identifies the key dimensions that have the greatest impact on its safety and quality. For example, for drilling through floor slabs, the key is to avoid reinforcing bars and ensure the verticality of the hole; therefore, the key reliability dimensions identified by the system include vertical safety margin reliability and the reliability of the relative position of the internal reinforcing bars. For slotting operations, the key is the accurate positioning of the slot; therefore, the key dimension includes the reliability of the planar offset direction.

[0088] Extract the components of individual credibility on the key credibility dimension and map them to generate job-related credibility for specific job types.

[0089] For example, the confidence level of a single entity is defined as a vector V that includes multi-dimensional components such as geometric location, size parameters, and safety margin. credibility Based on the identification results, i.e., the key credibility dimension, the corresponding components are extracted from the vector, and a task-related credibility score is generated through weighted combination. task The calculation formula can be expressed as:

[0090] Score task =Σ m=1 M (w m * V dim_m );

[0091] Among them, Score task Reliability scoring for specific task types; m V represents the attention weight of this task type to the m-th dimension; dim_m Let S be the component value of the m-th dimension in the individual credibility vector; M is the total number of key credibility dimensions involved in the individual credibility vector. For example, the credibility score of a drilling operation.drill This could be equal to 0.6 multiplied by the vertical safety margin component plus 0.4 multiplied by the relative position component of the reinforcing bars. Based on this, the system compares the generated job-related confidence score with the project management threshold. If the score... task If the value falls below a preset safety threshold (e.g., 0.85), the system will automatically generate a work stoppage warning and highlight the area where the component is located, prompting engineers to conduct a high-precision retest before proceeding with the work. This achieves closed-loop control from data model to on-site action.

[0092] In an exemplary embodiment, the method further includes: encapsulating the credibility data into a standardized service interface and visually representing it in a 3D BIM environment, specifically:

[0093] The individual credibility and spatial credibility fields are encapsulated into a credibility service interface dataset, and a multi-dimensional query mode is defined.

[0094] In this embodiment, after completing the spatial propagation calculation, the system performs a data encapsulation operation. Specifically, it integrates the multidimensional vector set of individual unit credibility with the spatial credibility field to construct a credibility service interface dataset. Each record in this dataset includes: a unique component identifier (Global_ID), a process block identifier (Block_ID), a structural block identifier (Struct_ID), and the multidimensional individual unit credibility vector V at the current moment. current and the propagation-corrected spatial field nodal state value V field For example, the system defines three standardized access modes to support external business calls. The first is a component-based query mode, where the external module inputs the component ID, and the interface returns the real-time credibility and risk level of that component. The second is a region-based query mode, where the external module inputs a 3D spatial bounding box, and the interface returns a heatmap data showing the credibility distribution of all components within that region. The third is a job-type-based query mode, where the external module inputs the job code (e.g., drilling), and the interface dynamically invokes the mapping logic to generate job-related credibility for a specific job type, returning a specific credibility score for that job type.

[0095] Based on spatial credibility field data, differentiated colors and transparency are rendered in a three-dimensional information model to achieve risk visualization.

[0096] Specifically, to visually demonstrate the risk distribution at the junction of the station building and the utility tunnel, the system incorporates a built-in visualization rendering engine. A mapping function between confidence values ​​and the RGBA color space is established. For example, the mapping rule for the confidence value C is set as follows: if C is greater than 0.9, it is rendered as green (safe) with a transparency of 0.8; if C is between 0.7 and 0.9, it is rendered as yellow (warning) with a transparency of 0.6; if C is less than 0.7, it is rendered as red (high risk) with a transparency of 0.3 (high opacity for emphasis). Spatial confidence field data is injected into the BIM browser in real time. In the 3D view, engineers can directly observe that although most areas appear green, the local utility tunnel interfaces crossing settlement joints appear dark red, and this red area shows a slight spreading trend along the process block direction. This transforms implicit algorithm results into explicit management views, effectively assisting in the planning of on-site inspection routes.

[0097] In one possible implementation, a method for modeling the reliability of components in a station building utility tunnel information model includes:

[0098] Acquire multi-source data and identify boundary components from the multi-source data to form a basic dataset of boundary components;

[0099] For each boundary component in the boundary component basic dataset, construction process data is integrated and the error knowledge base is queried to construct the construction process profile of the component to generate error pattern features, forming an extended multi-source feature vector set;

[0100] Based on the extended multi-source feature vector set, the conflict between multi-source data is explained to calculate the individual credibility of each boundary component, forming a multi-dimensional vector set of individual credibility.

[0101] Based on the multidimensional vector set of individual credibility and the topological and segmentation information in multi-source data, a two-layer graph containing process block layer and structural block layer is constructed, and the individual credibility is propagated anisotropically on the two-layer graph to generate a spatial credibility field.

[0102] In a preferred implementation, the method further includes: propagating credibility to identify error root cause components, specifically: constructing a process block layer and a structural block layer, and using each boundary component in the boundary component set as a boundary component node; connecting the boundary component nodes to process block nodes in the process block layer and structural block nodes in the structural block layer according to their respective process relationships and structural relationships, forming a two-layer graph; setting anisotropic propagation weights for connections between nodes in the two-layer graph based on the process logic relationships within the process block layer and the structural adjacency relationships within the structural block layer; propagating the initial credibility of each boundary component node in the two-layer graph based on the anisotropic propagation weights until the credibility values ​​of each node converge, and identifying the boundary component with the highest credibility after convergence as the error root cause component. The process block is formed by dividing the boundary components according to at least one construction grouping information identified in the construction process data, such as the same pouring batch, the same formwork system, or the same construction team; the structural block is formed by dividing the boundary components according to at least one structural boundary information identified in the design model in the multi-source data, such as structural joints, expansion joints, or settlement joints. Optionally, setting anisotropic propagation weights includes: setting a first basic weight for edges connecting component nodes within the same process block in the bilayer graph; setting a second basic weight for edges connecting component nodes in different process blocks in the bilayer graph, wherein the first basic weight is higher than the second basic weight; and setting a third basic weight lower than the second basic weight for edges connecting component nodes belonging to different structural boundaries in the bilayer graph. In a further implementation, setting anisotropic propagation weights also includes: identifying preset error pattern features of components associated with component nodes; and, when a specific error pattern feature is identified, selectively adjusting the propagation weights on specific dimensions of the multidimensional confidence vector related to that error pattern feature.

[0103] According to one aspect of this application, multi-source data is acquired, and boundary components are identified from the multi-source data to form a basic dataset of boundary components, including:

[0104] The design model and measured point cloud in the multi-source data are registered to generate registered multi-source data. Semantic and geometric association analysis is performed on the registered multi-source data to identify components with connection relationships between different professions or different construction stages as boundary components, forming a basic dataset of boundary components.

[0105] Furthermore, a construction sequence profile of the component is constructed to generate error mode features, including:

[0106] Based on construction process data, a construction sequence of construction procedures in time and space is established for each boundary component in the boundary component basic dataset, forming a construction procedure profile. The construction procedure profile is used to query the error knowledge base to obtain the potential error types associated with the construction procedure sequence. The potential error types and the geometric deviation of the boundary components in multi-source data are combined to form error pattern features.

[0107] In one embodiment of this application, calculating the credibility of a single entity can also involve: reading an extended multi-source feature vector set; for each boundary component, comparing the design geometry, as-built geometry, point cloud local geometry, manual measurement records, and document descriptions one by one; calculating conflict indicators such as planar deviation, vertical deviation, opening size differences, and temporal relationships; and, combined with the presence and absence of each data source, identifying various conflict modes such as only design without as-built, as-built but without point cloud, and significant directional deviation between point cloud and manual measurement. The difference information originally scattered across different fields is abstracted into a structured conflict mode description, generating a conflict mode feature set, which provides features such as the difference type, direction, and magnitude between the multiple sources for each boundary component, which can be used for subsequent interpretation. Simultaneously, the extended multi-source feature vector set and the conflict mode feature set are read; for each boundary component, process features and error mode features are first extracted from the extended multi-source feature vector set, and then these error modes are matched with the geometric conflict modes in the conflict mode feature set. For example, when the process characteristics indicate that the component has undergone post-excavation construction and is located in a densely reinforced area, if the conflict mode shows that the design opening and the point cloud opening have a large deviation in the planar direction but a small vertical deviation, the system interprets this conflict as the planar position of the design drawing being invalid, and the point cloud being more representative of the actual position. In another type of process characteristic, such as the overall upward movement of the formwork and the consistency between the as-built drawing and the design height, if the conflict mode shows a large difference between the point cloud and the manually measured height, the system prioritizes the assumption that a certain manual measurement has a large error. Through the linked interpretation of error mode and conflict mode, the reliability of different data sources such as design drawings, as-built drawings, point clouds, and manual measurements is directionally corrected, obtaining the reliability weights of each data source in multiple dimensions such as planar position, vertical position, and opening size, forming a source reliability correction result set.

[0108] The system reads the extended multi-source feature vector set, the source reliability correction result set, the time information for which current reliability needs to be evaluated, and the historical reliability records and their update event set. Utilizing the reliability weights of each data source in different dimensions within the source reliability correction result set, it weights and integrates the original measurement and calculation indicators from the extended multi-source feature vector set, such as geometric location, opening size, vertical safety margin, and planar offset direction, to obtain the initial individual reliability value for each boundary component in each dimension. Combining the historical retesting time, retesting accuracy level, and historical accident events recorded in the historical reliability records and their update event set, and using the time information for which current reliability needs to be evaluated as a benchmark, the system performs a time-series correction on the initial reliability value: for components that have not been retested for a long time and whose construction error mode is judged to be high-risk, the system applies a faster reliability decay in the relevant dimensions; for components that have recently undergone high-precision point cloud retesting or rigorous manual verification, the system increases their reliability value in the corresponding dimensions and appropriately slows down the subsequent decay rate. We obtain multidimensional initial confidence values ​​that reflect both the differences between multiple data sources and the evolution over time and the behavior of repeated testing, thus forming a multidimensional initial confidence value set for individual data.

[0109] The system reads the initial set of multidimensional unit reliability values ​​and the set of job types. For each job type, it defines the most relevant combination of reliability dimensions. For example, for the job type of drilling through a floor slab, the system focuses on the reliability of vertical safety margin, the reliability of opening size, and the reliability of distance from the floor slab edge; for grooving or partial cutting, it focuses more on the reliability of planar offset direction and the reliability of relative position with surrounding steel bars / pipelines. Based on these job-dimensional mapping rules, the system performs weighted combination and nonlinear processing of the reliability of each dimension in the initial set of multidimensional unit reliability values ​​according to job type to obtain the job-related unit reliability of each boundary component under each job type. When there are recent re-measurement or recent accident records, the system can also incrementally adjust the reliability of the corresponding job type according to the specific job events recorded in the historical reliability records and their update event sets. For example, for components that have experienced drilling deviation accidents, additional reductions are set on the reliability of the same type of job. Through the above mapping and updating, a multidimensional vector set of unit reliability is generated, which contains the multidimensional unit reliability of each boundary component and the job-related reliability for each job type.

[0110] In another embodiment of this application, generating a spatial credibility field can further involve: reading geometric and structural topology information, process segmentation information such as construction section division and pouring section, auxiliary data related to the division of process blocks and structural blocks, and basic dataset of boundary components. Using the process segmentation information such as construction section division and pouring section, and information such as pouring zones, formwork support range, and construction team responsibility areas in the auxiliary data related to the division of process blocks and structural blocks, the station building floor slab, shear wall, and pipe gallery structure are divided into multiple process blocks, and each process block is used as a node of the process block layer. Based on the structural topology information in the geometric and structural topology information, the floor slab is divided into structural blocks according to structural joints, expansion joints, settlement joints, etc., and the pipe gallery is divided according to structural segments and mileage segments, forming nodes of the structural block layer. Using the spatial location and structural information of each boundary component in the basic dataset of boundary components, each boundary component is connected to its corresponding process block and structural block, thereby forming a multi-layer graph structure among the process block layer, structural block layer, and component layer, generating a two-layer graph structure of process blocks and structural blocks.

[0111] The system reads the multidimensional vector set of individual credibility and the two-layer graph structure of the process block. It iterates through each boundary component node in the two-layer graph structure, finding the multidimensional individual credibility and operation-related credibility of the corresponding component in the multidimensional vector set of individual credibility, and binding these credibility values ​​as attributes to the corresponding component nodes. Based on the overall distribution of the credibility of boundary components within the same process block, it calculates the credibility overview of each process block in different dimensions. For example, if most components within a process block have low credibility in terms of vertical safety margin, this feature is written into the process block node. Similarly, the system performs statistics on the structure block layer, aggregating the credibility distribution of components within the same structure block into the initial credibility attribute of the structure block. Each layer node in the two-layer graph carries the initial credibility state, forming a two-layer graph state with credibility attributes.

[0112] The system reads the state of the two-layer graph with confidence attributes and expands the multi-source feature vector set. For edges within the process block layer, based on information such as the same pouring batch, the same formwork system, and the same construction team, a larger propagation weight is assigned to the propagation edges between components within the same process block. This reflects the fact that overall formwork offset or team construction habits can produce similar error patterns within the same process block. Within the structural block layer, based on structural continuity and stiffness coupling, a medium propagation weight is assigned to edges between different process blocks within the same structural block. This indicates that while there is some possibility of error propagation due to structural continuity but different processes, it is not as strong as within the process block itself. For edges crossing structural boundaries such as structural joints, expansion joints, and settlement joints, the system sets the propagation weight to a lower value based on geometric and structural topology information and the boundary type identified in auxiliary data related to the division of process blocks and structural blocks. In some cases, it even allows only unidirectional propagation from areas with higher precision control to areas with weaker control, prohibiting reverse propagation to prevent low-confidence areas from contaminating high-confidence areas. By combining error pattern features from the extended multi-source feature vector set, further adjustments are made to the dimensional propagation corresponding to certain specific error patterns (e.g., overall template upward movement). Under these patterns, the propagation weight in the vertical safety margin dimension along the pouring direction is enhanced, while the propagation in irrelevant dimensions is weakened. Anisotropic propagation weight configurations are generated to provide propagation coefficients and directional constraints for subsequent confidence propagation on the two-layer graph.

[0113] The system reads the state of the two-layer graph with credibility attributes and the anisotropic propagation weight configuration. Using the multidimensional individual credibility of component-level nodes as initial values, it iteratively exchanges credibility information between the process block layer and the structural block layer. When a process block node is determined to have a significant error pattern and the credibility of its internal components is generally low in a certain dimension, the system, based on the anisotropic propagation weight configuration, propagates this low credibility trend along the internal edges of the process block and its associated component nodes, causing a certain degree of credibility reduction in that dimension for components in the same process block that have not yet been retested. Conversely, when some components are confirmed to have high credibility through high-precision retesting, the system propagates positive correction along the structural continuity direction within their respective structural block layer, causing a certain degree of improvement in the relevant dimensions for adjacent components with similar error patterns. Throughout multiple iterations, the system consistently follows the anisotropic propagation weights determined by process relationships, structural boundaries, and error pattern characteristics, avoiding excessive diffusion across structural seams or unrelated error pattern regions until the overall credibility distribution converges or the preset number of iterations is reached. The converged node credibility distribution is mapped back to the boundary components and their surrounding spatial locations to form a spatial credibility field, providing a fine-grained, multi-dimensional credibility spatial distribution for the station building-utility corridor boundary area.

[0114] In summary, the modeling method for component credibility in the station building utility tunnel information model includes: generating error pattern features based on the construction process profile of the boundary components, using these features to provide causal explanations for conflicts between multi-source data, and calculating the individual credibility of the components through asymmetric correction; constructing a two-layer graph structure containing process block layers and structural block layers, mapping the boundary components to process logic nodes and physical structure nodes respectively; configuring anisotropic propagation weights according to the error pattern features, performing spatial propagation of individual credibility on the two-layer graph, and generating a spatial credibility field.

[0115] This invention addresses the problem of error propagation across physical boundaries by constructing a two-layer diagram of process blocks and structural blocks, along with an anisotropic propagation mechanism. By setting the propagation weights to be internal to cross-process > cross-structural boundaries and introducing a unidirectional blocking strategy, it successfully simulates the real propagation law of errors in the physical world constrained by construction batches and structural joints, avoiding the risk assessment distortion caused by traditional geometric smoothing algorithms. Simultaneously, through error pattern-driven conflict interpretation and asymmetric correction algorithms, it solves the problem of being unable to distinguish between systematic deviations and random noise. Specifically, by introducing an error knowledge base to match construction process profiles, it avoids blindly averaging conflict data. Instead, it differentiates (asymmetrically) corrects the reliability of design sources and point cloud sources based on physical causes (such as post-excavation construction and formwork offset), thereby giving the credibility calculation process clear physical interpretability and enhancing the guiding value of the evaluation results for engineering decisions.

[0116] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for modeling the reliability of components in a station building utility tunnel information model, characterized in that, include: Obtain and generate error pattern features based on the construction sequence profile of the boundary component, use the error pattern features to parse the conflict patterns between pre-stored multi-source data, and calculate the individual credibility of the boundary component. Construct a process block-structure block bilayer graph that includes a process block layer and a structure block layer, and associate the boundary components with the process block nodes and structure block nodes in the process block-structure block bilayer graph respectively; Based on the error mode characteristics, anisotropic propagation weights are configured, and spatial propagation of individual credibility is performed on the process block-structure block bilayer graph to generate a spatial credibility field.

2. The method according to claim 1, characterized in that, Construct a two-layer process block-structure block diagram, including a process block layer and a structure block layer, as follows: Acquire and aggregate regions with the same construction attributes to generate process block nodes. The construction attributes include pouring batch, formwork system and construction team. The physical space is divided based on the pre-stored structural boundary information of structural joints, expansion joints, and settlement joints, and structural block nodes are generated. Establish the topological mapping relationship between the boundary components and their respective process block nodes and structural block nodes, and generate a process block-structural block two-layer graph.

3. The method according to claim 2, characterized in that, Configure anisotropic propagation weights based on error mode characteristics, including: Assign a first basic weight to the propagation path within the same process block node, and assign a second basic weight to the propagation path that crosses structural boundary information. The first basic weight is higher than the second basic weight. Based on the first and second basic weights, the sensitive dimension of the error pattern feature association is analyzed, and the component of the anisotropic propagation weight on the sensitive dimension is increased.

4. The method according to claim 1, characterized in that, Calculating the individual reliability of the interface component includes: Quantify the differences in geometric location, size parameters, and document descriptions of multi-source data to generate conflict patterns; Based on the conflict pattern, obtain the source reliability adjustment parameters corresponding to the error pattern features in the preset error knowledge base; The source reliability weights of each data source in the multi-source data are asymmetrically corrected using source reliability adjustment parameters, and the individual credibility is calculated based on the corrected source reliability weights.

5. The method according to claim 4, characterized in that, Calculating the individual reliability of boundary components also includes: Determine the risk level associated with error pattern characteristics, and perform time decay calculation on the individual credibility based on the risk level and the time interval since the last assessment; In response to a retest event for a boundary component, the individual unit confidence level after execution time decay calculation is incrementally updated based on the measurement accuracy attribute of the retest event.

6. The method according to claim 1, characterized in that, Generate error pattern features, including: Acquire and integrate construction process data and structural segmentation information to reconstruct the construction sequence profile of the junction components. The construction sequence profile records the pouring section to which the junction components belong, the formwork system type, and the construction team attributes. The system queries a pre-defined error knowledge base and matches the construction process profile with typical process features in the error knowledge base. Based on the matching results, the corresponding error mode features are determined for the boundary components. The error mode features are defined with typical geometric deviation trends under the process combination.

7. The method according to claim 1, characterized in that, The individual confidence level is a multi-dimensional vector containing geometric location, size parameters, and safety margin dimensions; the method further includes: Obtain a set of job types for boundary components, and identify key credibility dimensions associated with specific job types in the set of job types; Extract the components of individual credibility on the key credibility dimension and map them to generate job-related credibility for specific job types.

8. The method according to claim 1, characterized in that, Generate a spatial credibility field, including: Initialize the node state values ​​of each node in the process block-structure block bilayer diagram with the individual node credibility; The node state values ​​are iteratively updated between adjacent nodes in the process block-structure block bilayer graph using anisotropic propagation weights. Calculate the maximum change in node state values ​​between adjacent iterations. If the maximum change is less than a preset convergence threshold, map the final node state values ​​to a spatial credibility field.

9. The method according to claim 4, characterized in that, Obtain the source reliability adjustment parameters corresponding to the error pattern characteristics in the preset error knowledge base, including: Analyze the typical conflict features corresponding to the error pattern characteristics; Calculate the conflict similarity between conflict patterns and typical conflict characteristics; If the conflict similarity is greater than a preset matching threshold, the error pattern feature is determined to be successfully matched, and the source reliability adjustment parameters associated with the error pattern feature are extracted.

Images