Method, device and equipment for processing concrete pouring construction data and medium

Abstract

The invention discloses a concrete pouring construction data processing method and device, equipment and a medium. The method comprises the steps of obtaining a knowledge file of a concrete pouring construction field; identifying the knowledge file to obtain a plurality of entities related to concrete pouring construction and a semantic relationship among the plurality of entities, and constructing a concrete pouring construction structured knowledge base based on the plurality of entities and the semantic relationship among the plurality of entities; obtaining multi-modal sensing data of a concrete pouring construction site; according to the concrete pouring construction structured knowledge base, diagnosing the multi-modal sensing data to obtain a diagnosis result; according to the diagnosis result, an adjustment strategy for the concrete pouring construction site is generated; the target construction simulation model is adopted to conduct simulation calculation on the strategy to obtain a simulation result, the simulation result serves as a simulation analysis result to be displayed to a user, through the embodiment of the invention, various changes in the actual construction process can be accurately reflected, and therefore suggestions needing to be adjusted are determined.

Description

Technical Field

[0001] This invention relates to the field of digital simulation, specifically to a method, apparatus, equipment, and medium for processing concrete pouring construction data. Background Technology

[0002] Digital simulation refers to the use of computer models to model and simulate physical, working, chemical, and other systems in the real world in order to predict their behavior, performance, or evolution under different conditions.

[0003] In related technologies, digital simulation has been widely used in the construction field. In the construction process, digital simulation technology is generally used in advance to simulate the working conditions, so that engineers can adjust the steps or plans when construction begins.

[0004] However, in actual production, the quality of concrete in construction is constrained by a variety of factors. Existing progress simulations cannot simulate the steps that require dynamic adjustments due to non-compliance with the process. Furthermore, even if simulation data is obtained, it still needs to be calculated by engineers before it can be applied, resulting in poor conversion efficiency of the obtained simulation data. Summary of the Invention

[0005] In view of the above problems, a method, apparatus, electronic equipment, storage medium, and product for processing concrete pouring construction data are proposed to overcome or at least partially solve the above problems, including: A method for processing concrete pouring construction data, the method comprising: Obtain knowledge documents related to concrete pouring construction; The knowledge file is identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between these entities. Based on these entities and their semantic relationships, a structured knowledge base for concrete pouring construction is constructed. Acquire multimodal sensor data at the concrete pouring construction site; Based on the structured knowledge base of concrete pouring construction, the multimodal sensing data is diagnosed to obtain diagnostic results; Based on the diagnostic results, an adjustment strategy is generated for the concrete pouring construction site. The adjustment strategy is simulated and calculated using a target construction simulation model to obtain simulation results, which are then presented to the user as simulation analysis results.

[0006] Optionally, after using a target construction simulation model to perform simulation calculations on the adjustment strategy and obtaining simulation results, the process includes: If the multimodal sensing data changes, then a first query information is constructed based on the changed multimodal sensing data; The first query information and the preset project adjustment prompts are input into the trained concrete pouring construction model for analysis to obtain the change process information output by the concrete pouring construction model. The target construction simulation model is adjusted based on the changed process information to obtain the adjusted target construction simulation model.

[0007] Optionally, the step of using a target construction simulation model to perform simulation calculations on the adjustment strategy and obtaining simulation results includes: The target construction simulation model is instantiated using the simulation parameter information, and the target construction simulation model is run several times in the simulation environment to obtain several candidate simulation construction results. The target simulation result is determined from several candidate simulation construction results based on the multimodal information.

[0008] Optionally, the knowledge document is identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, including: Obtain user-defined data; wherein, the user-defined data includes multiple entity types related to concrete pouring construction; Based on the user-defined data, the knowledge file is identified to obtain multiple entities and the semantic relationships between the multiple entities.

[0009] Optionally, the step of diagnosing the multimodal sensing data based on a structured knowledge base for concrete pouring construction to obtain diagnostic results includes: The first entity corresponding to the multimodal sensing data is determined from the structured knowledge base of concrete pouring construction. Based on the semantic relationship of the first entity, determine the second entity related to the first entity; Based on the knowledge content corresponding to the second entity, the multimodal sensing data is diagnosed to obtain a diagnostic result.

[0010] Optionally, determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction includes: Based on the concrete pouring construction model, the first entity corresponding to the multimodal sensing data is determined from the structured knowledge base of the concrete pouring construction. Before determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction based on the concrete pouring construction model and the multimodal sensing data, the method further includes: Training sample data is generated using a pre-built large model and the structured knowledge base of concrete pouring construction. The pre-set concrete pouring construction model is trained based on the training sample data.

[0011] Optionally, acquire multimodal sensing data from the concrete pouring construction site, including: Visual information about the concrete pouring site and the operational behavior of the personnel are obtained through construction vision sensors. Vibration sensors are used to acquire physical parameters during the concrete vibration process. The concrete vibration quality is determined based on visual information and the physical parameters. The concrete pouring parameters are obtained through a distance measuring sensor; The visual information, the operational behavior information, the physical parameters, the concrete vibration quality, and the pouring parameters are constructed into multimodal sensing data.

[0012] Optionally, the entity type includes any one or more of the following: construction action, construction equipment entity, construction method, construction material, technical indicators, construction environmental conditions, construction quality defects, construction quality specifications, and construction safety risks.

[0013] A device for processing concrete pouring construction data includes: The engineering knowledge acquisition module is used to acquire knowledge documents in the field of concrete pouring construction. The knowledge base construction module is used to identify the knowledge file, obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, and construct a structured knowledge base for concrete pouring construction based on the multiple entities and the semantic relationships between the multiple entities. The multimodal data acquisition module is used to acquire multimodal sensor data at the concrete pouring construction site. The diagnostic module is used to diagnose the multimodal sensing data based on the structured knowledge base of concrete pouring construction and obtain diagnostic results. The adjustment strategy module is used to generate an adjustment strategy for the concrete pouring construction site based on the diagnostic results. The simulation calculation module is used to perform simulation calculations on the adjustment strategy using the target construction simulation model, obtain simulation results, and display the simulation results to the user as simulation analysis results.

[0014] An electronic device includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the method described above.

[0015] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0016] A computing program product includes a computer program that, when executed by a processor, implements the method described above.

[0017] The embodiments of the present invention have the following advantages: In this embodiment of the invention, knowledge documents in the field of concrete pouring construction are acquired; the knowledge documents are identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between these entities; and a structured knowledge base for concrete pouring construction is constructed based on these entities and their semantic relationships; multimodal sensor data from the concrete pouring construction site is acquired; the multimodal sensor data is diagnosed according to the structured knowledge base for concrete pouring construction to obtain diagnostic results; an adjustment strategy for the concrete pouring construction site is generated based on the diagnostic results; a target construction simulation model is used to simulate and calculate the adjustment strategy to obtain simulation results, and the simulation results are displayed to the user as simulation analysis results. This achieves the acquisition of multimodal data from the construction site and the dynamic simulation of the multimodal data and its adjustment strategy, which can accurately reflect various changes in the actual construction process, improve the efficiency of simulation calculation and the efficiency of converting simulation results. Attached Figure Description

[0018] Figure 1 This is a flowchart of the steps of a method for processing concrete pouring construction data according to an embodiment of the present invention; Figure 2 This is a structural block diagram of a concrete pouring construction data processing device provided in an embodiment of the present invention. Detailed Implementation

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

[0020] Reference Figure 1 The diagram illustrates a flowchart of a method for processing concrete pouring construction data according to an embodiment of the present invention, which may specifically include the following steps: Step 101: Obtain knowledge documents related to concrete pouring construction.

[0021] In some examples, the knowledge document can be an authoritative document containing structured or unstructured knowledge generated throughout the entire lifecycle of concrete properties, production, quality, and construction.

[0022] Collect various types of knowledge in the field of concrete pouring construction. These types of knowledge may include any one or more of the following: characteristics of different types of concrete, pouring process standards, special construction plans, construction specifications, quality acceptance standards, equipment parameter manuals, and construction logs.

[0023] Step 102: Identify the knowledge file to obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, and construct a structured knowledge base for concrete pouring construction based on the multiple entities and the semantic relationships between the multiple entities.

[0024] In some examples, the entity can be a core concept or action that can be named or described in concrete pouring construction, and the semantic relationship can be a logical association between two entities.

[0025] The collected knowledge documents are scanned and identified. Multiple entities related to concrete pouring construction and the semantic relationships between them are extracted using a preset model. Based on the semantic relationships between entities, two entities are associated to form multiple triples of head entity-semantic relationship-tail entity. These triples are stored to construct a structured knowledge base for concrete pouring construction.

[0026] In practical applications, OCR (Optical Character Recognition) technology is used to convert scanned documents containing various types of knowledge into editable text. This editable text is then cleaned, denoised, and standardized. Pre-trained or fine-tuned named entity recognition models (such as BERT-based models) or LLM (Limited Least Meaning) are used for few-shot hints to identify and classify predefined entities from the text. Subsequently, relation extraction models or LLM are used to determine the semantic relationships between the identified entities in the sentence, forming triples of head entity—relation—tail entity. For example, from the standard text "The spacing between vibratory rod insertion points should not be greater than 1.5 times the effective radius," the following can be extracted: vibratory rod insertion point spacing—should not be greater than—1.5 times the effective radius.

[0027] Furthermore, the extracted entities can be cleaned and integrated. Aliases and abbreviations pointing to the same entity from different data sources can be aligned to eliminate ambiguity. Simultaneously, descriptive logic or rule-based inference engines are used to check for knowledge contradictions. For example, if the final extracted triple "maintenance time less than 3 days" conflicts with the specification clause "maintenance time at least greater than or equal to 7 days," all contradictory knowledge will be replaced with the standard specification content. The processed high-quality triples, along with corresponding traceability information (such as source document, version, extraction model, confidence level, and data entry personnel), are imported into a graph database (such as Neo4j) for storage.

[0028] In some embodiments of the present invention, the knowledge document is identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, including: Sub-step 11: Obtain user-defined data; wherein, the user-defined data includes multiple entity types related to concrete pouring construction.

[0029] In some examples, the entity type can be a class of concepts or actions that indicate concrete pouring construction.

[0030] The system retrieves data that engineers have predefined in the backend of the system. This data includes multiple entity types related to concrete pouring construction, such as construction procedures, construction methods, construction equipment, materials, technical indicators, environmental conditions, quality defects, quality specifications, and safety risks.

[0031] Sub-step 12: Based on the user-defined data, identify the knowledge file to obtain multiple entities and the semantic relationships between the multiple entities.

[0032] Based on the user-defined data, the knowledge file is identified. By identifying the text data in the knowledge file, entity texts that conform to multiple entity types are identified in the text data. Based on the entity texts, the semantic relationships between the entity texts are determined.

[0033] In practical applications, engineers prioritize defining core entity types in the operational backend, such as construction procedures (formwork erection, rebar tying, concrete pouring, vibration, curing), construction methods (e.g., roller-compacted concrete, shotcrete), construction equipment (pump truck, vibrator), materials (cement, aggregate, admixtures), technical indicators (strength, slump), environmental conditions (ambient temperature, wind speed, probability of precipitation), quality defects (honeycomb, pitting, cold joints), quality specifications (e.g., specific clauses of the "Code for Acceptance of Construction Quality of Concrete Structures" GB50204), and safety risks (collapse, electrical leakage), etc. Furthermore, to facilitate faster identification, engineers also... The semantic relationships between entities will be defined first, such as "temporal relationship" (reinforcement binding, preceding concrete pouring), "use relationship" (vibration process, using machinery, vibrator), "compliance relationship" (vibration, must be followed, specification A.1.2), causal relationship (insufficient vibration, resulting in honeycombing) (parameter P exceeds the threshold, resulting in defect Q), subordinate relationship (a certain type of concrete, belonging to a certain concrete category), and applicability relationship (a certain construction method, applicable to the concrete pouring process), etc. Then, the knowledge document will be identified, and multiple entities and the semantic relationships between them will be determined from the identified text data.

[0034] Step 103: Obtain multimodal sensor data from the concrete pouring construction site.

[0035] In some examples, the multimodal sensing data can be visual information of the concrete pouring construction site and physical parameters characterizing the concrete processing process, acquired by various types of sensors.

[0036] Multiple sensors are deployed around the construction site. Visual sensors are used to acquire real-time visual information about the concrete pouring site, which can characterize the changes in the concrete during the pouring process. Vibration sensors are used to acquire physical parameters of the concrete processing.

[0037] In some embodiments of the present invention, acquiring multimodal sensing data at a concrete pouring construction site includes: Sub-step 21: Obtain visual information of the concrete pouring construction site and information on the work behavior of the personnel through construction vision sensors.

[0038] In some examples, the construction vision sensor can be an industrial camera, the visual information can be visual scene information that can be captured at the concrete pouring construction site, and the work behavior information can be the actions that personnel are performing in real time.

[0039] Industrial cameras can acquire real-time visual information of the concrete pouring construction site, and can also capture the specific actions of the personnel carrying out the construction. These actions may include any one or more of the following: mixing, pouring, and movement trajectory.

[0040] In practical applications, high-definition cameras are deployed at key points on the concrete pouring surface, such as the four corners of the pouring area, main construction passages, and around the concrete placing boom. This allows for clear capture of visual information including the operational behavior of construction workers, the operating status of equipment, and the surface condition of the concrete during the pouring process. Drones, wearable devices, and robots can also be used as supplementary acquisition methods. A portion of the collected data is labeled for use in visual model training. The labeling process is divided into several levels: 1) Pixel-level: Instance segmentation and contour delineation of key equipment and personnel. 2) Frame-level: Multi-label process classification of images. 3) Video segment-level: Temporal behavior localization and recognition, accurately labeling the start and end times and types of behaviors. Using object detection models (such as YOLO), key elements on the work surface are identified and located in real time, such as workers, vibrators, pump pipes, and poured areas. Semantic segmentation models (such as U-Net) are used to accurately divide the areas into "areas being poured," "areas that have been vibrated," and "areas to be poured." Through posture estimation and behavior recognition models, the working behavior of construction equipment and workers (such as whether they are mixing, vibrating, or moving) is analyzed. Scene graph reasoning is introduced to not only identify objects but also understand the spatial relationships between them (e.g., a worker holding a vibrator touching a concrete surface is "vibrating," while a vibrator placed on the ground is "idle"), and this is converted into a textual scene description (e.g., "ID1 worker is using a vibrator to vibrate in area A for more than 30 seconds").

[0041] Sub-step 22: Obtain the physical parameters during the concrete vibration process using vibration sensors.

[0042] In some examples, the physical parameters can be physical parameters that characterize the operation of the vibrating machinery during the concrete compaction process.

[0043] Miniature vibration sensors are installed on each concrete vibrator to monitor key physical parameters such as the depth of insertion into the concrete, vibration frequency, amplitude, and vibration duration in real time during the vibration process.

[0044] Sub-step 23: Determine the concrete vibration quality based on visual information and the physical parameters.

[0045] Before determining the concrete vibration quality, the visual information and physical parameters related to the concrete collected by sensors can be compared with data in a pre-set database to determine the concrete vibration quality.

[0046] In practical applications, the system utilizes surface images of concrete after vibration captured by miniature cameras and key physical parameters obtained from miniature vibration sensors installed on each concrete vibrator. A standardized image database reflecting the surface and interface state of the concrete after vibration is first constructed (capturing quality characteristics such as the number and distribution of air bubbles, the fullness of the cement paste, the exposed and uniform nature of aggregates, surface smoothness, and bleeding and laitance caused by over-vibration). Under these requirements, surface images of concrete with different mix proportions and under different environmental conditions (such as day / night, dry / wet) after vibration are systematically collected.

[0047] Domain experts, based on standards and experience, classify vibration compaction quality into four levels: "Excellent," "Good," "Medium," and "Poor," or quantify specific defects (e.g., "bubble density <5% is acceptable"). Based on this standard, images in the dataset are finely annotated. Annotation types should include: image classification labels (directly providing the overall quality level); bounding box annotations (outlining defect areas, such as large areas of honeycomb or pitted surfaces); and more refined pixel-level segmentation (e.g., precisely delineating the outline of each bubble or oozing area). Building upon the database, convolutional neural networks (such as ResNet and EfficientNet) are used to classify the overall vibration compaction quality level. Based on the collected standardized image dataset, a deep learning quality evaluation model is constructed, consisting of data annotation and the definition of the quality level system. This step can serve as the foundation model, using object detection models (such as YOLO and Faster R-CNN) for defect localization; and image segmentation models (such as U-Net and DeepLab series) for quantitative analysis of bubble and oozing areas. It should be noted that the visual information is not limited to measuring the quality of concrete vibration, but can also be used to determine some aspects of the concrete construction site, such as real-time construction technology and construction materials. In addition, it can also determine the site environment characteristics such as the terrain, geology, and hydrology around the concrete construction site, or the storage location of construction materials and the specific placement of materials during the construction process. This invention does not impose any specific limitations.

[0048] Sub-step 24: Obtain concrete pouring parameters using a distance sensor.

[0049] In some examples, the pouring parameters may be the pouring range and pouring boundaries of the concrete on the plane, and / or the flatness and pouring height after pouring.

[0050] Distance sensors deployed at the concrete pouring site can acquire multiple pouring parameters after the concrete is poured.

[0051] In practical applications, multiple laser rangefinders and electronic levels (another type of distance sensor) are deployed at the concrete pouring site. The laser rangefinders are mounted on flexibly adjustable supports and, through rotational scanning, acquire high-precision, real-time distance data between different locations within the pouring area and the rangefinder, thereby determining the pouring range and boundaries of the concrete on the plane. The electronic levels are distributed at regular intervals around the perimeter of the pouring area to measure the flatness and height of the concrete surface.

[0052] Sub-step 25: The visual information, the operation behavior information, the physical parameters, the concrete vibration quality, and the pouring parameters are constructed into multimodal sensing data.

[0053] The sensor collects visual information, personnel operation behavior information, physical parameters during pouring, concrete vibration quality, pouring parameters, and other information to construct multimodal sensing information.

[0054] In some examples, the multimodal sensing information also includes ambient parameters of the construction environment.

[0055] By using a variety of environmental sensors pre-installed around the concrete pouring area, environmental parameters of the concrete pouring area can be collected in real time, and other parameters in the multimodal sensing data besides the target parameter can be determined.

[0056] In practical applications, the above information is aggregated in real time or near real time, including the location of construction targets based on visual perception, the division of pouring areas, the types of work behaviors of construction personnel and equipment, the spatial relationship analysis results of construction targets, the analysis results of concrete pouring range and pouring surface height based on laser rangefinders and electronic levels, vibration operation parameters, and vibration quality analysis results based on concrete surface images. Furthermore, environmental sensors can be used to collect parameters of the surrounding environment. Temperature and humidity sensors, wind speed sensors, and air pressure sensors are reasonably distributed around the concrete pouring area to collect environmental temperature and humidity data, measure wind speed at the construction site, and monitor atmospheric pressure. Other parameters in the multimodal sensing data besides the target parameters are determined. The "environmental temperature and humidity data, construction site wind speed, and atmospheric pressure" are converted into structured output to form a "snapshot of concrete pouring operation status" (multimodal sensing information).

[0057] Step 104: Based on the structured knowledge base of concrete pouring construction, diagnose the multimodal sensing data to obtain diagnostic results.

[0058] In some embodiments of the present invention, the step of diagnosing the multimodal sensing data based on a structured knowledge base for concrete pouring construction to obtain diagnostic results includes: Sub-step 31: Determine the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction.

[0059] Sub-step 32: Based on the semantic relationship of the first entity, determine the second entity related to the first entity.

[0060] Sub-step 33: Based on the knowledge content corresponding to the second entity, diagnose the multimodal sensing data to obtain a diagnostic result.

[0061] In some examples, the knowledge content may be a strategy or method that the second entity should correctly execute.

[0062] Based on the acquired multimodal sensing data, a first entity is identified from the structured knowledge base for concrete pouring construction that contains information or parameters from the multimodal sensing data, and either of these can correspond to it. Then, based on the first entity and its semantic relationships, a second entity related to the first entity is determined. For example, if the multimodal sensing data acquired by the sensor is "pouring", the first entity "pouring" will be retrieved from the structured knowledge base for concrete pouring construction based on this sensing data, and then the corresponding second entity will be determined based on the relationships of the first entity.

[0063] Step 105: Based on the diagnostic results, generate an adjustment strategy for the concrete pouring construction site.

[0064] After obtaining the correct strategies or methods to be executed, the multimodal sensing data is diagnosed to determine the steps or methods that are erroneous in the multimodal sensing data, and based on the results of the diagnosis, the correct adjustment strategy for the concrete pouring construction site is generated.

[0065] In practical applications, it is combined with preset prompt word templates to form a complete natural language prompt. The fine-tuned concrete construction big language model is called to perform knowledge retrieval and logical reasoning. Key information in the "concrete pouring operation status snapshot" (such as "process: mixing", "location: slab number 101", "temperature: 35℃, mixing time: 25 minutes") is used as query conditions to perform real-time retrieval in the concrete pouring construction knowledge graph and quickly obtain the most relevant specification clauses, construction plans and other knowledge subgraphs.

[0066] After obtaining the correct strategies or methods to be implemented, the multimodal sensing data is diagnosed to determine the steps or methods that are erroneous in the multimodal sensing data. Based on the results of the diagnosis, a correct adjustment strategy is generated for the concrete pouring construction site. For example, if it is found that "concrete mixing time < 30 minutes" conflicts with the knowledge content of "concrete mixing time ≥ 30 minutes" in the specification, the correct adjustment strategy for the concrete pouring construction site can be determined based on the requirement that the concrete mixing time should be greater than 30 minutes.

[0067] Step 106: Using the target construction simulation model, perform simulation calculations on the adjustment strategy to obtain simulation results, and display the simulation results to the user as simulation analysis results.

[0068] In an embodiment of the present invention, a target construction simulation model is used to perform simulation calculations on the adjustment strategy, including: Select the target construction simulation model from the existing engineering simulation model library; In some examples, the engineering simulation model library may include a variety of pre-built reusable simulation models, including construction element simulation components, process simulation components, and technology simulation components. Furthermore, the engineering simulation model library may consist of an atomic model layer, a combined model layer, and a technology template layer. The target simulation model may refer to an executable simulation program that is matched, selected, and assembled from the engineering simulation model library by a construction simulation modeling agent, and capable of fully representing the current construction technology logic.

[0069] In practical applications, construction element simulation components can be stored in the atomic model layer. These components simulate the most basic, indivisible physical processes, mechanical behaviors, or logical operations in construction technology. Construction elements can include elements related to construction resources such as construction machinery, construction materials, and construction steps. Process simulation components can be stored in the composite model layer. These components simulate typical construction processes with complete functionality. Technology simulation components can be stored in the technology template layer. These components simulate frequently occurring, complete technologies in hydropower engineering, such as dam concrete pouring and TBM tunnel excavation. They are out-of-the-box, parameterized, and driven construction technology templates.

[0070] The target construction simulation model can be configured by connecting and configuring several simulation model components that meet the process requirements according to the logical relationships defined in the construction process information. It has complete parameter interfaces and clear input and output definitions, and is the core entity that is directly used to carry simulation parameters and drive simulation calculations.

[0071] In one optional embodiment of the present invention, the engineering simulation model library includes several simulation model components.

[0072] Among them, simulation model components can refer to the basic functional units that make up the engineering simulation model library. Each component encapsulates the simulation logic of a specific construction process, physical behavior, or logical operation.

[0073] The step of selecting a first construction simulation model from the existing engineering simulation model library based on the construction process information includes: In some examples, the construction process information refers to the structured output generated by the concrete pouring construction model based on the second query information and project analysis prompts, used to drive the construction of the simulation model. The construction process information includes key process elements such as the identified construction sequence, logical relationships between processes (timing, dependency, parallelism, etc.), required resource types, and their constraints, which can constitute the direct input for the simulation modeling agent to perform model matching and assembly. The second query information can refer to a structured query request for the concrete pouring construction model constructed by the master control agent based on multimodal sensor data acquired from the concrete pouring construction site. The second query information can combine the multimodal sensor data with a preset engineering logic framework to form a formatted instruction that can trigger the model to perform domain-specific analysis. An example of a project analysis prompt is: "Please generate the corresponding construction process plan based on the following project description"; the master control agent can be the system control center responsible for overall coordination and requirement analysis.

[0074] A11, semantically match the construction process information with the metadata of several simulation model components respectively, and select at least one simulation model component; A12, determine the assembly sequence of the components among the at least one simulation model component based on the construction process information; A13. Connect and configure the at least one simulation model component according to the assembly sequence of the components to obtain the first construction simulation model.

[0075] In this embodiment of the invention, the construction process information can be semantically matched with the metadata of several simulation model components to select at least one simulation model component.

[0076] The metadata of the simulation model components refers to structured tag data describing the functional characteristics, interface specifications, and applicable scenarios of each simulation model component. According to this invention, the metadata includes at least: component functional descriptions, such as concrete vibration simulation; a list of input and output parameters, such as parameter names, types, and units; performance constraints, such as the maximum simulable workload; association rules, such as requiring use with the "concrete pouring" component; and semantic tags, such as pouring process and mechanized construction.

[0077] In this embodiment of the invention, the assembly order of at least one simulation model component can be determined based on the construction process information, and the at least one simulation model component can be connected and configured according to the assembly order to obtain the target simulation model.

[0078] The component assembly sequence refers to the technical solution by which the construction simulation modeling agent, based on the logical relationships of the work processes defined in the construction technology information (such as timing constraints, dependencies, and parallel rules), performs topological sorting and connection configuration of the selected simulation model components. This sequence not only determines the execution order between components but also defines the data flow transmission path, event triggering conditions, and resource coordination mechanisms, ensuring that the assembled target simulation model accurately reflects the dynamic logic of the actual construction process. The construction simulation modeling agent can serve as a control center responsible for simulation model matching and assembly.

[0079] In practical applications, after receiving structured construction process information transmitted by the master control agent, the construction simulation modeling agent can match component resources in the simulation model library based on semantic similarity algorithm, and load a preset model association rule library, which includes engineering constraints, such as: the concrete pouring process needs to be associated with the vibration equipment model, and then perform double verification.

[0080] Next, based on the time sequence relationships, dependencies, and parallel operation requirements clearly defined in the construction process information, the most suitable construction element simulation components, process simulation components, and technology simulation components are selected from the pre-built simulation model library. According to the logical relationships of the construction process, such as pre- and post-constraints and parallel operation conditions, these independent simulation model components are assembled into a complete target simulation model that can reflect the actual construction process, and the data interfaces between each simulation model component are configured.

[0081] In practical applications, snapshots of concrete pouring operations (multimodal sensor information) and retrieved knowledge subgraphs (knowledge content) can be used together as context to fill preset prompt templates, forming a complete natural language prompt. This prompt performs knowledge retrieval and logical reasoning, requiring the use of a "thinking chain" model. It involves step-by-step reasoning based on structured knowledge provided by the concrete pouring construction structured knowledge base to conduct compliance diagnostics of concrete pouring construction and generate structured simulation model adjustment content. For example, the natural language prompt might be: "You are a concrete construction expert. The current construction site situation is as follows: The construction location is 'Placement 101'. The vision system has identified 'ID1 worker using a vibrator to vibrate in area A for more than 30 seconds,' and detected 'Ambient temperature: 35℃,' 'Vibrator insertion spacing greater than 50cm,' 'Air bubble coverage rate 3.5%,' and 'Bleeding area 0.8 square decimeters.' Please determine whether there are potential compliance risks based on relevant construction specifications and plans, using the concrete pouring construction model, and provide suggestions for adjusting the concrete pouring simulation model."

[0082] In some embodiments of the present invention, the step of performing simulation calculations using a first construction simulation model based on the multimodal sensing data to obtain a first simulation result includes: Sub-step 41 involves instantiating the first construction simulation model using the simulation parameter information and running the first construction simulation model several times in the simulation environment to obtain several candidate simulation construction results.

[0083] In some examples, the simulation parameter information can be presented in a structured form, such as key-value pairs or parameter tables, containing the initial state values, performance index data, random distribution parameters, and engineering constraint thresholds required for each model component. It is the core input data for realizing the instantiation and dynamic operation of the simulation model.

[0084] Sub-step 42: Determine the first simulation result from several candidate simulation construction results based on the multimodal information.

[0085] In some embodiments of the present invention, the step of selecting a target simulation construction result from several candidate simulation construction results based on the multimodal information includes: A21, based on the multimodal information, determine at least one of the following: construction process requirements, construction quality requirements, meteorological environmental conditions, construction resource data, and multimodal management requirements.

[0086] A22, Based on the candidate simulation construction results, determine at least one of the following: construction period, construction cost, resource utilization rate, and construction intensity corresponding to the candidate simulation construction results.

[0087] A23, taking at least one of the construction period, the construction cost, the resource utilization rate, and the construction intensity as the optimization objective, and taking at least one of the construction process requirements, the construction quality requirements, the meteorological environment conditions, the construction resource data, and the multimodal management requirements as the constraint conditions.

[0088] A24. Based on the optimization objective and the constraints, select the target simulation construction result from the candidate simulation construction results.

[0089] In this embodiment of the invention, at least one of the following is determined based on multimodal information: construction process requirements, construction quality requirements, meteorological environmental conditions, construction resource data, and multimodal management requirements.

[0090] Among them, construction process requirements can refer to the technical methods, operating procedures and process standards that must be followed in the multimodal construction process, including normative constraints such as specific construction method selection, process arrangement and technical parameter control.

[0091] Construction quality requirements can refer to the design standards, acceptance specifications, and performance indicators that the project deliverables must meet, covering quantitative or qualitative quality criteria such as material strength, structural dimensions, and durability.

[0092] Meteorological environmental conditions can refer to the natural environmental factors that affect operations during construction, including meteorological data such as temperature, precipitation, and wind speed, as well as site environmental characteristics such as topography, geology, and hydrology.

[0093] Construction resource data can refer to multimodal information on the allocation of resources such as manpower, equipment, materials and funds, including the quantity, specifications, supply plan and dynamic availability of various resources.

[0094] Multimodal management requirements can refer to the constraints on organization, schedule, cost and risk control in multimodal implementation, involving management rules such as project milestones, budget limits, safety standards and coordination mechanisms.

[0095] In this embodiment of the invention, at least one of the following can be determined based on the candidate simulation construction results: construction period, construction cost, resource utilization rate, and construction intensity.

[0096] Construction period can refer to the total multimodal time or critical path duration obtained through simulation calculations, including the start and end times of each process, floating time, and overall progress distribution characteristics.

[0097] Construction cost can refer to the total cost of resources consumed during the simulation process, including direct costs of labor, materials, and machinery, as well as indirect costs such as management and risks, and is reflected as a cumulative curve of cost changes over time.

[0098] Resource utilization rate refers to the actual efficiency of the use of various construction resources (manpower, equipment, site, etc.) during the simulation period. It can be quantitatively characterized by the ratio of resource occupation time to total available time or load balancing.

[0099] Construction intensity refers to the amount of work completed or resources consumed per unit of time, reflecting the density of the construction process. It can be measured by indicators such as the average daily volume of concrete poured or the output rate of equipment shifts.

[0100] In this embodiment of the invention, at least one of the following can be used as the optimization objective: construction period, construction cost, resource utilization rate and construction intensity. At least one of the following can be used as the constraint: construction process requirements, construction quality requirements, meteorological environmental conditions, construction resource data and multimodal management requirements. Based on the optimization objective and the constraint, the target simulation construction result is selected from several candidate simulation construction results.

[0101] In practical implementation, after analyzing the simulation result dataset, the construction simulation optimization agent can define one or more optimization objectives, such as the shortest construction period and the lowest cost, define adjustable decision variables, such as resource allocation schemes and process time arrangements, and load all the constraints that must be met, such as construction process logic and resource availability limits.

[0102] Then, multi-objective optimization algorithms, such as NSGA II (Non-dominated Sorting Genetic Algorithm II) or MOEA / D (Multi-Objective Evolutionary Algorithm based on Decomposition), can be run to explore the decision space globally and optimize locally, identify and output the Pareto optimal solution set that achieves the best balance among multiple objectives.

[0103] Ultimately, based on actual engineering needs and decision-making preferences, several representative candidate solutions are selected from the Pareto solution set, such as frontier solutions that balance construction period and cost. The complete target simulation construction results corresponding to these solutions are then output to provide quantitative basis for construction decisions.

[0104] In an embodiment of the present invention, the engineering simulation model library is constructed in the following manner: A31, Obtain historical construction information, and determine foundation construction operations, process logic information, and technological logic information based on the historical construction information.

[0105] A32, the basic construction operations are parametrically encapsulated to obtain the construction element simulation component with standardized input and output interfaces.

[0106] A33, determine the assembly sequence of construction elements among at least one of the construction element simulation components based on the process logic information.

[0107] A34, assemble at least one of the construction element simulation components according to the construction element assembly sequence to obtain the process simulation component.

[0108] A35, determine the process assembly sequence between at least one of the process simulation components based on the process logic information.

[0109] A36, assemble at least one of the process simulation components according to the process assembly sequence to obtain the process simulation component.

[0110] A37. An engineering simulation model library is constructed based on the construction element simulation component, the process simulation component, and the technology simulation component.

[0111] In this embodiment of the invention, historical construction information can be obtained, and foundation construction operations, process logic information, and technological logic information can be determined based on the historical construction information.

[0112] Historical construction information refers to valuable construction process records accumulated from past projects, including heterogeneous data from multiple sources such as construction logs, progress reports, resource allocation tables, equipment operation records, quality inspection reports, and completion acceptance documents. Historical construction information can form the initial experience base for building a simulation model library.

[0113] Basic construction operations refer to indivisible construction action units abstracted from historical construction information, such as "rebar tying," "formwork installation," and "concrete vibration." Each operation has a clear action definition, resource requirements, and time consumption characteristics, serving as the direct basis for constructing the construction element simulation components in the atomic model layer.

[0114] Process logic information refers to the organizational rules and temporal relationships between multiple basic construction operations summarized from historical construction information, such as the sequence of "formwork erection - rebar tying - pouring - curing", as well as parallel operation conditions and resource conflict constraints. Process logic information can be used to guide the assembly logic of process simulation components in the composite model layer.

[0115] Process logic information can refer to the macro-level construction process and coordination rules across multiple processes extracted from complete historical engineering cases, such as the full-process collaborative mechanism of "concrete pouring process in dams, including surface preparation, transportation scheduling, pouring and vibration, and temperature-controlled curing". Process logic information can support the top-level design and parametric configuration of process simulation components in the process template layer.

[0116] In this embodiment of the invention, basic construction operations can be parameterized and encapsulated to obtain a construction element simulation component with standardized input and output interfaces.

[0117] In practical applications, indivisible basic construction operations are extracted from historical construction information and abstracted into reusable parametric construction element simulation components. These basic construction operations include physical processes (such as mechanical movement), material transfer, chemical reactions (such as cement hydration), and geometric operations (such as excavation and shaping).

[0118] Secondly, standardized mathematical logic and parameter interfaces are established for each construction element simulation component. Taking the "crane hoisting" model as an example, it encapsulates the calculation rules for hoisting speed, slewing radius, load and stability. The input parameters include the hoisting weight, boom length, and start and end coordinates, and the output results are operation time, energy consumption index and safety status.

[0119] Model development can be implemented on a professional simulation platform or using object-oriented programming languages ​​such as C++ / Python, following a unified interface specification. Each construction element simulation component is equipped with structured input and output data, including functional descriptions, input and output parameter definitions, and applicable scope descriptions, to support model retrieval, matching, and invocation. This ultimately forms an atomic model layer with rich types, reliable logic, and flexible configuration capabilities.

[0120] In this embodiment of the invention, the assembly order of construction elements among at least one construction element simulation component can be determined according to the process logic information, and at least one construction element simulation component can be assembled according to the construction element assembly order to obtain a process simulation component.

[0121] The assembly sequence of construction elements refers to the execution sequence and dependencies formed between multiple construction element simulation components according to the construction process logic when constructing process simulation components. This sequence is based on the micro-logic within the process. For example, in the "concrete pouring process", the step sequence of "placement-leveling-vibration-finishing" must be strictly followed, and the triggering conditions, data transfer, and resource handover rules between steps must be defined.

[0122] In practical applications, multiple construction element simulation components can be systematically integrated according to construction procedure logic to form a complete procedure simulation component. First, the combination logic between the construction element simulation components is defined, including determining their temporal relationships, such as a strict sequence of drilling-charging-blasting; dependency conditions, such as not initiating slag removal before ventilation is completed; and triggering mechanisms, such as triggering safety monitoring after blasting vibration data reaches the standard. Second, a data flow channel is established between the construction element simulation components, allowing upstream model output parameters, such as drilling depth and blasting volume, to serve as input variables for downstream models. Finally, the integrated model is encapsulated into an interface, presenting it as a unified black-box module. Inputs are procedure-level macroscopic parameters, such as tunnel cross-sectional dimensions and surrounding rock grade, while outputs are procedure-level performance indicators, such as single-cycle time, material consumption, and safety state set, thereby achieving standardized calling and parameterized driving of the procedure simulation components.

[0123] In this embodiment of the invention, the assembly sequence of at least one process simulation component can be determined based on process logic information, and at least one process simulation component can be assembled according to the assembly sequence to obtain a process simulation component.

[0124] The assembly sequence of processes refers to the macroscopic organizational relationship formed between multiple combined models, i.e., multiple process simulation components, based on the complete construction process flow when constructing process simulation components. This sequence reflects the collaborative logic between processes. For example, in the "dam concrete construction process", the processes need to be organized according to the process chain of "foundation treatment - formwork installation - rebar binding - pouring - curing - quality inspection", and the resource scheduling, time and space constraints and risk control mechanisms across processes should be coordinated.

[0125] In practical applications, multiple process simulation components can be systematically integrated to form a complete process simulation component that can be driven by parameters, targeting typical process scenarios in hydropower projects.

[0126] First, a process flow diagram can be established based on the standardized construction plan. This flow diagram should cover the main processes, auxiliary processes, and emergency handling processes, and establish a mapping relationship between each node and the corresponding process simulation component.

[0127] Secondly, a parametric driving engine can be developed. This engine supports centralized adjustment of process-level global parameters, such as dam height, concrete strength grade, and TBM tunneling diameter, by modifying the top-level configuration file or graphical interface. The parametric driving engine has an automatic parameter distribution mechanism, which can decompose global parameters level by level and pass them to all related process simulation components and construction element simulation components at the bottom level, enabling collaborative updates of process simulation components.

[0128] Finally, the output of the process simulation component can be standardized and encapsulated, and the key performance indicator output set of the process simulation component can be predefined, such as the total project duration distribution curve, the component resource requirement matrix, the cost accumulation function, etc., to form a ready-to-use process simulation component with complete input and output specifications.

[0129] In this embodiment of the invention, an engineering simulation model library can be constructed based on construction element simulation components, process simulation components, and technology simulation components.

[0130] In practical applications, the construction element simulation components, process simulation components, and technology simulation components in the engineering simulation model library are all processed using parametric, reusable semantic encapsulation and intelligent assembly technologies, thus all having standardized interfaces.

[0131] This invention extracts basic construction operations, procedural logic information, and technological logic information from historical construction information and encapsulates them into construction element simulation components, procedural simulation components, and technological simulation components, ultimately constructing an engineering simulation model library. The simulation model library built based on historical construction information possesses high realism and reliability, accurately reflecting the actual construction process. The parametric encapsulation and standardized interface design of this invention achieve high reusability and flexible configuration capabilities for the simulation components. The assembly sequence design of the procedural and technological processes ensures that the simulation model can fully reflect the logical relationships of the construction process. The engineering simulation model library constructed by this invention provides rich component resources and efficient modeling tools for construction simulation, significantly improving the construction efficiency and accuracy of construction simulation models.

[0132] In some embodiments of the present invention, after using a target construction simulation model to perform simulation calculations on the adjustment strategy and obtaining simulation results, the process includes: Sub-step 51: If the multimodal data changes, construct the first query information according to the adjustment strategy of the change.

[0133] Sub-step 52 involves inputting the first query information and preset project adjustment prompts into the trained concrete pouring construction model for analysis, thereby obtaining the modified process information output by the concrete pouring construction model.

[0134] Sub-step 53: Adjust the target construction simulation model according to the changed process information to obtain the adjusted target construction simulation model.

[0135] The first query information refers to a dynamic query command, reconstructed by the master control agent based on the changed project description and oriented towards the construction process-enhanced language model, when changes in adjustment strategies and / or multimodal data are detected. The first query information can combine updated engineering conditions, modified process requirements, or newly added constraints with preset project adjustment prompts to form a structured request that can trigger the model to incrementally correct or reconstruct existing simulation schemes.

[0136] In this embodiment of the invention, the first query information and the preset project adjustment prompts can be input into the trained concrete pouring construction model for analysis to obtain the change process information output by the concrete pouring construction model.

[0137] The preset project adjustment prompts are basic instruction templates that indicate the need to modify the existing scheme when multimodal data changes. These prompts do not contain specific change analysis logic; they are only used to activate the model's scheme correction mode. A typical example is "Adjust the original construction process scheme according to the following changes." Based on this basic instruction, the model can autonomously identify the impact of changes and deduce process updates in conjunction with the specific changes.

[0138] Change process information refers to the structured process update description output by the language model based on the first query information and project adjustment prompts, which addresses project changes. Change process information can represent the process steps that need to be modified, the adjusted logical relationships, the updated resource requirements, and the new constraints.

[0139] In this embodiment of the invention, the target simulation model can be adjusted according to the changed process information to obtain the adjusted target simulation model.

[0140] The adjusted target simulation model can refer to the target simulation model generated by the construction simulation modeling agent after making partial modifications or reconstructions to the original target simulation model based on the changed process information. The adjustments include, but are not limited to: replacing or adding / deleting some simulation model components, modifying the connection relationships between components, and updating parameter configuration logic, ultimately forming a target simulation model that is consistent with the changed multimodal data and / or adjustment strategy.

[0141] In practical applications, when the construction environment changes, such as a shift in the pouring area location or an adjustment in the number of construction equipment due to design changes, users can input new adjustment commands through natural language descriptions or real-time changes in multimodal sensor data. The master control agent will re-parse the command and send model adjustment suggestions to the construction simulation modeling agent. Based on these suggestions, the modeling agent quickly reconstructs the assembled target simulation model. Specific operations may include adjusting the spatial location parameters of sub-models, updating the logical connections between components, and modifying resource allocation logic. Through this dynamic adjustment mechanism, the simulation model remains synchronized with the changed construction environment, thereby continuously providing accurate and real-time construction process simulation. For example, at a concrete pouring construction site, if a conflict is found between the requirement of "concrete mixing time < 30 minutes" and the specification of "concrete mixing time ≥ 30 minutes," the implementers may directly extend the real-time concrete mixing time. In this case, the multimodal sensor data will show real-time changes, and the real-time changes in the multimodal sensor data will input new adjustment commands. The master control agent will re-parse the command and send model adjustment suggestions to the construction simulation modeling agent. Based on this suggestion, the modeling agent can quickly reconstruct the assembled target simulation model, so as to keep the simulation model synchronized with the changed construction environment, thereby continuously providing accurate and real-time simulation of the construction process.

[0142] Furthermore, for feedback data regarding different types of non-compliance processes, the simulation model can be adjusted to reflect the non-compliance feedback data. This allows the simulation model to reflect the impact of non-compliance issues on the time nodes of each construction stage. For example, if the feedback data indicates a deviation in the pouring position, the material placement path and quantity in the material placement sub-process of the simulation model can be adjusted to simulate the concrete distribution under such deviations, and to determine the extension of subsequent finishing and curing time caused by uneven material placement. As another example, when "insufficient vibration" is diagnosed, the simulation model will integrate the current pouring progress, resource consumption, and compliance status to dynamically predict the expected completion time and resource requirements of the pouring operation after adding rework steps. It can also predict the duration of construction delays, changes in the critical path, and additional resource costs caused by delays due to quality issues, comprehensively assessing the impact of non-compliance processes on project costs.

[0143] This invention, by introducing a concrete pouring construction model, achieves dynamic adjustment and real-time updating of the simulation model after changes in multimodal data. By constructing a first query message and combining it with preset project adjustment prompts, this invention can accurately identify project changes and trigger model correction. Based on the changed process information, the concrete pouring construction model automatically generates a structured process update description, ensuring the accuracy and efficiency of the adjustment. Ultimately, the adjusted target simulation model remains consistent with the changed multimodal data and / or adjustment strategies, providing dynamic adaptability for construction simulation, significantly improving the flexibility and practicality of the simulation model, and providing strong support for the dynamic management of complex engineering projects.

[0144] In some embodiments of the present invention, the step of using a target construction simulation model to perform simulation calculations on the adjustment strategy and obtain simulation results includes: Sub-step 61: Instantiate the preset construction simulation model using the simulation parameter information, and run the preset construction simulation model several times in the simulation environment to obtain several candidate simulation construction results; In this embodiment of the invention, simulation parameter information is used to inject parameters into each component of a pre-defined construction simulation model structure, thereby instantiating the target simulation model. The target simulation model is then run several times in a simulation environment to obtain several candidate simulation construction results.

[0145] The candidate simulation construction results can refer to the set of project performance index data obtained through multiple independent simulation runs. Each result corresponds to the simulation output under a specific set of parameter configurations, which may include indicators such as construction period, construction cost, resource utilization rate, and construction intensity.

[0146] In practical applications, the construction simulation calculation scheduling agent monitors the simulation status through an event-driven mechanism. When it senses that the model instantiation is complete, it automatically starts the simulation execution process. First, based on the confidence interval calculation method in statistical principles and combined with the preset simulation accuracy requirements, the minimum number of repetitions for the Monte Carlo simulation is dynamically determined. Furthermore, experimental design methods such as Latin hypercube sampling can be selectively used to optimize the exploration efficiency of the parameter space.

[0147] In terms of resource allocation, the construction simulation computing scheduling agent detects the availability of local computing clusters and cloud elastic resources in real time, decomposes large-scale simulation tasks into independent computing units that can be executed in parallel, and establishes a task distribution queue. During task execution, the system continuously monitors the running status, resource utilization, and task progress of each computing node through a heartbeat detection mechanism, and dynamically adjusts the task allocation strategy based on a load balancing algorithm. For task units that fail, the system automatically records the failure status and triggers a retry mechanism, reallocating computing resources when necessary.

[0148] Meanwhile, the intelligent agent for construction simulation calculation scheduling integrates a cloud platform resource management interface, which can dynamically adjust the scale of resource configuration according to the real-time computing load, automatically expand cloud server instances during peak simulation computing periods, and release idle resources in a timely manner during off-peak computing periods, thereby achieving optimal control of computing costs.

[0149] All output data from completed simulation tasks, including time-series data, resource consumption records, and cost distribution information, are automatically collected and stored in a time-series database according to a standardized format, forming a complete simulation results dataset. This dataset not only contains the original simulation output but also task execution metadata, providing a structured data foundation for subsequent optimization analysis and decision support.

[0150] Sub-step 62: Determine the target simulation result from several candidate simulation construction results based on the multimodal information, and display the target simulation construction result as the simulation analysis result to the user.

[0151] In this embodiment of the invention, a target simulation construction result can be selected from several candidate simulation construction results based on the multimodal information, and the target simulation construction result can be displayed to the user as a simulation analysis result. The target simulation construction result can be one or more.

[0152] The target simulation construction result refers to the simulation output selected from the candidate result set based on preset optimization criteria, such as shortest construction period, optimal cost, and most balanced resources. The target simulation construction result not only includes the optimal performance index value, but also retains the corresponding complete construction process, which can support scheme traceability and decision verification.

[0153] In some embodiments of the present invention, the step of acquiring multimodal sensing data from a concrete pouring construction site and determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of the concrete pouring construction includes: Based on the concrete pouring construction model, and according to the multimodal sensing data, the first entity corresponding to the multimodal sensing data is determined from the structured knowledge base of the concrete pouring construction.

[0154] Before determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction based on the concrete pouring construction model and the multimodal sensing data, the method further includes: Sub-step 71: Using a pre-set large model, training sample data is generated based on the structured knowledge base of concrete pouring construction.

[0155] In some examples, the training sample data may be paired training sample data that characterizes the input instructions to the model and the outputs that characterize the corresponding responses to the input instructions.

[0156] Based on accurate facts in the structured knowledge base of concrete pouring construction, various forms of instructions or questions are generated in batches through a pre-set large language model, thereby combining a large amount of high-quality instruction-output pairing data (training sample data), where the "output" part is strictly derived from the structured knowledge base of concrete pouring construction.

[0157] Sub-step 72: Train the pre-set concrete pouring construction model based on the training sample data.

[0158] Based on the output paired data (training sample data), the pre-set concrete pouring construction model is optimized and trained.

[0159] Based on the generated (instruction, output) paired dataset, a suitable base model (such as Qwen2-7B or InternLM2-7B) is selected, and the general large language model is optimized by combining transfer learning with fine-tuning methods such as QLoRA, so that it can better adapt to the language characteristics and professional knowledge in the field of concrete pouring construction.

[0160] After obtaining the trained target concrete pouring construction model, the real-time construction situation or user-input natural language can be analyzed through the target concrete pouring construction model to obtain the necessary simulation actions and implicit requirements.

[0161] In practical applications, the user's natural language description is input into the target concrete pouring construction model, which has been fine-tuned using a concrete construction knowledge graph (a structured knowledge base for concrete pouring construction). Entity recognition, intent deconstruction, and implicit information reasoning are then performed to deconstruct the implicit simulation sequence. Based on a preset output template, all the parsed information is organized into a structured simulation model call content in JSON format. This includes a list of involved simulation models (e.g., process sequence, resource types, and technical requirements), logical relationships between models (timing, dependencies), triggering conditions, and all necessary input parameters.

[0162] For example, if a user inputs "Simulate the concrete pouring of an arch dam during the dry season, the dam is 80 meters high, two cable cranes are used, and temperature control needs to be considered," the system accurately identifies key elements such as "arch dam," "dry season," "concrete pouring," "cable cranes (2 units)," and "temperature control." Understanding the user's deeper simulation intent, "simulate concrete pouring" is deconstructed into a series of necessary simulation actions, such as "simulate concrete production," "simulate transportation," "simulate surface pouring," and "simulate temperature control process." Implicit requirements are inferred; for example, "dry season" might mean higher construction intensity and less weather interference; "temperature control" means the need to implement relevant cooling or insulation measures. These implicit requirements are then used as other parameters to further calculate the simulation results when performing simulation calculations on the construction site using the simulation model.

[0163] In some embodiments of the present invention, the entity type includes any one or more of the following: construction action, construction equipment entity, construction method, construction material, technical indicators, construction environmental conditions, construction quality defects, construction quality specifications, and construction safety risks.

[0164] In this embodiment of the invention, knowledge documents in the field of concrete pouring construction are acquired; the knowledge documents are identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between these entities; and a structured knowledge base for concrete pouring construction is constructed based on these entities and their semantic relationships; multimodal sensor data from the concrete pouring construction site is acquired; the multimodal sensor data is diagnosed according to the structured knowledge base for concrete pouring construction to obtain diagnostic results; an adjustment strategy for the concrete pouring construction site is generated based on the diagnostic results; a target construction simulation model is used to simulate and calculate the adjustment strategy to obtain simulation results, and the simulation results are displayed to the user as simulation analysis results. This achieves the acquisition of multimodal data from the construction site and the dynamic simulation of the multimodal data and its adjustment strategy, which can accurately reflect various changes in the actual construction process, improve the efficiency of simulation calculation and the efficiency of converting simulation results.

[0165] Reference Figure 2 The diagram shows a structural schematic of a concrete pouring construction data processing device according to an embodiment of the present invention, which may specifically include the following modules: Engineering knowledge acquisition module 201 is used to acquire knowledge documents in the field of concrete pouring construction. The knowledge base construction module 202 is used to identify the knowledge file, obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, and construct a structured knowledge base for concrete pouring construction based on the multiple entities and the semantic relationships between the multiple entities. The multimodal data acquisition module 203 is used to acquire multimodal sensing data at the concrete pouring construction site. Diagnostic module 204 is used to diagnose the multimodal sensing data based on the structured knowledge base of concrete pouring construction and obtain diagnostic results. The adjustment strategy module 205 is used to generate an adjustment strategy for the concrete pouring construction site based on the diagnostic results. The simulation calculation module 206 is used to perform simulation calculations on the adjustment strategy using the target construction simulation model, obtain simulation results, and display the simulation results to the user as simulation analysis results.

[0166] In some embodiments of the present invention, after using a target construction simulation model to perform simulation calculations on the adjustment strategy and obtaining simulation results, the process includes: The query information construction module is used to construct query information based on the changed adjustment strategy if the adjustment strategy changes. The process change module is used to input the query information and preset project adjustment prompts into a trained concrete pouring construction model for analysis, and obtain the process change information output by the concrete pouring construction model. The target construction simulation model determination module is used to adjust the target construction simulation model according to the changed process information to obtain the adjusted target construction simulation model.

[0167] In some embodiments of the present invention, the simulation calculation module 206 includes: The simulation construction results submodule is used to instantiate the target construction simulation model using the simulation parameter information, and to run the target construction simulation model several times in the simulation environment to obtain several candidate simulation construction results; The target simulation result determination submodule is used to determine the target simulation result from several candidate simulation construction results based on the multimodal information.

[0168] In some embodiments of the present invention, the knowledge base construction module 202 includes: A defined data acquisition submodule is used to acquire user-defined data; wherein, the user-defined data includes multiple entity types related to concrete pouring construction; The semantic relationship determination submodule is used to identify the knowledge file based on the user-defined data to obtain multiple entities and the semantic relationships between the multiple entities.

[0169] In some embodiments of the present invention, the knowledge base construction module 202 further includes: The first entity determination submodule is used to determine the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction. The second entity determination submodule is used to determine a second entity related to the first entity based on the semantic relationship of the first entity. The diagnostic result acquisition module is used to diagnose the multimodal sensing data based on the knowledge content corresponding to the second entity and obtain a diagnostic result.

[0170] In some embodiments of the present invention, the first entity determining submodule includes: The first entity determination unit is used to determine the first entity corresponding to the multimodal sensing data from the structured knowledge base of the concrete pouring construction based on the concrete pouring construction model. Before determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction based on the concrete pouring construction model and the multimodal sensing data, the method further includes: A training sample unit is generated to generate training sample data based on the structured knowledge base of concrete pouring construction using a pre-set large model. The model training unit is used to train a pre-set concrete pouring construction model based on the training sample data.

[0171] In some embodiments of the present invention, the apparatus further includes: The visual information and work behavior acquisition module is used to acquire visual information of the concrete pouring construction site and work behavior information of the personnel through construction visual sensors.

[0172] The vibration parameter acquisition module is used to acquire physical parameters during the concrete vibration process through vibration sensors. The vibration quality determination module is used to determine the concrete vibration quality based on the physical parameters. The pouring parameter acquisition module is used to acquire concrete pouring parameters through a distance sensor; The multimodal sensing data construction module is used to construct multimodal sensing data from the visual information, the operational behavior information, the physical parameters, the concrete vibration quality, and the pouring parameters.

[0173] In some embodiments of the present invention, the entity type includes any one or more of the following: construction action, construction equipment entity, construction method, construction material, technical indicators, construction environmental conditions, construction quality defects, construction quality specifications, and construction safety risks.

[0174] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0175] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0176] Some embodiments of the present invention also provide an electronic device, including a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the method described above.

[0177] Some embodiments of the present invention also provide a computer-readable storage medium on which a computer program is stored, and which, when executed by a processor, implements the method described above.

[0178] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this invention are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0179] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0180] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0181] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0182] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0184] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0185] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the aforementioned element.

[0186] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

[0187] The above provides a detailed description of the method, apparatus, electronic equipment, and program product for processing concrete pouring construction data. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. The content of this specification should not be construed as limiting the present invention.

Claims

1. A method for processing concrete pouring construction data, characterized in that, The method includes: Obtain knowledge documents related to concrete pouring construction; The knowledge file is identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between these entities. Based on these entities and their semantic relationships, a structured knowledge base for concrete pouring construction is constructed. Acquire multimodal sensor data at the concrete pouring construction site; Based on the structured knowledge base of concrete pouring construction, the multimodal sensing data is diagnosed to obtain diagnostic results; Based on the diagnostic results, an adjustment strategy is generated for the concrete pouring construction site. The adjustment strategy is simulated and calculated using a target construction simulation model to obtain simulation results, which are then presented to the user as simulation analysis results.

2. The method according to claim 1, characterized in that, After using the target construction simulation model to perform simulation calculations on the adjustment strategy and obtaining the simulation results, the following steps are included: If the multimodal sensing data changes, then a first query information is constructed based on the changed multimodal sensing data; The first query information and the preset project adjustment prompts are input into the trained concrete pouring construction model for analysis to obtain the change process information output by the concrete pouring construction model. The target construction simulation model is adjusted based on the changed process information to obtain the adjusted target construction simulation model.

3. The method according to claim 1, characterized in that, The target construction simulation model is used to perform simulation calculations on the adjustment strategy, and the simulation results are obtained, including: The target construction simulation model is instantiated using simulation parameter information, and the target construction simulation model is run several times in the simulation environment to obtain several candidate simulation construction results; The target simulation result is determined from several candidate simulation construction results based on the multimodal information.

4. The method according to claim 1, characterized in that, The knowledge document is identified to obtain multiple entities related to concrete pouring construction and the semantic relationships between these entities, including: Obtain user-defined data; wherein, the user-defined data includes multiple entity types related to concrete pouring construction; Based on the user-defined data, the knowledge file is identified to obtain multiple entities and the semantic relationships between the multiple entities.

5. The method according to claim 1, characterized in that, The diagnostic results obtained by diagnosing the multimodal sensor data based on the structured knowledge base of concrete pouring construction include: The first entity corresponding to the multimodal sensing data is determined from the structured knowledge base of concrete pouring construction. Based on the semantic relationship of the first entity, determine the second entity related to the first entity; Based on the knowledge content corresponding to the second entity, the multimodal sensing data is diagnosed to obtain a diagnostic result.

6. The method according to claim 5, characterized in that, Determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction includes: Based on the concrete pouring construction model, the first entity corresponding to the multimodal sensing data is determined from the structured knowledge base of the concrete pouring construction. Before determining the first entity corresponding to the multimodal sensing data from the structured knowledge base of concrete pouring construction based on the concrete pouring construction model and the multimodal sensing data, the method further includes: Training sample data is generated using a pre-built large model and the structured knowledge base of concrete pouring construction. The pre-set concrete pouring construction model is trained based on the training sample data.

7. The method according to claim 1, characterized in that, Acquire multimodal sensor data from the concrete pouring construction site, including: Visual information about the concrete pouring site and the operational behavior of the personnel are obtained through construction vision sensors. Vibration sensors are used to acquire physical parameters during the concrete vibration process. The concrete vibration quality is determined based on the visual information and the physical parameters. The concrete pouring parameters are obtained through a distance measuring sensor; The visual information, the operational behavior information, the physical parameters, the concrete vibration quality, and the pouring parameters are constructed into multimodal sensing data.

8. The method according to claim 4, characterized in that, The entity types include any one or more of the following: construction actions, construction equipment entities, construction methods, construction materials, technical indicators, construction environmental conditions, construction quality defects, construction quality specifications, and construction safety risks.

9. A device for processing concrete pouring construction data, characterized in that, The device includes: The engineering knowledge acquisition module is used to acquire knowledge documents in the field of concrete pouring construction. The knowledge base construction module is used to identify the knowledge file, obtain multiple entities related to concrete pouring construction and the semantic relationships between the multiple entities, and construct a structured knowledge base for concrete pouring construction based on the multiple entities and the semantic relationships between the multiple entities. The multimodal data acquisition module is used to acquire multimodal sensor data at the concrete pouring construction site. The diagnostic module is used to diagnose the multimodal sensing data based on the structured knowledge base of concrete pouring construction and obtain diagnostic results. The adjustment strategy module is used to generate an adjustment strategy for the concrete pouring construction site based on the diagnostic results. The simulation calculation module is used to perform simulation calculations on the adjustment strategy using the target construction simulation model, obtain simulation results, and display the simulation results to the user as simulation analysis results.

10. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the method as described in any one of claims 1-8.

12. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 8.

Images