Knowledge graph-based composite shock-absorbing structure construction procedure optimization method and system

Abstract

The application discloses a composite shock-absorbing structure construction procedure optimization method and system based on a knowledge graph, and belongs to the technical field of intelligent construction, wherein the method comprises the following steps: constructing a shock-absorbing structure construction procedure knowledge graph; performing rule reasoning filtering on the current engineering structure parameters on the shock-absorbing structure construction procedure knowledge graph; matching a candidate construction procedure scheme set from a composite shock-absorbing structure engineering case library according to the current engineering structure parameters; constructing a feasible construction procedure scheme space according to an initial construction procedure scheme set and the candidate construction procedure scheme set; performing global search optimization on the feasible construction procedure scheme space, determining a target construction procedure optimization scheme, and performing composite shock-absorbing structure construction control through the target construction procedure optimization scheme. The application solves the technical problems that, in the prior art, a construction procedure scheme depends on manual experience and is difficult to integrate scattered knowledge for multi-target collaborative optimization, thereby resulting in low scheme generation efficiency and the inability to take into account multiple targets such as construction period and quality.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction technology, specifically to a method and system for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs. Background Technology

[0002] With the continuous improvement of seismic fortification standards for building engineering, composite damping structures are widely used in engineering scenarios such as high-rise buildings and large-span public buildings. Their construction process is long and the time and space constraints are complex. The rationality of the process plan directly determines the construction period, cost input and structural quality. Process optimization is the core link in the management and control of damping structure engineering.

[0003] Currently, the construction procedures for composite vibration reduction structures largely rely on manual experience, resulting in low efficiency, poor fault tolerance, and low knowledge reusability. Existing optimization methods struggle to systematically integrate construction knowledge across the field, fail to balance procedural logic compliance with engineering practice experience, and lack global optimization capabilities under multiple constraints, making it difficult to adapt to the customized optimization needs of differentiated projects. Summary of the Invention

[0004] This application provides a knowledge graph-based method and system for optimizing the construction process of composite vibration reduction structures. It aims to solve the technical problems in the prior art where construction process schemes rely on manual experience and are difficult to integrate scattered knowledge for multi-objective collaborative optimization, resulting in low scheme generation efficiency and an inability to balance multiple objectives such as construction period and quality.

[0005] In view of the above problems, this application provides a method and system for optimizing the construction process of composite vibration reduction structures based on knowledge graphs.

[0006] The first aspect disclosed in this application provides a method for optimizing the construction procedures of composite vibration-damping structures based on knowledge graphs. This method includes: extracting entities and modeling knowledge based on construction procedure information of the composite vibration-damping structure to construct a knowledge graph of construction procedures for the vibration-damping structure; inputting current engineering structural parameters and performing rule-based reasoning filtering on the knowledge graph of construction procedures for the vibration-damping structure based on the current engineering structural parameters to obtain an initial set of construction procedure schemes; matching candidate sets of construction procedure schemes from a composite vibration-damping structure engineering case library according to the current engineering structural parameters; constructing a feasible construction procedure scheme space based on the initial set of construction procedure schemes and the candidate sets of construction procedure schemes; performing a global search and optimization on the feasible construction procedure scheme space to determine a target optimized construction procedure scheme; and controlling the construction of the composite vibration-damping structure through the target optimized construction procedure scheme.

[0007] Another aspect of this application discloses a knowledge graph-based system for optimizing the construction procedures of composite vibration-damping structures. This system includes: a knowledge graph construction module for extracting entities and modeling knowledge based on construction procedure information of the composite vibration-damping structure to construct a knowledge graph of construction procedures for the vibration-damping structure; a rule-based reasoning and filtering module for inputting current engineering structural parameters and performing rule-based reasoning and filtering on the knowledge graph of construction procedures for the vibration-damping structure based on these parameters to obtain an initial set of construction procedure schemes; a scheme space construction module for matching candidate construction procedure schemes from a composite vibration-damping structure engineering case library according to the current engineering structural parameters, and constructing a feasible construction procedure scheme space based on the initial set of construction procedure schemes and the candidate set of construction procedure schemes; and a construction procedure optimization module for performing a global search and optimization of the feasible construction procedure scheme space to determine a target optimized construction procedure scheme, and controlling the construction of the composite vibration-damping structure through the target optimized construction procedure scheme.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: This paper adopts a technical solution that integrates knowledge graph construction, rule-based reasoning and filtering, case matching, and multi-objective global optimization. By extracting entities from construction process information to construct a knowledge graph, an initial solution set is generated on the graph using rule-based reasoning based on current engineering parameters. The solution set is then matched with a case library to construct a feasible solution space. Finally, the final process solution is determined through multi-objective function evaluation and global search optimization. This solution solves the technical problems in existing technologies where construction process solutions rely on human experience and are difficult to integrate scattered knowledge for multi-objective collaborative optimization, resulting in low solution generation efficiency and an inability to balance multiple objectives such as schedule and quality. The solution achieves the technical effect of automatically generating logically compliant and optimally efficient composite vibration reduction structure construction process solutions, thereby improving the efficiency of construction decision-making and the level of multi-objective collaborative control.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 A flowchart illustrating the method for optimizing the construction process of a composite vibration reduction structure based on a knowledge graph is provided for embodiments of this application.

[0011] Figure 2 A schematic diagram of a knowledge graph-based composite vibration reduction structure construction process optimization system is provided for embodiments of this application.

[0012] Figure labeling: Knowledge graph construction module 11, rule reasoning and filtering module 12, solution space construction module 13, construction procedure optimization module 14. Detailed Implementation

[0013] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0014] The overall concept of the technical solution provided in this application is as follows: This application provides a method and system for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs. A procedure knowledge graph is constructed from construction texts. An initial set of solutions is generated by rule-based reasoning on the graph according to current engineering parameters. Simultaneously, similar cases are retrieved from a historical case library as candidate solutions. These two sets are merged to form a feasible solution space. A global search and optimization of the solution space is performed using a multi-objective function, automatically outputting the optimal construction procedure solution that balances schedule, cost, and quality.

[0015] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0016] Example 1, as Figure 1 As shown in the embodiments of this application, a method for optimizing the construction process of composite vibration reduction structures based on knowledge graphs is provided. The method includes: Step S100: Based on the construction process information of the composite damping structure, entity extraction and knowledge modeling are performed to construct a knowledge graph of the construction process of the damping structure.

[0017] Furthermore, entity extraction and knowledge modeling are performed based on the construction process information of the composite damping structure to construct a knowledge graph of the construction process of the damping structure. This includes: performing text cleaning on the construction process information of the composite damping structure to obtain a structured construction process text library, wherein the text cleaning steps include non-text noise removal, sentence segmentation, and terminology standardization; constructing a dual-channel entity extraction strategy, wherein the dual-channel entity extraction strategy consists of a rule matching channel and a semantic annotation channel executed in series; extracting entities from the structured construction process text library according to the dual-channel entity extraction strategy to obtain a set of construction knowledge entities; and performing semantic relationship analysis and knowledge modeling output based on the set of construction knowledge entities to construct a knowledge graph of the construction process of the damping structure.

[0018] Specifically, the dual-channel entity extraction strategy refers to an entity recognition architecture composed of a rule-matching channel and a semantic annotation channel. First, rule matching quickly recalls candidate entities to ensure coverage, and then semantic annotation refines entity attributes to ensure accuracy, achieving a balance between extraction efficiency and precision. The rule-matching channel is a processing link that quickly locates candidate entities from text through string matching based on a pre-defined database of construction domain entity types and keywords. Its core function is to filter out the candidate entity range with high recall. The semantic annotation channel is a processing link that uses natural language semantic understanding technology to perform semantic recognition, classification, and attribute annotation on candidate entities. It is used to clarify the semantic features and specific classification of entities and correct deviations in rule matching. The construction knowledge entity set refers to the summary of all standardized knowledge units extracted from construction process text, including four core entity categories: processes, materials, constraints, and attributes. Each entity category is accompanied by corresponding semantic attribute information. The knowledge graph of construction procedures for vibration-damping structures is a structured construction knowledge network stored in the form of "head entity-relationship-tail entity" triples. It carries the process logic, constraints, and resource associations in the field of vibration-damping structure construction and supports subsequent rule reasoning and path retrieval.

[0019] Specifically, the process involves acquiring original information on various construction procedures for composite vibration reduction structures. First, text cleaning is performed on the original information, including removing non-textual noise, sentence segmentation, and standardizing technical terms, resulting in a structured text library of standardized construction procedures. Then, a dual-channel entity extraction strategy is built, consisting of a rule-matching channel and a semantic annotation channel. Four types of construction entities are predefined: procedure entities, material entities, constraint entities, and attribute entities. The rule-matching channel first performs keyword matching on the structured construction procedure text library according to entity type, obtaining keyword sets for the corresponding categories of construction entity types. Finally, the semantic annotation channel performs semantic annotation on each keyword set. The process involves identifying and labeling attributes to obtain a set of semantic attributes for construction entities. These attributes are then linked and integrated with their corresponding entity types to create a complete set of construction knowledge entities. Based on this, three types of entity relationships are predefined: temporal dependencies, spatial constraints, and process resource requirements. Semantic relationship analysis is performed on the construction knowledge entity set according to these relationship types to obtain a set of construction entity semantic relationships. Entities and their corresponding relationships are then organized into standardized triplet units, which are aggregated to form a set of construction entity triplet units. Finally, the triplet unit set is instantiated and populated into a predefined graph ontology model to complete knowledge modeling and output a knowledge graph of vibration-damping structure construction processes.

[0020] Preferably, the construction of the rule matching channel proceeds sequentially in four steps: corpus collection, dictionary construction, engine integration, and rule configuration. First, national construction specifications, construction method manuals, historical project construction plans, and technical disclosure documents related to composite vibration reduction structures are collected. Terminology is then sorted into four pre-defined categories: process entities, material entities, constraint entities, and attribute entities. Standard expressions, commonly used engineering aliases, on-site abbreviations, and common spelling errors for each field's terminology are extracted, completing the classification and collection of terminology materials. Next, a hierarchical keyword dictionary is built based on the categorized terminology, binding corresponding entity type tags to each category of entries, establishing a one-to-one mapping relationship between entries and entity categories. Then, the AC automaton multi-pattern matching algorithm is integrated as the core matching engine, configuring full-text sentence-by-sentence scanning, entry location positioning, and automatic category mapping logic to achieve rapid batch retrieval of candidate entities. Finally, three types of verification rules—entity length threshold, contextual part-of-speech constraints, and entity category mutual exclusion—are configured to build a post-rule verification unit for deduplication, error correction, and boundary adjustment of the preliminary matching results. For example, taking the extraction of process entities and constraint entities for composite damping structures as an example, when constructing the dictionary, standard process terms such as viscous damper installation, buckling restrained brace hoisting, and joint weld inspection are included in the process entity dictionary. At the same time, commonly used engineering aliases such as damper placement and BRB hoisting are added. Expressions such as seismic fortification intensity of 8 degrees and weld quality standard of level II are included in the constraint entity dictionary. When the matching engine scans the construction plan text, it can quickly locate all corresponding entries and automatically map their respective entity types when the seismic fortification intensity of this project is 8 degrees and joint weld inspection is carried out after the BRB hoisting is completed. Finally, after rule verification, general words without context that appear alone are removed, entity boundaries are corrected, and the keyword set of construction entity types corresponding to the category is output.

[0021] Preferably, during the construction of the semantic annotation channel, the following steps are taken: First, high-quality domain texts such as special construction plans for composite vibration reduction structures, national construction specifications, and technical briefing records are selected as the corpus base. Standardized manual annotation is carried out according to three dimensions: entity boundaries, entity categories, and semantic attributes. The start and end positions, categories, and corresponding engineering attribute characteristics of four types of entities—annotation procedures, materials, constraints, and attributes—are clearly defined to form a compliant construction domain entity annotation dataset. Second, a general pre-trained language model is selected as the base, with named entity recognition and attribute extraction as the joint task as the training objective. The parameters are fine-tuned using the domain annotation dataset to enhance the model's ability to semantically parse construction professional terms, resulting in a sequence annotation model adapted to the vibration reduction construction domain. Finally, homonymous and polysemous terminology scenarios in the construction domain are sorted out, a semantic disambiguation knowledge base and entity attribute mapping table are constructed, and a semantic disambiguation and attribute completion module is built. After connecting it with the annotation model, the overall channel is integrated and verified.

[0022] The sequence labeling model adapted for vibration reduction construction adopts a three-level cascaded architecture of BERT-BiLSTM-CRF, consisting of a character embedding layer, a bidirectional context encoding layer, and a CRF sequence constraint output layer connected in series. The character embedding layer uses a Chinese BERT-base model with a 12-layer Transformer encoder as its base, with a hidden layer dimension of 768. The input is a construction procedure text sequence with a length not exceeding 512 characters, and the output is a 768-dimensional character-level semantic vector that integrates character semantic features, positional encoding, and segmentation encoding. This vector maps professional terms in the construction text into standardized high-dimensional semantic representations. The bidirectional context encoding layer is a bidirectional long short-term memory network with a hidden unit dimension of 256. The input is a semantic vector sequence output by the character embedding layer. It captures long-distance word order dependencies and contextual semantic associations of entities in construction statements through two independent forward and backward temporal encoding paths, outputting a 512-dimensional bidirectional fused semantic feature vector to enhance the representation of long-span procedures. The model improves the boundary recognition accuracy of entities and constraints. The CRF sequence constraint output layer incorporates a learnable label transition probability matrix and uses a BIO annotation system to construct a label set, specifically including nine categories: B-process, I-process, B-material, I-material, B-constraint, I-constraint, B-attribute, I-attribute, and non-entity label O. The input is the feature vector output by the bidirectional context encoding layer. By learning the legal transition rules between labels, the model outputs the globally optimal entity label sequence, ensuring the boundary coherence and category legality of entity labels. The model's construction and training are divided into two stages: domain-based pre-training and supervised fine-tuning. In the domain-based pre-training stage, unlabeled texts from national construction specifications, industry construction method manuals, historical engineering construction plans, and technical disclosure documents in the field of vibration reduction construction are first collected. After sentence segmentation, deduplication, and terminology unification preprocessing, a domain-based pre-training corpus of no less than 100,000 sentences is constructed. A masked language model is used as the training task, with a mask ratio set to 15%. The AdamW optimizer is used, with an initial learning rate set to 2×10⁻⁶. -5 The batch size was set to 32, and the training epochs were set to 10. The full parameters of the Chinese BERT-base model were further pre-trained. Semantic and contextual features of professional terminology in the field of seismic mitigation construction were injected into the model to obtain a domain-adapted pre-trained base model. In the supervised fine-tuning stage, the manually annotated seismic mitigation construction entity dataset was first divided into training, validation, and test sets in an 8:1:1 ratio. The annotations included the start and end boundary positions and corresponding entity categories for each entity. The AdamW optimizer was used, with an initial learning rate set to 3×10⁻⁶. -5The batch size was set to 16, and the training rounds were set to 15. The weighted sum of cross-entropy loss and CRF path loss was used as the total loss function. End-to-end fine-tuning training was performed on all parameters of the model. During the training process, the entity-level F1 score of the validation set was used as the core evaluation index. An early stopping strategy was adopted. Training was terminated when the F1 score of the validation set did not improve for 3 consecutive rounds. The model parameters with the highest F1 score on the validation set were selected as the optimal parameters. After the accuracy was verified on the test set, a sequence labeling model adapted to the field of vibration reduction construction was obtained.

[0023] Preferably, during sequential execution, the rule matching channel performs a full and rapid scan of the construction process text library based on the AC automaton and a hierarchical keyword dictionary. This enables the precise location and extraction of candidate fragments related to construction entities from massive amounts of text, thereby compressing the processing object of the subsequent semantic annotation channel from the entire text to candidate entity context fragments. This effectively avoids the bottlenecks of high computational resource consumption and high inference latency faced by deep learning models when performing word-by-word sequence annotation on long construction documents, and improves the efficiency of knowledge graph construction for large-scale engineering documents. At the same time, after identifying entity boundaries and categories, the high-confidence annotation results output by the semantic annotation channel can be continuously fed back to the rule matching channel to iteratively optimize the coverage of the keyword dictionary and the accuracy of the matching rules, forming a positive feedback loop of rule coarse screening to improve efficiency, semantic refinement to maintain accuracy, and results feeding back into better rules.

[0024] This step transforms the dispersed, unstructured knowledge of vibration reduction structure construction procedures into a structured, reasonable knowledge network, providing a stable domain knowledge foundation for the intelligent generation and optimization of subsequent construction procedure schemes.

[0025] Furthermore, entity extraction is performed on the structured construction process text library according to the dual-channel entity extraction strategy to obtain a set of construction knowledge entities, including: defining construction entity types, which include process entities, material entities, constraint entities, and attribute entities; using the rule matching channel to perform keyword matching on the structured construction process text library according to the construction entity types to obtain a set of construction entity type keywords; performing semantic recognition annotation on the set of construction entity type keywords based on the semantic annotation channel to obtain a set of construction entity semantic attributes; and associating and integrating the set of construction entity semantic attributes with the construction entity types to obtain a set of construction knowledge entities.

[0026] Specifically, the construction entity type serves as the classification criterion for entity extraction, categorizing core knowledge objects in the field of composite vibration reduction structure construction. This application includes four categories: process, material, constraint, and attribute, used to clarify the definition boundaries of each type of entity and provide a unified classification basis for dual-channel extraction. Process entities refer to action-type knowledge units with independent operational logic and clearly defined implementation nodes in the construction process, covering various construction operations such as vibration reduction component hoisting, node welding, and quality inspection. Material entities refer to physical knowledge units involved in the construction process, including vibration reduction components such as buckling-restrained braces and viscous dampers, as well as various construction materials and equipment such as welding materials and hoisting equipment. Constraint entities refer to constraint-type knowledge units that restrict the implementation of construction processes, including specifications and working conditions such as seismic fortification intensity, weld quality grade, construction environment requirements, and construction period limitations. Attribute entities are attribute-type knowledge units used to describe the characteristics of various construction entities, including quantitative or qualitative characteristic parameters such as yield bearing capacity, installation deviation threshold, and component specifications and dimensions.

[0027] Specifically, the rules for classifying construction entity types and category boundaries are predefined, clarifying the definition scope and judgment criteria of the four major categories: process entities, material entities, constraint entity entities, and attribute entities. The rule matching channel is invoked, and according to the hierarchical keyword dictionary corresponding to each of the four construction entity types, sentence-by-sentence keyword matching and location are performed on all text in the structured construction process text library. The matching results are then categorized and collected according to entity type, resulting in a keyword set for each category of construction entity type. Next, the keyword sets for each type of construction entity, along with the complete context statements containing the keywords, are input into the semantic annotation channel. A sequence annotation model adapted to the field of vibration reduction construction is used to perform semantic recognition and attribute annotation on candidate keywords, correcting category misjudgments and entity boundary deviations caused by rule matching, supplementing the semantic attribute information of entities, and outputting a semantic attribute set of construction entities by category. All entities in the semantic attribute set of construction entities are then mapped and integrated with the predefined construction entity types, simultaneously completing entity deduplication, boundary calibration, and type label normalization processing, ultimately obtaining a complete and standardized set of construction knowledge entities.

[0028] Preferably, keyword matching is implemented based on a hierarchical keyword dictionary for construction entities and an AC automaton multi-pattern matching algorithm, specifically through five steps: text preprocessing, automaton construction, full-sentence scanning and matching, conflict filtering, and classification and aggregation. First, all text in the structured construction process text library is formatted with a unified character encoding, split into independent sentence units according to punctuation, and blank lines and meaningless characters are removed to form a sequence of text to be matched, arranged sentence by sentence. Second, the hierarchical keyword dictionary corresponding to four types of entities—process, material, constraint, and attribute—is loaded. An AC automaton state transition tree is constructed using all keywords as pattern strings. At each termination state node, the standard terminology name and entity type label of the corresponding keyword are bound, and a one-to-one mapping relationship is established between project aliases, site abbreviations, and standard terms. Then, the text is read sentence by sentence. The text sequence to be matched is taken and traversed along the state transition path of the AC automaton, character by character. When the traversal reaches the termination state, it is determined that the match is successful. The matched keyword text, the start and end character positions in the sentence, the corresponding standard terminology, and the entity type are recorded simultaneously. After the single sentence matching is completed, conflict filtering is performed. For scenarios where there are multiple overlapping matching results at the same character position, the unique optimal matching result is retained according to the rule that long keywords take precedence over short keywords and entity type priority constraints > process > material > attribute. At the same time, invalid matching items with a character length of less than 2 and redundant matching items that are repeated throughout the text are removed. Finally, the verified matching results are collected and stored according to four categories: process entity, material entity, constraint entity, and attribute entity, which generate four keyword sets for construction entity types.

[0029] This step employs a dual-channel sequential extraction mode combining rule-matching for initial screening and semantic annotation for refinement. This mode balances the recall rate and recognition accuracy of construction professional entity extraction, achieving standardized extraction of multi-dimensional construction knowledge entities and providing reliable, high-quality basic data support for the subsequent construction of a knowledge graph of vibration reduction structure construction procedures.

[0030] Furthermore, based on the set of construction knowledge entities, semantic relationship analysis and knowledge modeling are performed to construct a knowledge graph of construction procedures for vibration-damping structures. This includes: defining entity relationship types, which include temporal dependencies, spatial constraints, and process resource requirements; performing semantic relationship analysis on the set of construction knowledge entities according to the entity relationship types to obtain a set of semantic relationships of construction entities; associating and storing the set of construction knowledge entities according to the set of semantic relationships of construction entities to obtain a set of triplet units of construction entities; and instantiating the set of triplet units of construction entities and filling it into the graph ontology model for knowledge modeling to construct a knowledge graph of construction procedures for vibration-damping structures.

[0031] Specifically, entity relationship types are classification standards for defining the logical association between two entities in a knowledge graph. They are the core basis for constructing entity relationship networks. This application includes three categories: temporal dependency, spatial constraint, and process resource requirement, covering the three core association dimensions of construction process logic, spatial boundaries, and resource input. Temporal dependency relationships describe the relationship type of execution order and logical dependence between different process entities, reflecting the sequential rules of the construction process, such as the correspondence that a subsequent process can only be started after the previous process is completed. Spatial constraint relationships describe the relationship type of restriction and being restricted between process / component entities and site conditions and spatial working conditions, reflecting the spatial boundary conditions of construction implementation, such as the restriction relationship of working height and installation clearance on construction processes. Process resource requirement relationships describe the relationship type of demand and being demanded between process entities and material / equipment entities, reflecting the resource input correspondence of construction processes, such as the matching relationship of vibration damping components and construction equipment required for a certain process. The construction entity triplet unit set refers to a standardized knowledge data set with head entity-relationship-tail entity as the basic unit. It is the smallest storage unit of the knowledge graph and is used to structurally carry the association information between entities.

[0032] Specifically, a system of entity relationship types in the construction field is predefined, clarifying the core connotations, applicable entity pair scopes, and judgment rules of three types of relationships: temporal dependency, spatial constraint, and process resource demand. This provides a unified classification basis for semantic relationship analysis. Using a set of construction knowledge entities as the analysis object, and combining the contextual semantics of the original text containing the entity with the inherent logical rules of the construction field, semantic relationship matching analysis of entity pairs is conducted one by one according to the three relationship types. Temporal dependency is extracted by identifying the sequential and pre- and post-process logic between processes; spatial constraint is extracted by identifying the restrictive correspondence between processes / components and site conditions / environmental conditions; and spatial constraint is extracted by identifying the correspondence between processes and the input of materials and equipment. The process involves extracting the resource requirements of each construction process and summarizing all valid entity associations to obtain a set of semantic relationships for construction entities. Based on this, the two entities corresponding to each set of semantic relationships are designated as the head and tail entities, respectively. The corresponding relationship types are used as association edges, and the relationships are stored and formatted according to a standardized structure of head entity-relationship-tail entity. After deduplication, conflict verification, and normalization, a set of triplet units for construction entities is obtained. Finally, a predefined ontology model of the construction process of vibration-damping structures is loaded, and the entity instances and relationship instances in the triplet unit set are mapped and filled into the corresponding class slots and relationship slots of the ontology model. This completes knowledge modeling and consistency verification, ultimately constructing a knowledge graph of the construction process of vibration-damping structures.

[0033] Preferably, when defining entity relationship types, the business logic of the construction process of composite vibration reduction structures should be the core, following four basic principles: business relevance, boundary exclusivity, coverage integrity, and computability. Specifically, this is executed in four steps: dimension extraction, boundary limitation, rule formulation, and verification and solidification. The system systematically sorts out the core relational dimensions between entities in the entire construction process of vibration reduction structures. Starting from three core business scenarios—the sequential logic of process execution, the spatial constraints of construction implementation, and the resource input of process implementation—three core relationship types are extracted: temporal dependency, spatial constraint, and process resource demand relationship. This ensures that the relationship types fully meet the actual business needs of construction process optimization. Secondly, the entity type adaptation boundaries of each relationship type are clarified. In the temporal dependency relationship, the head and tail entities are limited to process entities, and the head entity of the spatial constraint relationship is either a process entity or a material entity. The tail entity is limited to constraint entities, and the head entity of the process resource demand relationship is limited to process entities, while the tail entity is a material entity or attribute entity. This ensures that the applicable scope of the three types of relationships does not overlap and that the boundaries are clear and mutually exclusive. Subsequently, semantic judgment rules for each type of relationship are formulated. The temporal dependency relationship is judged based on the sequential succession logic that the subsequent process can only be started after the preceding process is completed. The spatial constraint relationship is judged based on the constraint logic that the constraint conditions restrict the implementation or installation of entities. The process resource demand relationship is judged based on the logic that the execution of the process requires the consumption or use of the corresponding entity's resources. This provides a feasible and quantitative judgment standard for subsequent relationship extraction. Finally, a small batch of trial extraction is carried out using typical construction process data to verify the scenario coverage and judgment consistency of the relationship types, correct ambiguous boundary rules, and finally solidify the unified definition and judgment specification of the three types of entity relationship types.

[0034] Preferably, when performing semantic relationship analysis, all entities within the construction knowledge entity set are used as the basis. Candidate entity pairs are generated based on the co-occurrence context of the same sentence or paragraph, limiting relationship determination to entities within the same semantic context and reducing the scope of invalid calculations. Secondly, corresponding determination logic is executed for the three types of entity relationships. For temporal dependency relationships, pairs of candidates consisting entirely of process entities are extracted, a pre-defined dictionary of construction temporal trigger words is loaded, and dependency parsing is used to identify temporal connective components in the statement and syntactic associations between entities. At the same time, the inherent process sequence logic rules in the construction specifications for composite vibration reduction structures are matched to comprehensively determine whether there is a sequential dependency relationship and the direction of dependency between two process entities. For spatial constraint relationships, candidate pairs consisting of process entities or material entities and constraint condition entities are extracted, and constraint conditions are identified through semantic role labeling. The system calculates the degree of association between the limiting modifiers of entity in the statement and the corresponding entity by combining semantic similarity calculation, and determines whether there is a constraint relationship between the two. For process resource demand relationships, candidate pairs composed of process entities and material entities or attribute entities are extracted. Based on a pre-set resource predicate dictionary, and with the help of syntactic dependency analysis, the system identifies the objects and supporting resource components corresponding to the process actions, and determines whether there is a demand relationship for resource input between the process and the corresponding entity. Then, a relationship conflict check is performed. For scenarios where the same candidate entity pair matches multiple types of relationships, a unique attribution determination is made based on the core semantic priority and construction business logic, and weakly related entity pairs without clear semantic support are eliminated. Finally, all valid entity association pairs are standardized and collected according to the format of head entity, relationship type, and tail entity, and output to form a set of construction entity semantic relationships.

[0035] This step transforms discrete construction knowledge entities into a domain knowledge network with structured relationships. The system accumulates the process logic, constraints, and resource correspondence of vibration reduction structure construction, providing a computable and reasonable knowledge foundation for rule-based reasoning and intelligent optimization of subsequent process schemes.

[0036] Step S200: Input the current engineering structure parameters, and perform rule-based reasoning and filtering on the knowledge graph of the vibration reduction structure construction process based on the current engineering structure parameters to obtain an initial construction process scheme set.

[0037] Furthermore, based on the current engineering structural parameters, rule-based reasoning and filtering are performed on the knowledge graph of the vibration-damping structure construction process to obtain an initial construction process scheme set. This includes: inputting the current engineering structural parameters into the knowledge graph of the vibration-damping structure construction process for mapping and transformation to obtain a set of initial engineering anchor nodes; pre-setting process node reasoning rules and process constraint filtering rules according to the current engineering structural parameters, wherein the process node reasoning rules include triggering conditions and reasoning actions; triggering from the set of initial engineering anchor nodes, performing path reasoning and filtering on the knowledge graph of the vibration-damping structure construction process according to the process node reasoning rules and process constraint filtering rules to obtain the initial construction process scheme set.

[0038] Specifically, the current engineering structural parameters refer to the core design and operating parameters of the composite vibration reduction structure project to be implemented, covering seismic fortification level, vibration reduction component type and specifications, building structure form, construction site conditions, etc., and serve as the input basis for the customized generation of process schemes. The initial engineering anchor node set: After the current engineering structural parameters are mapped and matched with the entity nodes in the vibration reduction structure construction process knowledge graph, the corresponding set of entity nodes is located, serving as the starting anchor point for process path reasoning. Process node reasoning rules are used to expand the reasoning rules of the process path on the knowledge graph. They consist of two parts: triggering conditions and reasoning actions. When the triggering conditions are met, the corresponding node expansion action is executed, driving the forward derivation of the process path. Triggering conditions refer to the preconditions that trigger the process node reasoning action, typically requiring the current node's entity type, attribute value, or association relationship to meet preset requirements; these are prerequisites for initiating the reasoning action. Reasoning actions refer to the node expansion operations performed on the knowledge graph after the triggering conditions are met, typically involving jumping along a specified relationship type to an associated node, supplementing the corresponding process node, and extending the process path. Process constraint filtering rules are used to verify the legality of process nodes and paths. They cover constraints such as quality standards, schedule requirements, and spatial conditions, and are used to eliminate process paths that do not meet engineering requirements. Path reasoning refers to the traversal process of starting from the initial anchor node, following the entity relationship links of the knowledge graph, and gradually expanding the process nodes according to reasoning rules to generate a complete process path.

[0039] Specifically, the seismic fortification level, damping component specifications, structural form, site conditions, and other engineering structural parameters of the current project are input into the system. An entity semantic matching algorithm maps each parameter to constraint entities, material entities, and attribute entities in the knowledge graph of the damping structure construction process, locating the set of anchor nodes corresponding to the current project. Then, based on the specific requirements of the current project's structural parameters, pre-set process node inference rules and process constraint filtering rules. The process node inference rules specify the triggering conditions and inference actions for each rule. The triggering conditions are set to ensure that the attribute values ​​and relationships of the current node meet the engineering parameter requirements. The inference actions are set to jump along the temporal dependency relationship and the process resource requirement relationship to the corresponding node. The process should be associated with process nodes. Process constraint filtering rules are used to clarify the constraint judgment criteria such as weld quality level, construction environment threshold, and component installation deviation. Then, starting from the initial project anchor node set, a breadth-first traversal strategy is used to carry out path reasoning on the knowledge graph. For each process node expanded, it is checked whether it meets the trigger conditions of the process node reasoning rules. If the conditions are met, the corresponding reasoning action is executed to extend the process path. At the same time, the process constraint filtering rules are called to check the compliance of the current node and path, and nodes and invalid paths that do not meet the project constraints are eliminated until all extendable compliant process paths are traversed. Finally, all complete and compliant process paths are sorted and deduplicated, and an initial construction process scheme set adapted to the current project is output.

[0040] Preferably, during the mapping and transformation, the input current engineering structural parameters are structurally parsed and broken down into several parameter entries according to four categories: seismic design parameters, damping component parameters, structural form parameters, and construction site parameters. Each entry includes a parameter category label, a standard parameter value, and an additional constraint description, completing the conversion of unstructured parameters into standardized structured parameters. Secondly, for each type of parameter entry, the corresponding entity type in the knowledge graph of damping structure construction procedures is matched. Seismic design parameters and site condition parameters correspond to constraint entities, damping component parameters correspond to material entities, and structural dimensions and performance index parameters correspond to attribute entities. First, a precise string matching method is used to match the parameter entries one-to-one with the standard terms of the graph entities, directly matching the parameters with the standard terms. The entries are directly mapped to nodes. For parameter entries that do not match precisely, a domain-adapted pre-trained language model is used to semantically encode the text of the parameter entry and all entity texts of the same type in the graph, generating semantic vectors with consistent dimensions. The semantic matching degree between the parameter and each entity is obtained by calculating the cosine similarity between the vectors. Then, a similarity threshold of 0.85 is preset, and entities with semantic matching degree greater than or equal to the threshold are selected as candidate matching nodes. The entity with the highest matching degree is selected as the corresponding mapping node of the parameter entry, while invalid candidates with matching degree lower than the threshold are removed. Finally, the entity nodes obtained by mapping all parameter entries are deduplicated and consistency is checked to exclude nodes with attribute conflicts. After summarizing all valid mapping nodes, the initial project anchor node set is obtained.

[0041] Preferably, the process node reasoning rules are production rules used to forward expand process paths. Each rule adopts a fixed binary structure of trigger conditions and reasoning actions, pre-fixed into the graph reasoning engine based on the construction specifications and construction method logic of composite vibration reduction structures. The trigger conditions use the entity type, attribute values, and relationships of the current node, as well as the current engineering structure parameters, as the judgment elements. The reasoning actions correspond to node jumps, path extensions, and branch selection operations on the knowledge graph, and are mainly divided into three categories: Temporal progression reasoning rules: These deduce the sequential links of processes based on the temporal dependencies between entities. The trigger condition is that the attributes of the current process node match the engineering parameters, and the reasoning action is to jump to the subsequent process node along the temporal dependency edge. For example, if the current node is a buckling-restrained support hoisting and positioning process entity, and the corresponding node attribute is a welded connection node, then jump to the support node welding process node along the temporal dependency. Resource matching reasoning rules: These complete the matching preceding processes based on the process resource demand relationship. The trigger condition is that the current process matches the corresponding material or equipment entity, and the reasoning action is to add the corresponding resource preparation process node to the current path. For example: If the current process involves installing and matching a viscous damper to a material displacement monitoring sensor, then the preceding process node for sensor accuracy calibration will be added to the path. Parameter branch inference rules: Differentiated process branches are triggered based on engineering parameters. The triggering condition is that the engineering parameters meet a preset threshold, and the inference action is to select the corresponding process node. For example: If the seismic fortification intensity of the current project is ≥8 degrees, and the current node is weld quality inspection, then the process will jump to the first-level weld ultrasonic non-destructive testing process branch, replacing the conventional second-level visual inspection process.

[0042] Process constraint filtering rules are negative rules used to reverse-verify the legality of work paths. They are pre-set based on construction acceptance standards and engineering condition limitations to eliminate process nodes and invalid paths that do not meet the current project requirements. The core rules are divided into four categories: Quality Standard Constraint Rules: Verify whether the process matches the corresponding quality acceptance level of the project. For example, if the weld inspection level in the process path is Level II, and the seismic fortification intensity of the current project is 8 degrees or above, then the path is determined to be inconsistent with the seismic quality code and is eliminated. Spatial Condition Constraint Rules: Verify whether the process is suitable for the spatial limitations of the construction site. For example, if the process path includes a 25t truck crane hoisting process, and the current project site's clear height is <7m, then the path is determined to be unsuitable for implementation and is eliminated. Environmental Condition Constraint Rules: Verify whether the process is suitable for the construction environment parameter requirements. For example, if the process path includes a viscous damper hydraulic system oil injection process, and the construction environment temperature is below 0℃, then the path is determined to be unsuitable for process implementation and is eliminated. Schedule Threshold Constraint Rules: Verify whether the total schedule of the path meets the project schedule limit. For example, if the total construction period of the vibration reduction sub-project corresponding to the process path exceeds the construction period threshold set for the current project, then the path is determined not to meet the construction period requirements and is removed.

[0043] This step, based on a knowledge graph, enables targeted rule-based reasoning and constraint filtering of construction process schemes. It can quickly generate compliant process schemes that adapt to the current project parameters, effectively ensuring the correctness of the process logic and the compliance with project constraints, and providing a high-quality initial scheme space for subsequent global optimization.

[0044] Step S300: Match candidate construction procedure schemes from the composite vibration reduction structure engineering case library according to the current engineering structural parameters, and construct a feasible construction procedure scheme space based on the initial construction procedure scheme set and the candidate construction procedure scheme set.

[0045] Furthermore, matching a set of candidate construction procedures from the composite vibration reduction structure engineering case library according to the current engineering structural parameters includes: retrieval encoding of the current engineering structural parameters to generate a multi-dimensional retrieval vector; performing retrieval matching within the composite vibration reduction structure engineering case library based on the multi-dimensional retrieval vector to obtain a graph structure similarity set; and selecting and extracting cases within the composite vibration reduction structure engineering case library according to the graph structure similarity set to determine the set of candidate construction procedures.

[0046] Specifically, the composite vibration reduction structure engineering case library refers to a structured dataset of historically implemented composite vibration reduction structure engineering projects. Each case contains core information such as structural parameters, complete construction procedures, and implementation effect data. All case procedures are stored in a graph structure, serving as the data source for case retrieval and matching. The multidimensional retrieval vector refers to a high-dimensional feature vector output through retrieval encoding. Each dimension corresponds to an engineering feature dimension such as seismic fortification level, vibration reduction component type, building structure height, and site conditions, quantifying the core attribute information of the current project and used for similarity matching calculations. Graph structure similarity is a comprehensive index used to quantify the matching degree of engineering features and the similarity of the process topology between the current project and historical cases. It integrates the calculation results of node attribute similarity and edge structure similarity to accurately reflect the reusability of the process solutions. The feasible construction procedure solution space refers to the total set of all valid construction procedure solutions formed after merging, deduplicating, and verifying compliance of the initial construction procedure solution set generated by knowledge graph rule reasoning and the candidate construction procedure solution set obtained from case matching. This represents the search scope for subsequent global optimization.

[0047] Specifically, structural parameters such as the seismic fortification level, damping component type and specifications, building structure form, and construction site conditions of the current project are obtained. These parameters are then normalized and numerically mapped using feature dimensions. A pre-defined feature coding model is used to perform retrieval coding, assigning a corresponding vector dimension to each project feature to generate a multi-dimensional retrieval vector with unified dimensions. Subsequently, a composite damping structure project case library is loaded. The multi-dimensional retrieval vector is compared with the project feature vector of each case in the library to calculate attribute similarity. Simultaneously, the structural similarity between the node composition and relational links of each case's process diagram and the expected process topology of the current project is weighted to obtain a comprehensive diagram for each case. The similarity scores of all cases are aggregated to form a graph structure similarity set. Then, historical cases are sorted from high to low similarity, and a similarity threshold and a selection threshold are preset to select the top few historical cases with similarity scores higher than the threshold. The complete construction procedure schemes of the corresponding cases are extracted to obtain a candidate construction procedure scheme set. Finally, the initial construction procedure scheme set obtained by reasoning through knowledge graph rules is merged with the candidate construction procedure scheme set obtained by case matching. Duplicate procedure schemes are deduplicated, and the basic constraint compliance of all schemes is checked simultaneously. Invalid schemes with logical conflicts and constraint inconsistencies are eliminated to construct a complete feasible construction procedure scheme space.

[0048] Preferably, the feasible construction procedure scheme space adopts a three-layer structured storage architecture: a scheme index layer, a procedure sequence layer, and a constraint attribute layer. The overall structure is based on a directed acyclic graph (DAG) as the basic storage unit. The scheme index layer stores the unique identifier, source category, and initial fit score of each procedure scheme for rapid retrieval and classification management. The procedure sequence layer stores the core construction procedure nodes, temporal dependency links, supporting resources, and constraint nodes of each scheme in an ordered topological node sequence, fully carrying the execution logic and relationships of the procedures. The constraint attribute layer carries multi-dimensional quantitative attribute parameters for each scheme, such as estimated construction period, construction cost, quality acceptance level, and workspace requirements, providing an evaluation basis for subsequent multi-objective optimization. During construction, the initial construction procedure scheme set output by knowledge graph rule reasoning and the candidate construction procedure scheme set obtained by matching the case library are first obtained. The two scheme sets are then executed... The process involves standardizing the format of all schemes, including the terminology of process nodes, the coding rules for relationship types, and the dimensions of attribute parameters. All schemes are then converted into a uniformly formatted directed acyclic process topology diagram. Next, a scheme deduplication operation is performed. The overlap between schemes is determined by calculating the graph edit distance between process diagrams of different schemes. Schemes with edit distances less than a preset threshold are retained for those with higher adaptability, while redundant and duplicate schemes are eliminated. Then, based on the three mandatory constraints corresponding to the current engineering structural parameters—quality standards, project deadlines, and spatial conditions—compliance checks are performed on all remaining schemes, eliminating invalid schemes that do not meet the hard constraints. Finally, all verified process schemes are aggregated and stored according to a three-layer architecture. A unique index identifier and initial adaptability mark are generated for each scheme, ultimately constructing a complete and feasible construction process scheme space covering logically compliant schemes and practically verified schemes.

[0049] Preferably, the screening threshold is composed of a graph structure similarity admission threshold and a candidate scheme number upper limit threshold, determined by a combination of historical data statistical calibration and dynamic adaptation to engineering level. At least 50 sets of labeled test cases verified by actual engineering are extracted from the composite vibration reduction structure engineering case library. Using the engineering structural parameters of each test case as the search input, the graph structure similarity is calculated by traversing the entire library. The engineering reusability rate of the screening results under different similarity values ​​is statistically analyzed; that is, the proportion of selected case schemes that can be applied to the target project without major adjustments, only parameter adaptation is required. With an engineering reusability rate of not less than 85% as the core criterion, the lowest similarity value meeting this condition is selected as the basic similarity admission threshold, with a calibration range of 0.70~0.80. Based on this, the following is concluded... The similarity threshold is dynamically adjusted according to the structural importance level of the current project. When the project is a key project with a seismic fortification intensity of 8 degrees or above and a building height of over 100 meters, the similarity threshold is increased by 0.05 to 0.10 to improve the matching accuracy of candidate schemes. When it is a conventional small and medium-sized seismic reduction structure project, the threshold is decreased by 0.05 to appropriately expand the range of candidate schemes. The upper limit of the number of candidate schemes is set to 5 to 10 sets based on the computational efficiency and scheme diversity requirements of subsequent global optimization. At the same time, a fallback correction mechanism is set. When the number of cases that meet the similarity threshold is less than 3 sets, the similarity threshold is gradually decreased in increments of 0.05 until the number of candidate cases meets the minimum requirement of 3 sets, so as to avoid the empty candidate construction procedure scheme set from affecting the construction of the feasible scheme space.

[0050] This step supplements the engineering practice-verified process solutions through case matching, complementing the solutions derived from knowledge graph rule reasoning. It balances the logical compliance and engineering practicality of the process solutions, effectively expanding the coverage of feasible solutions and constructing a high-quality solution space for subsequent multi-objective global optimization.

[0051] Step S400: Perform a global search and optimization of the feasible construction procedure scheme space to determine the target construction procedure optimization scheme, and use the target construction procedure optimization scheme to control the construction of the composite vibration reduction structure.

[0052] Furthermore, a global search and optimization is performed on the feasible construction procedure scheme space to determine the target construction procedure optimization scheme, including: constructing a multi-objective function for construction effect evaluation based on the construction requirements of the composite vibration reduction structure; using the multi-objective function for construction effect evaluation to evaluate the fitness of the feasible construction procedure scheme space and obtain the fitness of multiple construction procedure schemes; and performing a global search and optimization on the feasible construction procedure scheme space based on the fitness of the multiple construction procedure schemes to determine the target construction procedure optimization scheme.

[0053] Specifically, the multi-objective function for construction effect evaluation is a mathematical evaluation system used to quantitatively judge the comprehensive performance of construction procedure schemes. It is composed of a weighted integration of multiple mutually constraining optimization sub-objective functions such as construction period, cost, construction quality, and safety risks. It is the core criterion for measuring the merits of schemes and driving iterative optimization. The fitness of construction procedure schemes refers to the comprehensive evaluation value calculated after substituting each construction procedure scheme into the multi-objective function. It quantitatively reflects the degree to which the scheme meets the optimization objectives. The higher the fitness, the better the comprehensive construction effect of the scheme. It is the core basis for scheme selection and iteration in the optimization process.

[0054] Specifically, based on the construction management requirements of composite vibration reduction structure engineering, four core optimization objectives are determined: total construction period, cost of sub-projects, construction quality compliance rate, and on-site safety risk level. Quantitative calculation models for corresponding sub-objectives are constructed for each. The weight coefficients of each sub-objective are assigned using the analytic hierarchy process (AHP) in conjunction with the project level and owner requirements. After normalization, these are integrated to form a multi-objective function for construction effect evaluation. Subsequently, each set of construction procedure schemes within the feasible construction procedure scheme space is substituted into this multi-objective function. Based on the scheme's procedure topology, resource allocation parameters, and constraint attribute information, the quantitative values ​​of each sub-objective are calculated sequentially. After dimensionless processing, a weighted sum is obtained to obtain the fitness of the construction procedure scheme corresponding to each scheme, forming a fitness evaluation result set for the entire scheme space. Then, all feasible construction procedure schemes within the feasible construction procedure scheme space are evaluated. The proposed scheme serves as the initial population. An improved genetic algorithm with an elite retention strategy is used for global search and optimization. A roulette wheel selection operation is performed based on fitness to select high-quality schemes as parents. A new generation of process schemes is generated through process node crossover and local process sequence mutation operations. Simultaneously, mandatory constraint compliance verification is performed on the new schemes to eliminate invalid schemes that do not meet the engineering requirements. The population and corresponding fitness values ​​are updated iteratively generation by generation. The iteration is terminated when the number of iterations reaches the preset maximum number of iterations or when the optimal fitness of the population does not improve significantly for five consecutive generations. Finally, the process scheme with the highest fitness is selected from the terminated generation as the target construction process optimization scheme. Subsequently, based on the process sequence arrangement, resource scheduling plan and quality control requirements of this scheme, precise control of the progress, quality and safety of the entire process of composite vibration reduction structure construction is carried out.

[0055] Preferably, the multi-objective function for evaluating construction effectiveness is a comprehensive fitness function. This function is used to quantify the overall construction effect of the x-th construction procedure scheme. The function ranges from [0,1]. A larger value indicates a better overall performance of the scheme. The specific form of the function is as follows: ; in, , , , These are the weighting coefficients for the schedule target, cost target, quality target, and safety target, respectively, satisfying... + + + =1, the weighting coefficient is determined by the analytic hierarchy process in combination with the seismic resistance level, structural importance level and owner control requirements of the current project; For the normalization sub-function of the project duration target, For the cost target normalization subfunction, For the quality target normalization subfunction, For the safety target normalization sub-function, all four sub-functions take values ​​in the range [0,1], and the dimensional differences are eliminated through extremum normalization. Wherein: The normalization sub-function for the project duration target is: In the formula, The total duration of the critical path for the xth construction procedure scheme is calculated from the longest operation path duration of the procedure sequence; , These represent the maximum and minimum total construction periods for all possible construction procedures within the feasible construction procedure space. The function value increases as the total construction period of each procedure decreases, reflecting the optimization objective of time compression.

[0056] The cost target normalization subfunction is: In the formula, The total cost of the vibration reduction sub-project of the xth construction procedure scheme includes the sum of component installation labor costs, construction machinery shift costs, auxiliary consumable costs, and special testing costs; , These represent the maximum and minimum costs of all feasible construction procedures within the feasible construction procedure space. The function value increases as the total cost of the procedures decreases, reflecting the optimization objective of cost control.

[0057] The quality target normalization sub-function is: In the formula, The construction quality compliance rate of the xth construction procedure plan is calculated by weighting the pass rate of all quality acceptance nodes in the plan, covering core acceptance items such as component installation deviation, weld quality grade, and vibration damping performance verification. These represent the highest and lowest quality compliance rates among all feasible construction procedures within the available space. The function value increases as the quality compliance rate of the chosen procedures improves, reflecting the optimization objective of quality assurance.

[0058] The security target normalization sub-function is: In the formula, The safety risk quantification value of the xth construction procedure plan is obtained by weighted summation of the risk levels of high-risk procedures within the plan, covering risky procedures such as hoisting of heavy vibration damping components, high-altitude operations, and hot work. , These represent the maximum and minimum safety risk values ​​for all possible construction procedures within the feasible construction procedure space. This function value increases as the safety risk of a procedure decreases, reflecting the optimization objective of safety management.

[0059] in, , , , An improved analytic hierarchy process (AHP) combined with a dynamic assessment of project importance was employed. First, a three-layer AHP structure was constructed, with the optimal overall construction benefit as the target layer and the time, cost, quality, and safety targets as criterion layers. The relative importance of the four optimization targets was compared pairwise using a 1-9 scale, and the geometric mean of the expert scores was used to construct a fourth-order judgment matrix. Second, the judgment matrix was solved using a sum-product method to obtain the initial weight vectors corresponding to each criterion. Simultaneously, the maximum eigenvalue and consistency ratio (CR) of the judgment matrix were calculated. When CR < 0.1, the judgment matrix was deemed to meet the consistency test requirements, and the initial weights were considered valid. If consistency was not met... If the weighting ratio does not meet the requirements, the judgment matrix is ​​fed back to experts for scoring and adjustment until it passes the consistency check. Then, an engineering importance correction coefficient is introduced. Based on the current project's seismic fortification intensity, building height, and safety level, the correction coefficient for each objective is calculated. For super high-rise or large-span key projects with a seismic fortification intensity of 8 degrees or above, the correction coefficient for the quality objective is set to 1.2, the correction coefficient for the safety objective is set to 1.15, and the correction coefficient for the cost objective is set to 0.8. For conventional small and medium-sized seismic reduction projects, a balanced correction coefficient is used. Finally, the initial weights are multiplied by the corresponding correction coefficients and then normalized so that the sum of the four weight coefficients is 1, ultimately yielding a weighting coefficient suitable for the current project. , , , Weighting coefficients.

[0060] This step, through multi-objective comprehensive evaluation and intelligent global optimization, achieves collaborative optimization of multi-dimensional construction objectives within the feasible solution space. It can output the optimal process solution that takes into account the construction period, cost, quality and safety, effectively improving the comprehensive benefits and the level of refined management of composite vibration reduction structure construction.

[0061] Furthermore, based on the fitness of the multiple construction process schemes, a global search and optimization is performed on the space of feasible construction process schemes to determine the target construction process optimization scheme, including: optimizing the space of feasible construction process schemes according to the fitness of the multiple construction process schemes to obtain a frontier construction process scheme set; and performing population growth expansion and global search and optimization based on the frontier construction process scheme set to determine the target construction process optimization scheme.

[0062] Specifically, the frontier construction process solution set refers to the set of non-dominated high-quality process solutions selected from the feasible solution space based on the Pareto optimality criterion. No solution within this set can improve other objectives without degrading at least one optimization objective. It represents the set of high-quality solutions under multi-objective optimization and serves as the parent generation for subsequent population expansion. Population growth and expansion refers to the process of generating a large number of new feasible solutions using the frontier high-quality solutions as parents through operations such as process node cross-recombination, local process sequence mutation, and parameter adaptation adjustment. This is used to expand the search range in the neighborhood of high-quality solutions and discover potential better solutions. Pareto dominance refers to the rule for determining the superiority of solutions in multi-objective optimization. If solution A performs no worse than solution B in all optimization objectives (time, cost, quality, and safety), and performs better in at least one objective, then solution A is considered to dominate solution B, and the dominated solution lacks frontier competitiveness. The elite retention strategy is a protection mechanism in the global optimization process. In each iteration, several solutions with the highest overall fitness in the current population are directly retained to the next generation population to prevent the loss of high-quality process solutions during cross-mutation and ensure optimization convergence.

[0063] Specifically, based on the multi-objective values ​​and comprehensive fitness of all schemes within the feasible construction procedure scheme space, the Pareto dominance criterion is used for scheme optimization. The performance of different schemes in four dimensions—schedule, cost, quality, and safety—is compared one by one. Inferior schemes completely dominated by other schemes are eliminated, and all non-dominated high-quality schemes are retained, forming a frontier construction procedure scheme set as the initial high-quality parent population for global optimization. Subsequently, based on this frontier scheme set, population growth and expansion are carried out. Using each frontier scheme as a basis, a large number of new procedure schemes are generated through three methods: cross-recombination of procedure sequences, local procedure node mutation, and fine-tuning of resource allocation parameters. Simultaneously, all... The newly generated schemes undergo mandatory constraint compliance verification and fitness calculation, eliminating invalid schemes that violate engineering hard constraints, resulting in an expanded and complete optimization population. Then, based on the expanded population, iterative global search optimization is carried out. In each iteration, scheme selection, cross-recombination, and mutation update operations are performed sequentially. Simultaneously, an elite retention strategy is adopted to directly retain the best scheme of each generation to the next generation, continuously updating the Pareto front and optimal fitness of the population. When the number of iterations reaches the preset limit or the optimal fitness of the population does not significantly improve for several consecutive generations, the iteration is terminated. Finally, the scheme with the highest comprehensive fitness is selected from the frontier schemes of the terminated generation and determined as the target construction process optimization scheme.

[0064] Preferably, the Pareto optimality criterion in this application is a rule for judging the merits of solutions in the multi-objective optimization scenario of composite vibration reduction structure construction procedures. It uses four core optimization objectives—time reduction, cost reduction, quality compliance rate improvement, and safety risk reduction—as a unified evaluation dimension. For any two sets of construction procedure solutions in the feasible construction procedure solution space, if solution A performs no worse than solution B in all four optimization dimensions, and at least one dimension shows significantly better performance (i.e., shorter construction period, lower cost, higher quality compliance rate, or lower safety risk), then solution A is Pareto-dominated, and solution B is a degraded solution that can be directly eliminated. If no other feasible solution for a certain set of procedures can improve the other optimization objectives without worsening at least one of its optimization objectives, then the solution meets the Pareto optimality criterion and belongs to the Pareto non-dominated solution. Based on this criterion, this application selects all non-dominated solutions from the feasible solution space to form a frontier construction procedure solution set, which serves as a high-quality parent population for subsequent population growth and deep global optimization, ensuring that the optimization process takes into account both multi-objective balance and solution diversity.

[0065] This step, through a two-stage optimization model of first screening high-quality frontier solutions and then expanding the population in a targeted manner, effectively compresses the scope of ineffective searches. At the same time, it achieves in-depth mining in the neighborhood of high-quality solutions, taking into account both the efficiency and optimality of global optimization, and can efficiently output a comprehensive optimal construction procedure scheme that meets the engineering requirements.

[0066] Furthermore, based on the set of cutting-edge construction procedure schemes, population growth expansion and global search optimization are performed to determine the target construction procedure optimization scheme, including: using the set of cutting-edge construction procedure schemes as parent solutions for crossover and mutation growth to generate a construction procedure scheme growth population; and using the multi-objective function for construction effect evaluation to perform a global search optimization on the construction procedure scheme growth population to determine the target construction procedure optimization scheme.

[0067] Specifically, the parent solution refers to the cutting-edge construction procedure scheme obtained through Pareto selection. As the initial high-quality parent for population expansion and iterative evolution, its process logic and resource allocation mode have multi-objective comprehensive advantages and form the basis for generating derivative solutions. Crossover and mutation growth refers to the population expansion operation that simulates the mechanism of biological evolution. It includes two core operations: crossover and mutation. Crossover involves exchanging local process sequences of different schemes to generate new schemes, while mutation involves adjusting local process nodes or execution order of a single scheme to generate new schemes. This is used to expand the solution search space and discover potential better solutions.

[0068] Specifically, all solutions in the frontier construction process solution set are used as the initial parent solutions. The process nodes of each solution are linearly encoded according to topological order, converting them into ordered encoding strings that support genetic operations, while preserving the temporal dependency constraints and attribute parameters of each process. Then, crossover and mutation growth operations are performed. The crossover operation uses a two-point crossover method, randomly selecting two process sites from two parent solutions, exchanging local process sequence fragments between the sites, simultaneously verifying the temporal dependency validity of the new sequence, repairing logically conflicting process sequences, and generating crossover-derived solutions. The mutation operation uses a site mutation method, randomly selecting local process segments from a single parent solution, and generating mutated derivative solutions by fine-tuning the process sequence, replacing alternative processes with the same function, and adding or removing auxiliary quality control processes, while controlling the crossover probability and mutation... Within a preset probability range, the diversity and stability of the population are balanced. All newly generated schemes are merged with the original parent schemes, and invalid schemes are eliminated after mandatory constraint compliance verification to form a growth population of construction procedure schemes. Then, the multi-objective function for construction effect evaluation is called to calculate the comprehensive fitness of each scheme in the growth population. Based on this, a global search optimization is carried out in multiple rounds of iteration. In each round of iteration, fitness evaluation, elite selection, crossover mutation, and population update operations are performed in sequence. The elite schemes with the highest fitness in each generation are directly retained to the next generation to continuously improve the overall performance of the population. When the number of iterations reaches the preset maximum value, or when the optimal fitness of the population does not improve significantly for several consecutive generations, the iteration is terminated. Finally, the scheme with the highest comprehensive fitness is selected from the terminated generation population and determined as the target construction procedure optimization scheme.

[0069] This step expands the population by using high-quality cutting-edge solutions as parent generations, and combines multi-objective iterative optimization to achieve solution evolution. It not only relies on high-quality foundations to ensure the rationality of the optimization direction, but also expands the search boundary through cross-mutation. It can output composite vibration reduction structure construction procedure solutions with better overall performance while taking into account computational efficiency.

[0070] In summary, the knowledge graph-based construction process optimization method for composite vibration reduction structures provided in this application has the following technical effects: 1. By linking knowledge graph construction, rule reasoning and filtering, case matching and global multi-objective optimization into a closed-loop optimization process, knowledge-driven approach ensures the logical compliance of the solution, and data-driven approach reuses historical experience. Finally, it automatically generates an optimized solution for the construction process of composite vibration reduction structure that takes into account multiple objectives such as schedule, cost and quality, solving the problems of low efficiency and difficulty in global collaborative optimization in traditional manual arrangement.

[0071] 2. By first defining four types of construction entities—process, material, constraint, and attribute—and then using a rule matching channel to quickly screen keywords and a semantic annotation channel to finely identify semantic attributes and integrate the output, rough text is transformed into a precise set of construction knowledge entities with types and attributes. This overcomes the shortcomings of pure rule methods, which have many ambiguities, and pure semantic methods, which have low recall.

[0072] 3. By constructing a multi-objective evaluation function oriented towards construction needs such as schedule, cost, and quality, the fitness of feasible construction procedure schemes is evaluated, and the global search for optimization is driven by multi-objective fitness to select the construction procedure scheme with the best comprehensive efficiency from a large number of feasible schemes, thus solving the problem that experience-based optimization considers only one objective and is prone to getting trapped in local optima.

[0073] Example 2, based on the same inventive concept as the knowledge graph-based composite vibration reduction structure construction process optimization method in the aforementioned examples, such as... Figure 2As shown in the figure, this application provides a knowledge graph-based construction procedure optimization system for composite vibration reduction structures. The system includes: a knowledge graph construction module 11, used to extract entities and model knowledge based on the construction procedure information of the composite vibration reduction structure to construct a knowledge graph of the construction procedure of the vibration reduction structure; a rule reasoning and filtering module 12, used to input the current engineering structure parameters, and perform rule reasoning and filtering on the knowledge graph of the construction procedure of the vibration reduction structure based on the current engineering structure parameters to obtain an initial set of construction procedure schemes; a scheme space construction module 13, used to match candidate construction procedure schemes from the composite vibration reduction structure engineering case library according to the current engineering structure parameters, and construct a feasible construction procedure scheme space based on the initial set of construction procedure schemes and the candidate set of construction procedure schemes; and a construction procedure optimization module 14, used to perform a global search and optimization on the feasible construction procedure scheme space to determine the target construction procedure optimization scheme, and to control the construction of the composite vibration reduction structure through the target construction procedure optimization scheme.

[0074] Furthermore, the knowledge graph construction module 11 is also used to perform the following steps: perform text cleaning processing on the construction process information of the composite damping structure to obtain a structured construction process text library, wherein the text cleaning steps include non-text noise removal, sentence segmentation processing, and terminology unification; construct a dual-channel entity extraction strategy, wherein the dual-channel entity extraction strategy is executed in series by a rule matching channel and a semantic annotation channel; extract entities from the structured construction process text library according to the dual-channel entity extraction strategy to obtain a set of construction knowledge entities; and perform semantic relationship analysis and knowledge modeling output based on the set of construction knowledge entities to construct a knowledge graph of the damping structure construction process.

[0075] Furthermore, the knowledge graph construction module 11 is also used to perform the following steps: defining construction entity types, which include process entities, material entities, constraint entities, and attribute entities; using the rule matching channel to perform keyword matching on the structured construction process text library according to the construction entity type to obtain a construction entity type keyword set; performing semantic recognition annotation on the construction entity type keyword set based on the semantic annotation channel to obtain a construction entity semantic attribute set; and associating and integrating the construction entity semantic attribute set with the construction entity type to obtain a construction knowledge entity set.

[0076] Furthermore, the knowledge graph construction module 11 is also used to perform the following steps: defining entity relationship types, including temporal dependency relationships, spatial constraint relationships, and process resource requirement relationships; performing semantic relationship analysis on the construction knowledge entity set according to the entity relationship types to obtain a construction entity semantic relationship set; associating and storing the construction knowledge entity set according to the construction entity semantic relationship set to obtain a construction entity triplet unit set; and instantiating the construction entity triplet unit set and filling it into the graph ontology model for knowledge modeling to construct a knowledge graph of vibration reduction structure construction processes.

[0077] Furthermore, the rule reasoning and filtering module 12 is also used to perform the following steps: inputting the current engineering structure parameters into the vibration reduction structure construction process knowledge graph for mapping and transformation to obtain the initial engineering anchor node set; based on the current engineering structure parameters, pre-setting process node reasoning rules and process constraint filtering rules, wherein the process node reasoning rules include triggering conditions and reasoning actions; triggering from the initial engineering anchor node set, performing path reasoning and filtering on the vibration reduction structure construction process knowledge graph according to the process node reasoning rules and process constraint filtering rules to obtain the initial construction process scheme set.

[0078] Furthermore, the scheme space construction module 13 is also used to perform the following steps: retrieval encoding of the current engineering structure parameters to generate a multi-dimensional retrieval vector; retrieval matching in the composite vibration reduction structure engineering case library based on the multi-dimensional retrieval vector to obtain a graph structure similarity set; and case screening and extraction in the composite vibration reduction structure engineering case library according to the graph structure similarity set to determine a candidate construction procedure scheme set.

[0079] Furthermore, the construction procedure optimization module 14 is also used to perform the following steps: constructing a multi-objective function for evaluating construction effect based on the construction requirements of the composite vibration reduction structure; using the multi-objective function for evaluating construction effect to evaluate the fitness of the feasible construction procedure scheme space and obtain the fitness of multiple construction procedure schemes; and performing a global search and optimization on the feasible construction procedure scheme space based on the fitness of the multiple construction procedure schemes to determine the target construction procedure optimization scheme.

[0080] Furthermore, the construction procedure optimization module 14 is also used to perform the following steps: optimize the feasible construction procedure scheme space according to the fitness of the multiple construction procedure schemes to obtain a frontier construction procedure scheme set; and perform population growth expansion and global search optimization based on the frontier construction procedure scheme set to determine the target construction procedure optimization scheme.

[0081] Furthermore, the construction procedure optimization module 14 is also used to perform the following steps: cross-mutate and grow the frontier construction procedure scheme set as the parent solution to generate a construction procedure scheme growth population; and use the construction effect evaluation multi-objective function to perform a global search and optimization on the construction procedure scheme growth population to determine the target construction procedure optimization scheme.

[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs, characterized in that, The method includes: Based on the construction process information of composite damping structures, entity extraction and knowledge modeling are performed to construct a knowledge graph of the construction process of damping structures. Input the current engineering structure parameters, and perform rule-based reasoning and filtering on the knowledge graph of the construction process of the vibration reduction structure based on the current engineering structure parameters to obtain an initial set of construction process schemes; According to the current engineering structural parameters, a set of candidate construction procedures is matched from the composite vibration reduction structure engineering case library. Based on the initial set of construction procedures and the set of candidate construction procedures, a space of feasible construction procedures is constructed. A global search and optimization of the feasible construction procedure schemes is performed to determine the target construction procedure optimization scheme, and the construction control of the composite vibration reduction structure is carried out through the target construction procedure optimization scheme.

2. The method for optimizing the construction process of composite vibration reduction structures based on knowledge graphs as described in claim 1, characterized in that, Based on the construction process information of composite damping structures, entity extraction and knowledge modeling are performed to construct a knowledge graph of the construction process of damping structures, including: Text cleaning processing is performed on the construction process information of the composite vibration reduction structure to obtain a structured construction process text library. The text cleaning steps include non-text noise removal, sentence segmentation, and terminology standardization. A dual-channel entity extraction strategy is constructed, wherein the dual-channel entity extraction strategy is executed in series by a rule matching channel and a semantic annotation channel; The structured construction process text library is used to extract entities according to the dual-channel entity extraction strategy described above to obtain a set of construction knowledge entities. Based on the aforementioned set of construction knowledge entities, semantic relationship analysis and knowledge modeling are performed to output a knowledge graph of construction procedures for vibration reduction structures.

3. The method for optimizing the construction process of composite vibration reduction structures based on knowledge graphs as described in claim 2, characterized in that, The structured construction procedure text library is subjected to entity extraction according to the aforementioned dual-channel entity extraction strategy to obtain a set of construction knowledge entities, including: Define construction entity types, which include process entities, material entities, constraint entities, and attribute entities; The rule matching channel is used to perform keyword matching on the structured construction process text library according to the construction entity type to obtain a keyword set for construction entity type; Based on the semantic annotation channel, semantic recognition and annotation are performed on the keyword set of the construction entity type to obtain the semantic attribute set of the construction entity; The semantic attribute set of the construction entity is associated and integrated with the construction entity type to obtain a set of construction knowledge entities.

4. The method for optimizing the construction process of composite vibration reduction structures based on knowledge graphs as described in claim 3, characterized in that, Based on the aforementioned set of construction knowledge entities, semantic relationship analysis and knowledge modeling are performed to output a knowledge graph of construction procedures for vibration-damping structures, including: Define entity relationship types, including temporal dependency relationships, spatial constraint relationships, and process resource requirement relationships; Semantic relationship analysis is performed on the set of construction knowledge entities according to the entity relationship type to obtain the set of construction entity semantic relationships. Based on the semantic relationship set of construction entities, the set of construction knowledge entities is associated and stored to obtain a set of construction entity triplet units; The construction entity triplet unit set is instantiated and filled into the graph ontology model for knowledge modeling, thus constructing a knowledge graph of the construction process of the vibration reduction structure.

5. The method for optimizing the construction process of composite vibration reduction structures based on knowledge graphs as described in claim 1, characterized in that, Based on the current engineering structural parameters, rule-based reasoning and filtering are performed on the knowledge graph of the vibration reduction structure construction process to obtain an initial set of construction process schemes, including: The current engineering structural parameters are input into the knowledge graph of the construction process of the vibration reduction structure for mapping and transformation to obtain the initial engineering anchoring node set; Based on the current engineering structure parameters, preset process node reasoning rules and process constraint filtering rules are established, wherein the process node reasoning rules include triggering conditions and reasoning actions; Starting from the initial project anchor node set, path reasoning and filtering are performed on the knowledge graph of the vibration reduction structure construction process according to the process node reasoning rules and process constraint filtering rules to obtain the initial construction process scheme set.

6. The method for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs as described in claim 1, characterized in that, Based on the current engineering structural parameters, a set of candidate construction procedure schemes is matched from the composite vibration reduction structure engineering case library, including: The current engineering structure parameters are retrieved and encoded to generate a multi-dimensional retrieval vector; Based on the multidimensional retrieval vector, a retrieval and matching process is performed within the composite vibration reduction structure engineering case library to obtain a graph structure similarity set. Based on the structural similarity set described above, case studies are extracted from the composite vibration reduction structure engineering case library to determine the candidate construction procedure scheme set.

7. The method for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs as described in claim 1, characterized in that, A global search and optimization of the feasible construction procedure scheme space is performed to determine the target construction procedure optimization scheme, including: Based on the construction requirements of composite vibration reduction structures, a multi-objective function for evaluating construction effectiveness is constructed. The fitness of the feasible construction procedure scheme space is evaluated using the aforementioned multi-objective function for construction effect evaluation, and the fitness of multiple construction procedure schemes is obtained. Based on the adaptability of the multiple construction process schemes, a global search and optimization is performed on the space of feasible construction process schemes to determine the target construction process optimization scheme.

8. The method for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs as described in claim 7, characterized in that, Based on the fitness of the multiple construction process schemes, a global search and optimization is performed on the feasible construction process scheme space to determine the target construction process optimization scheme, including: Based on the adaptability of the multiple construction process schemes, the feasible construction process scheme space is optimized to obtain the frontier construction process scheme set; Based on the aforementioned set of cutting-edge construction procedure schemes, population growth expansion and global search optimization are performed to determine the optimal scheme for the target construction procedure.

9. The method for optimizing construction procedures of composite vibration reduction structures based on knowledge graphs as described in claim 8, characterized in that, Based on the aforementioned set of cutting-edge construction procedure schemes, population growth expansion and global search optimization are performed to determine the optimal scheme for the target construction procedure, including: The aforementioned set of cutting-edge construction procedure schemes is used as the parent solution for crossover and mutation growth to generate a population of construction procedure scheme growth. The multi-objective function for evaluating construction effectiveness is used to perform a global search and optimization on the growth population of the construction procedure scheme to determine the optimal scheme for the target construction procedure.

10. A knowledge graph-based system for optimizing construction procedures of composite vibration-damping structures, characterized in that, The system is used to execute the knowledge graph-based construction procedure optimization method for composite vibration reduction structures according to any one of claims 1 to 9, the system comprising: The knowledge graph construction module is used to extract entities and model knowledge based on the construction process information of composite vibration reduction structures, and to construct a knowledge graph of the construction process of vibration reduction structures. The rule reasoning and filtering module is used to input the current engineering structure parameters and perform rule reasoning and filtering on the knowledge graph of the construction process of the vibration reduction structure based on the current engineering structure parameters to obtain an initial set of construction process schemes. The scheme space construction module is used to match a set of candidate construction procedures from the composite vibration reduction structure engineering case library according to the current engineering structural parameters, and to construct a feasible construction procedure scheme space based on the initial construction procedure scheme set and the candidate construction procedure scheme set. The construction procedure optimization module is used to perform a global search and optimization of the feasible construction procedure scheme space, determine the target construction procedure optimization scheme, and control the construction of the composite vibration reduction structure through the target construction procedure optimization scheme.

Images