Multi-source heterogeneous data fusion and treatment method for steel box girder design and manufacturing

Abstract

The invention relates to the technical field of intelligent manufacturing and data management of steel box girders, and provides a multi-source heterogeneous data fusion and management method for design and manufacturing of a steel box girder. According to the method, an intelligent parameter relationship network is constructed, the relationship strength between entities is calculated by using a standardized co-occurrence strength formula Sij = Nij / square root of (Ni.Nj), and automatic association and parameter binding are realized by using a threshold value 0.7; a graded early warning mechanism is introduced, first-level early warning faces data layer normalization and physical constraint verification, and second-level early warning adopts a risk index model R = 0.6 * (deformation / maximum deformation) + 0.4 * (stress / yield strength), when Rgt; and 0.8, alarming and handling are triggered. A dynamic limiting and shrinking model with allowable thickness error = standard error * (1-0.15 * strength deviation ratio) is adopted as a numerical value checking rule, and material strength-deformation sensitivity demonstration is taken as a basis. According to the method, the problems of inconsistent parameter definition, failure of a fixed threshold value, insufficient change risk pre-judgment, difficulty in experience reuse and the like in steel box girder design and manufacturing are solved, and the data consistency and decision reliability are improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing and data governance technology for steel box girders, specifically involving a method for the fusion and governance of multi-source heterogeneous data in the design and manufacturing of steel box girders. Background Technology

[0002] The multi-source data of steel box girders from design to manufacturing, quality inspection and operation and maintenance have long been scattered in heterogeneous systems, resulting in problems such as data silos, semantic inconsistencies and caliber deviations, which leads to low efficiency of cross-stage comparison and difficulty in timely blocking of error propagation.

[0003] Existing methods often rely on a single rule system or static knowledge base, making it difficult to simultaneously address both correlation discovery and conflict resolution. When faced with discrepancies between design parameters and actual production data, they often lack a reliable optimal decision-making mechanism.

[0004] Fixed threshold control methods are difficult to adapt to the differentiated requirements of different materials, component parts and working conditions, and are prone to missed inspections in important locations or excessive warnings in minor locations, affecting quality and schedule.

[0005] The fragmentation and difficulty in reusing experience and knowledge, coupled with insufficient structured accumulation of historical cases, lead to the recurrence of similar problems in different projects, making it difficult to form effective organizational learning and continuous improvement.

[0006] Based on this, the present invention proposes a method for multi-source heterogeneous data fusion and governance for steel box girder design and manufacturing to solve the above problems. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for multi-source heterogeneous data fusion and governance for steel box girder design and manufacturing, so as to solve the problems existing in the background art.

[0008] This invention provides the following technical solution: a method for multi-source heterogeneous data fusion and governance in the design and manufacturing of steel box girders, comprising the following steps:

[0009] S1: Construct an intelligent parameter relationship network, perform solid modeling and association of geometric parameters, material properties and process parameters, calculate the relationship strength Sij=Nij / √(Ni·Nj), and trigger automatic association and parameter binding when Sij>0.7;

[0010] S2: Set up tiered early warning rules, perform data layer standardization and physical constraint verification, and configure model layer risk assessment;

[0011] S3: When data is accessed or design changes are made, numerical inspection rules are executed, and the deviation of critical dimensions is dynamically limited according to the allowable thickness error = standard error × (1 - 0.15 × strength deviation rate);

[0012] S4: Situation assessment is performed based on the risk index R = 0.6·(deformation / maximum deformation) + 0.4·(stress / yield strength). When R > 0.8, an alarm is triggered and the conflict handling process is initiated.

[0013] S5: Trigger closed-loop optimization for abnormal data, call the historical case library for similarity matching, generate parameter revision suggestions and form a version record;

[0014] S6: Incrementally update the knowledge graph according to the principle of strong correlation, and keep irrelevant nodes frozen to achieve accurate resource allocation and robust knowledge evolution;

[0015] S7: Feed the optimization results back into the rule engine and standard library to complete the closed loop from anomaly detection to standard iteration.

[0016] Furthermore, the intelligent parameter relationship network includes the linkage between geometric parameters and process constraints. The process constraints include at least mold processing capacity, welding heat input and assembly tolerance. When the physical limits are exceeded, a hard constraint alarm is triggered and the process is terminated.

[0017] Furthermore, the coefficient 0.15 in the numerical inspection rule is determined based on the material strength-deformation sensitivity matrix and is used to map the material strength deviation to the thickness error limiting ratio.

[0018] Furthermore, the tiered early warning system includes:

[0019] Level 1 warning: Verify data format consistency, unit conversion accuracy, and physical boundary constraints;

[0020] Level 2 warning: Dynamic assessment is performed using the aforementioned risk index model. If R > 0.8, an alarm is triggered and the process enters a consultation or optimization phase.

[0021] Furthermore, the closed-loop optimization includes similarity matching between abnormal scenarios and the case library. When the matching degree is greater than or equal to a preset threshold, it enters the fast track to generate a revised solution; otherwise, it enters the small sample simulation and expert review process.

[0022] Furthermore, incremental updates to the knowledge graph only adjust the weights or revise the rules for nodes and edges that are strongly correlated with the abnormality, while uncorrelated nodes remain frozen to save resources and maintain stability.

[0023] Furthermore, the dynamic rule engine includes both hard constraint rules and soft constraint rules. Hard constraints are used to intercept operating conditions that exceed the physical limits of the equipment or materials, while soft constraints are used to perform dynamic compensation or threshold adjustment based on the material grade (including Q345, Q420 and above) and the importance of the component.

[0024] Furthermore, when the system detects that the design camber exceeds the mold's maximum processing capacity or the risk index exceeds the threshold, it automatically generates rectification suggestions, including strategies such as mold replacement, segmented design, and compensation joint setting, and forms a controlled version.

[0025] Furthermore, the knowledge graph relationship strength threshold and risk threshold can be adaptively updated according to the project scenario, and the update strategy is driven by closed-loop evaluation indicators.

[0026] Furthermore, the method can be executed by a device including a processor and a memory, the memory storing computer program instructions that, when executed by the processor, cause it to perform the method according to any one of claims 1 to 9.

[0027] The technical effects and advantages of this invention are as follows:

[0028] Enhanced association discovery capabilities: Stable automatic association is achieved using the strength metric Sij = Nij / √(Ni·Nj) and a threshold of 0.7, reducing manual configuration costs;

[0029] More sensitive risk identification: The introduction of a risk index R comprehensively considers the proportion of deformation and stress, which has better adaptability to working conditions than a fixed threshold.

[0030] More reasonable precision control: Based on the dynamic limiting model of thickness error due to material strength deviation, the process consistency of critical dimensions is improved;

[0031] Controllable knowledge evolution: The strategy of incremental updating of strongly related nodes and freezing of unrelated nodes is adopted to ensure the stability of the knowledge base and the efficiency of computing power.

[0032] Improved closed-loop governance: Anomaly—matching—revision—reinjection form a traceable, end-to-end closed loop, supporting the accumulation and reuse of organizational experience. Attached Figure Description

[0033] Figure 1 A schematic diagram illustrating the data flow and interaction relationships between the various functional modules of this invention is provided.

[0034] Figure 2 The complete flowchart of this invention, from data access to closed-loop governance, is shown. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0036] Example 1: Overall Process of Dual-Driven Data Governance

[0037] Step 1: Intelligent Parameter Relationship Network Construction. Entity extraction and synonym normalization are performed on geometric parameters (such as radius of curvature, camber), material properties (such as yield strength, grade), and process parameters (such as welding current, heat input, assembly tolerance). The correlation strength is calculated as Sij = Nij / √(Ni·Nj). When Sij > 0.7, automatic correlation and parameter binding are established, and edge weights and version numbers are recorded.

[0038] Step 2: Tiered Early Warning Configuration. Level 1 early warning performs data standardization and physical boundary checks; Level 2 early warning uses R = 0.6 * (deformation / maximum deformation) + 0.4 * (stress / yield strength) for risk assessment, and R > 0.8 triggers an alarm and process escalation.

[0039] Step 3: Numerical Inspection and Dynamic Limitation. For critical dimensions, a dynamic limitation is applied: allowable thickness error = standard error × (1 - 0.15 × strength deviation rate). The strength deviation is determined based on incoming material inspection and batch statistics.

[0040] Step 4: Closed-loop optimization and version management. When an abnormal event occurs (such as welding parameters exceeding limits or camber deviation exceeding standards), the system calls the case library for similarity matching. If the matching degree is greater than or equal to the threshold, it enters the fast track to automatically generate a revision proposal; otherwise, it enters a small-sample simulation and initiates an expert review meeting. After the revision is approved, only the graph nodes and edges strongly correlated with the abnormality are updated, and a version record is generated.

[0041] Step 5: Knowledge Feedback and Rule Iteration. The revision results are written back to the rule engine and standard library, enabling the model and rules to form an adaptive iteration.

[0042] Example 2: Synergistic Application of Hard and Soft Constraints

[0043] Hard constraint rule: When the design camber exceeds the maximum processing capacity of the mold, the system will immediately intercept and prompt for feasible strategies such as replacing the mold or segmented design, forming a mandatory rectification task order.

[0044] Soft constraint rule: For high-strength steel (such as Q420 and above) components, dynamic compensation and threshold adjustment are performed on thickness, tolerance, etc. based on importance level; if the strength deviation is 5%, the upper limit of thickness error is reduced by 15%.

[0045] Example 3: Conflict Classification and Handling

[0046] Level 1 Handling (Automatic Compensation): When a material strength deviation is detected, dynamic limiting is triggered, and the welding process parameter library is automatically linked and a correction parameter package is issued.

[0047] Secondary handling (consultation and decision-making): When R>0.8 or a hard constraint conflict occurs, an inter-departmental consultation is initiated to form a technical comparison report that includes alternative solutions such as segmented manufacturing and compensation joint settings.

[0048] Level 3 Response (Intelligent Inference): When the historical matching degree is lower than the threshold, conduct relationship network inference, risk labeling and experimental data package generation, and initiate expert review; after passing the review, write back the knowledge graph and freeze irrelevant nodes.

[0049] Example 4: Key Points of Data and System Implementation

[0050] Data Acquisition: Structured acquisition of design and process parameters, parsing of process annotations in unstructured data such as CAD, and establishment of unified data standards.

[0051] Specifications: Deformation is based on the maximum value of actual measurement points, and the maximum deformation is taken as the design allowable limit; stress is based on actual measurement or simulation results, and yield strength is taken according to the batch of materials.

[0052] Similarity matching: Case matching can use a multi-feature weighted cosine similarity / learned retrieval model, and set a threshold for entering the fast track (e.g., ≥85%).

[0053] Versioning and Retrospection: Generates a version number, a list of differences, and the scope of effect for each rule or parameter revision, supporting rollback and auditing.

[0054] Example 5: Operational Device Form

[0055] The method can be implemented by a computing device, including a processor, a memory, and a communication interface; the memory stores program instructions, and the processor executes the instructions to implement the steps described in claims 1 to 9. The system functional modules include: a data access module, a knowledge graph module, a rule engine module, a risk assessment module, a case library and version library, a closed-loop optimization module, and a visualization interface.

[0056] Optional parameters and thresholds

[0057] The correlation strength threshold of 0.7 and the risk threshold of 0.8 can be calibrated based on the project-level historical performance and acceptance criteria.

[0058] The dynamic constraint coefficient of 0.15 can be adaptively adjusted through regression or Bayesian updates based on the material system and process capabilities.

[0059] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for multi-source heterogeneous data fusion and governance in the design and manufacturing of steel box girders, characterized in that, Includes the following steps: S1: Construct an intelligent parameter relationship network, perform solid modeling and association of geometric parameters, material properties and process parameters, calculate the relationship strength Sij=Nij / √(Ni·Nj), and trigger automatic association and parameter binding when Sij>0.7; S2: Set up tiered early warning rules, perform data layer standardization and physical constraint verification, and configure model layer risk assessment; S3: When data is accessed or design changes are made, numerical inspection rules are executed, and the deviation of critical dimensions is dynamically limited according to the allowable thickness error = standard error × (1 - 0.15 × strength deviation rate); S4: Situation assessment is performed based on the risk index R = 0.6·(deformation / maximum deformation) + 0.4·(stress / yield strength). When R > 0.8, an alarm is triggered and the conflict handling process is initiated. S5: Trigger closed-loop optimization for abnormal data, call the historical case library for similarity matching, generate parameter revision suggestions and form a version record; S6: Incrementally update the knowledge graph according to the principle of strong correlation, and keep irrelevant nodes frozen to achieve accurate resource allocation and robust knowledge evolution; S7: Feed the optimization results back into the rule engine and standard library to complete the closed loop from anomaly detection to standard iteration.

2. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: The intelligent parameter relationship network includes the linkage between geometric parameters and process constraints. The process constraints include at least mold processing capacity, welding heat input and assembly tolerance. When the physical limits are exceeded, a hard constraint alarm is triggered and the process is stopped.

3. The method according to claim 1, characterized in that: The coefficient 0.15 in the numerical inspection rule is determined based on the material strength-deformation sensitivity matrix and is used to map the material strength deviation to the thickness error limiting ratio.

4. A method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1 or 2, characterized in that: The tiered early warning system includes: Level 1 warning: Verify data format consistency, unit conversion accuracy, and physical boundary constraints; Level 2 warning: Dynamic assessment is performed using the aforementioned risk index model. If R > 0.8, an alarm is triggered and the process enters a consultation or optimization phase.

5. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: Closed-loop optimization involves similarity matching between abnormal scenarios and the case library. When the matching degree is greater than or equal to a preset threshold, it enters the fast track to generate a revised solution; otherwise, it enters the small sample simulation and expert review process.

6. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: Incremental updates to the knowledge graph only adjust the weights or revise the rules for nodes and edges that are strongly correlated with the graph. Uncorrelated nodes remain frozen to save resources and maintain stability.

7. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: The dynamic rule engine includes both hard constraint rules and soft constraint rules. Hard constraints are used to intercept operating conditions that exceed the physical limits of the equipment or materials, while soft constraints are used to perform dynamic compensation or threshold adjustment based on the material grade (including Q345, Q420 and above) and the importance of the component.

8. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: When the system detects that the design camber exceeds the mold's maximum processing capacity or the risk index exceeds the threshold, it automatically generates rectification suggestions, including strategies such as mold replacement, segmented design, and compensation joint settings, and forms a controlled version.

9. The method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: The knowledge graph relationship strength threshold and risk threshold can be adaptively updated according to the project scenario, and the update strategy is driven by closed-loop evaluation indicators.

10. A method for multi-source heterogeneous data fusion and governance in steel box girder design and manufacturing according to claim 1, characterized in that: The method can be executed by a device including a processor and a memory, the memory storing computer program instructions that, when executed by the processor, cause it to perform the method according to any one of claims 1 to 9.

Images