Construction process digital twin online simulation method and device

Abstract

This application relates to the field of digital twin technology, and in particular to a method and apparatus for online simulation of construction process digital twins. The method includes: constructing a structured knowledge base encompassing safety risks, quality hazards, and dynamic interaction mechanisms of work processes, providing a core knowledge foundation for generating scheduling responses; designing a reinforcement learning network specifically for dynamic scheduling of construction processes, addressing the unique characteristics of construction process scheduling, to achieve intelligent, near real-time construction scheduling decisions; and researching adaptive model training methods to address the difficulty of reinforcement learning models in effectively learning and responding to sudden construction disturbances, enabling the model to learn how to dynamically adjust the process network when disturbances occur, minimizing the impact of process adjustments while ensuring the project's construction objectives. This solves the problem that existing technologies primarily consider dynamic changes in construction sites such as resource fluctuations and weather changes, failing to address responses to events such as safety issues or quality hazards at construction sites.

Description

Technical Field

[0001] This application relates to the field of digital twin technology, and in particular to a method and apparatus for online simulation of a construction process using a digital twin. Background Technology

[0002] Traditional construction scheduling methods typically build optimization models based on static constraints and generate initial scheduling schemes through offline calculations. While these methods can optimize resource allocation and schedule management to some extent, they often require recalculation of the scheduling scheme when encountering unforeseen problems, causing decision-making delays and impacting the safety and efficiency of the construction process. Furthermore, scheduling methods based on heuristic algorithms and mathematical programming still face problems such as premature convergence and local optima in optimization problems, limiting their effectiveness in complex construction scenarios. Therefore, a new scheduling method is urgently needed that can adapt to the dynamically changing construction environment.

[0003] Digital twin technology has received widespread attention in recent years and is considered an effective means to address the challenges of dynamic scheduling at construction sites. Digital twins achieve a virtual-real mapping of physical systems through high-fidelity virtual models, enabling real-time synchronization of the physical system's state and reflecting its operating conditions. This allows construction sites to acquire data in real time for comprehensive monitoring and optimization of the construction process. Digital twin technology significantly enhances the ability to collect real-time status data on personnel, machinery, materials, and the environment at construction sites, more accurately reflecting dynamic changes and providing strong support for construction scheduling. Despite its high-precision virtual simulation capabilities, digital twins still face a series of challenges in practical applications, including massive data volumes and complex elements.

[0004] Reinforcement Learning (RL) provides an effective decision-making framework for addressing the challenges of dynamic scheduling in construction sites. As an adaptive dynamic decision-making method, RL continuously adjusts its decision-making strategies through ongoing interaction with the environment, enabling it to adapt to rapidly changing scenarios. However, traditional RL, relying on manually designed state features, struggles to handle high-dimensional action spaces (such as parallel adjustment of work processes and multi-objective allocation of resources), while the spatiotemporal coupling between work processes in construction scheduling scenarios can lead to an explosion in action dimensions. Deep Reinforcement Learning (DRL) technology automatically extracts high-dimensional state features through deep neural networks and uses algorithms such as Q-learning and policy gradients to achieve dimensionality reduction mapping of high-dimensional action spaces. Its adaptive advantages have been validated in dynamic decision-making problems. Deep reinforcement learning can continuously interact with the environment and optimize its strategies through trial and error, which is highly effective for dynamic scheduling. However, the application of deep reinforcement learning in construction is still in its developmental stage.

[0005] Currently, existing technologies can be based on adaptive scheduling methods for construction cash flow using deep reinforcement learning. The main goal is to optimize the balance between schedule, resources, and cost, and to introduce a cash flow perspective to assist in the adaptive optimization of scheduling strategies. Alternatively, a hybrid approach combining reinforcement learning and agent-based modeling (ABM) can be used, combining graph embedding networks to optimize the scheduling of work sequence and resource constraints in construction projects, which can simulate the dynamic evolution process in complex construction environments. In addition, existing technologies can also combine valid action sampling (VAS) and DRL scheduling methods of graph convolutional networks to accelerate training convergence through reward shaping, demonstrating good scheduling and rescheduling capabilities in large-scale construction projects.

[0006] Therefore, existing construction sequence scheduling technologies have the following drawbacks:

[0007] 1. The analysis of the coupling effects of multiple factors is not comprehensive:

[0008] Existing technologies do not provide a detailed discussion of how factors such as safety, quality, and procedures interact and constrain each other during the construction process.

[0009] 2. The potential of digital twin technology in process scheduling has not yet been realized:

[0010] The existing scheduling process is not integrated with the actual construction process, and for example, it cannot respond quickly when a sudden event occurs on site.

[0011] 3. The method for dynamic scheduling of work processes is still immature:

[0012] There is currently no method for dynamically scheduling work processes to address potential quality hazards and safety issues during construction.

[0013] In summary, existing technologies mainly consider dynamic changes at construction sites, such as resource fluctuations and weather changes, but they do not address safety issues or potential quality hazards at construction sites, which urgently need to be resolved. Summary of the Invention

[0014] This application provides a digital twin online simulation method and apparatus for construction processes, which addresses the problem that existing technologies mainly consider dynamic changes in construction sites such as resource changes and weather changes, and do not respond to events such as safety issues or quality hazards at construction sites.

[0015] The first aspect of this application provides a method for online digital twin simulation of a construction process, applied in the offline training phase, comprising the following steps: constructing a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, a pre-set safety-process coupling knowledge base, and a quality-process coupling knowledge base, and establishing a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base; modeling the coupling relationship knowledge graph, determining the corresponding state space, action space, state transition mechanism, and disturbance mechanism, and constructing a multi-objective reward function corresponding to the original construction schedule plan, so as to construct a process scheduling optimization model based on the state space, the action space, the state transition mechanism, the multi-objective reward function, and the disturbance mechanism; determining a hierarchical deep reinforcement learning network model through the process scheduling optimization model, constructing a construction simulation environment using the hierarchical deep reinforcement learning network model, and training the hierarchical deep reinforcement learning network model based on the construction simulation environment, so as to perform online digital twin simulation of the original construction schedule plan based on the trained hierarchical deep reinforcement learning network model in the online simulation phase.

[0016] Optionally, in one embodiment of this application, the step of constructing a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, a preset safety-process coupling knowledge base and a quality-process coupling knowledge base, and establishing a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base includes: acquiring multiple safety risks and multiple safety accidents during the target construction process, determining the risk level and risk category of each of the multiple safety risks, and determining the corresponding process scheduling impact type based on the risk level and the risk category; establishing an accident type mapping relationship between the risk category and the safety accident type, a level mapping relationship between each safety risk and the risk level, a risk category mapping relationship between each safety risk and the risk category, and a process mapping relationship between each safety risk and the process scheduling impact type based on the multiple safety accidents, so as to determine the corresponding process scheduling impact type based on the accident type mapping relationship, the level mapping relationship, and the risk category mapping relationship. The system constructs a safety-process coupling knowledge sub-base based on the mapping relationship between the system and the process; it acquires various quality hazards during the target construction process and quantifies the severity of each quality hazard to obtain a corresponding severity level; based on the severity level, it establishes various process interaction influence modes and determines the state index corresponding to each quality hazard; it establishes mapping relationships between the state index and the quality hazard, the severity level, and the process interaction influence mode, and constructs the quality-process coupling knowledge sub-base based on the mapping relationships; based on the safety-process coupling knowledge sub-base and the quality-process coupling knowledge sub-base, it constructs a safety-quality-process coupling evolution knowledge base, and through a preset graph database strategy, it establishes safety-process knowledge graphs and quality-process knowledge graphs corresponding to the safety-process coupling knowledge sub-base and the quality-process coupling knowledge sub-base, and constructs corresponding visualized coupling relationship knowledge graphs based on the safety-process knowledge graphs and the quality-process knowledge graphs.

[0017] Optionally, in one embodiment of this application, the step of modeling the coupling relationship knowledge graph, determining the corresponding state space, action space, state transition mechanism, and disturbance mechanism, and constructing a multi-objective reward function corresponding to the original construction schedule plan, to construct a process scheduling optimization model based on the state space, the action space, the state transition mechanism, the multi-objective reward function, and the disturbance mechanism, includes: calculating the global environment state and process feature state in the construction process corresponding to the original construction schedule plan, and determining the state space based on a preset two-layer state representation strategy, combined with the global environment state and the process feature state; constructing the action space based on a preset binary action representation strategy, and determining whether all predecessor processes of the current process have been completed according to the construction progress of the original construction schedule plan, wherein, if all predecessor processes have been completed, the state of the current process is updated to determine the state transition mechanism; obtaining disturbance data in the construction process corresponding to the original construction schedule plan, and generating corresponding processing measures according to the disturbance data and a preset response strategy to determine the disturbance mechanism.

[0018] Optionally, in one embodiment of this application, the step of determining a hierarchical deep reinforcement learning network model through the process scheduling optimization model, constructing a construction simulation environment using the hierarchical deep reinforcement learning network model, and training the hierarchical deep reinforcement learning network model based on the construction simulation environment includes: constructing the hierarchical deep reinforcement learning network model based on a pre-constructed input layer, encoding layer, process attention layer, feature fusion layer, two-stream architecture layer, and output layer; extracting component geometric information and spatial relationships from a preset building information model, and obtaining construction activity data corresponding to the component geometric information and spatial relationships based on a preset four-dimensional work decomposition structure; determining multiple process constraint rules based on the construction activity data, constructing a perturbation simulation generation algorithm according to the multiple process constraint rules and the coupling relationship knowledge graph, and constructing the construction simulation environment through the perturbation simulation generation algorithm, so as to train the hierarchical deep reinforcement learning network model in the construction simulation environment through preset conventional construction paths and perturbation processing paths.

[0019] Optionally, in one embodiment of this application, the mathematical expression of the multi-objective reward function is:

[0020]

[0021] in, Indicates a reward for improved status; Indicates a reward for parallel processing steps; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

[0022] A second aspect of this application provides a digital twin online simulation method for construction processes, applied in the online simulation stage, comprising the following steps: monitoring the current construction site status corresponding to the target construction site, and obtaining corresponding disturbance information based on the current construction site status; inputting the disturbance information into a pre-constructed coupling relationship knowledge graph to output corresponding process scheduling schemes and process adjustment suggestions; converting the process scheduling schemes and process adjustment suggestions into corresponding target format data, and inputting the target format data into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

[0023] The third aspect of this application provides a method for online digital twin simulation of a construction process, applied in the offline training phase, comprising the following steps: a graph construction module, used to construct a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, a preset safety-process coupling knowledge base and a quality-process coupling knowledge base, and to establish a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base; a modeling module, used to model the coupling relationship knowledge graph, determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and construct a multi-objective reward function corresponding to the original construction schedule plan, so as to construct a process scheduling optimization model based on the state space, the action space, the state transition mechanism, the multi-objective reward function and the disturbance mechanism; and a training module, used to determine a hierarchical deep reinforcement learning network model through the process scheduling optimization model, and to construct a construction simulation environment using the hierarchical deep reinforcement learning network model, and to train the hierarchical deep reinforcement learning network model based on the construction simulation environment, so as to perform online digital twin simulation of the original construction schedule plan based on the trained hierarchical deep reinforcement learning network model in the online simulation phase.

[0024] Optionally, in one embodiment of this application, the map construction module includes: an acquisition unit, configured to acquire multiple safety risks and multiple safety accidents during the target construction process, determine the risk level and risk category of each of the multiple safety risks, and determine the corresponding process scheduling impact type based on the risk level and the risk category; a first mapping unit, configured to establish, based on the multiple safety accidents, an accident type mapping relationship between the risk category and the safety accident type, a level mapping relationship between each safety risk and the risk level, a risk category mapping relationship between each safety risk and the risk category, and a process mapping relationship between each safety risk and the process scheduling impact type, so as to construct the safety-process coupled knowledge sub-base according to the accident type mapping relationship, the level mapping relationship, the risk category mapping relationship, and the process mapping relationship; and a determination unit, configured to acquire multiple safety risks and multiple safety accidents during the target construction process. The system identifies several quality hazards and quantifies the severity of each hazard to obtain a corresponding severity level. Based on the severity level, it establishes multiple process interaction influence modes and determines the corresponding state index for each quality hazard. A second mapping unit is used to establish mapping relationships between the state index and the quality hazard, the severity level, and the process interaction influence mode, respectively, to construct the quality-process coupling knowledge sub-base based on the mapping relationships. A building unit is used to construct the safety-quality-process coupling evolution knowledge base based on the safety-process coupling knowledge sub-base and the quality-process coupling knowledge sub-base, and to establish safety-process knowledge graphs and quality-process knowledge graphs corresponding to the safety-process coupling knowledge sub-base and the quality-process coupling knowledge sub-base, respectively, using a preset graph database strategy, to construct corresponding visualized coupling relationship knowledge graphs based on the safety-process knowledge graphs and the quality-process knowledge graphs.

[0025] Optionally, in one embodiment of this application, the modeling module includes: a calculation unit, configured to calculate the global environment state and process characteristic state during the construction process corresponding to the original construction schedule, and determine the state space by combining the global environment state and the process characteristic state based on a preset two-layer state representation strategy; a judgment unit, configured to construct the action space based on a preset binary action representation strategy, and determine whether all predecessor processes of the current process have been completed according to the construction progress of the original construction schedule, wherein, if all predecessor processes are completed, the state of the current process is updated to determine the state transition mechanism; and a generation unit, configured to acquire disturbance data during the construction process corresponding to the original construction schedule, and generate corresponding processing measures according to the disturbance data and a preset response strategy to determine the disturbance mechanism.

[0026] Optionally, in one embodiment of this application, the training module includes: a first construction unit, used to construct the hierarchical deep reinforcement learning network model based on a pre-constructed input layer, encoding layer, process attention layer, feature fusion layer, two-stream architecture layer, and output layer; an extraction unit, used to extract component geometric information and spatial relationships from a preset building information model, and obtain construction activity data corresponding to the component geometric information and spatial relationships based on a preset four-dimensional work decomposition structure; and a second construction unit, used to determine multiple process constraint rules based on the construction activity data, to construct a perturbation simulation generation algorithm according to the multiple process constraint rules and the coupling relationship knowledge graph, and to construct the construction simulation environment through the perturbation simulation generation algorithm, so as to train the hierarchical deep reinforcement learning network model in the construction simulation environment through preset conventional construction paths and perturbation processing paths.

[0027] Optionally, in one embodiment of this application, the mathematical expression of the multi-objective reward function is:

[0028]

[0029] in, Indicates a reward for improved status; Indicates a reward for parallel processing steps; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

[0030] The fourth aspect of this application provides a digital twin online simulation method for construction processes, applied in the online simulation stage, comprising the following steps: a monitoring module for monitoring the current construction site status corresponding to the target construction site and obtaining corresponding disturbance information based on the current construction site status; an input module for inputting the disturbance information into a pre-constructed coupling relationship knowledge graph to output corresponding process scheduling schemes and process adjustment suggestions; and a simulation module for converting the process scheduling schemes and process adjustment suggestions into corresponding target format data and inputting the target format data into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

[0031] A fifth aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the online digital twin simulation method for construction processes as described in the above embodiments.

[0032] A sixth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described online digital twin simulation method for construction processes.

[0033] Therefore, the embodiments of this application have the following beneficial effects:

[0034] The embodiments of this application can construct a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, a pre-set safety-process coupling knowledge base and a quality-process coupling knowledge base, and establish a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base; model the coupling relationship knowledge graph, determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and construct a multi-objective reward function corresponding to the original construction schedule plan, so as to construct a process scheduling optimization model based on the state space, action space, state transition mechanism, multi-objective reward function and disturbance mechanism; through the process scheduling optimization model, determine a hierarchical deep reinforcement learning network model, and use the hierarchical deep reinforcement learning network model to construct a construction simulation environment, and train the hierarchical deep reinforcement learning network model based on the construction simulation environment, so as to perform digital twin online simulation of the original construction schedule plan based on the trained hierarchical deep reinforcement learning network model in the online simulation stage. This application enables minimal modification and adjustment of the process network in the event of sudden external influences or disturbances, ensuring that cost, schedule, and other objectives remain as close to the original plan as possible, while minimizing the impact of process adjustments on personnel organization, material procurement, and other aspects. Furthermore, this application boasts advantages such as fast online simulation algorithm spacing calculation speed and minimal scope and impact of process adjustments. Therefore, it addresses the problem that existing technologies primarily consider dynamic changes in construction sites, such as resource fluctuations and weather changes, and fail to respond to events such as safety issues or quality hazards at the construction site.

[0035] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0036] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0037] Figure 1 This is a flowchart of a digital twin online simulation method for construction processes applied in the offline training phase, according to an embodiment of this application.

[0038] Figure 2 A schematic diagram of a process state transition mechanism provided for one embodiment of this application;

[0039] Figure 3 A schematic diagram of a disturbance modeling mechanism provided for one embodiment of this application;

[0040] Figure 4 A schematic diagram of the overall execution framework for online digital twin simulation of a construction process is provided as an embodiment of this application;

[0041] Figure 5 A schematic diagram of building information model component information processing is provided as an embodiment of this application;

[0042] Figure 6 A schematic diagram of a model training process is provided for one embodiment of this application;

[0043] Figure 7 This is a flowchart illustrating a digital twin online simulation method for construction processes applied in the online simulation phase, according to an embodiment of this application.

[0044] Figure 8 A schematic diagram of an online simulation process is provided as an embodiment of this application;

[0045] Figure 9 An online simulation diagram of a reinforced concrete residential building complex project in a specific embodiment of this application is provided;

[0046] Figure 9 Image (a) is a three-dimensional view of an example project provided in one embodiment of this application;

[0047] Figure 9 (b) is a schematic diagram of XY axis grid division for an example project provided in an embodiment of this application;

[0048] Figure 10 An example (1) of scheduling change Gantt chart provided for one embodiment of this application;

[0049] Figure 11 A case (1) cumulative progress curve comparison diagram provided for one embodiment of this application;

[0050] Figure 12 A case (1) showing a comparison of demand curves for different types of workers provided as an embodiment of this application;

[0051] Figure 13 A case (1) showing a comparison of demand curves for various types of materials provided in one embodiment of this application;

[0052] Figure 14 An example (2) of scheduling change Gantt chart is provided for one embodiment of this application;

[0053] Figure 15 A case (2) cumulative progress curve comparison diagram provided for one embodiment of this application;

[0054] Figure 16 A case (2) showing a comparison of demand curves for different types of workers, provided as an embodiment of this application;

[0055] Figure 17 A case (2) comparison diagram of demand curves for various types of materials provided in one embodiment of this application;

[0056] Figure 18 An example (3) of scheduling change Gantt chart is provided for one embodiment of this application;

[0057] Figure 19 A case (3) cumulative progress curve comparison diagram provided for one embodiment of this application;

[0058] Figure 20 A case (3) comparison diagram of demand curves for various types of workers provided as an embodiment of this application;

[0059] Figure 21 A case (3) comparison diagram of demand curves for various types of materials provided in one embodiment of this application;

[0060] Figure 22 This is an example diagram of a construction process digital twin online simulation device applied in the offline training phase, according to an embodiment of this application;

[0061] Figure 23 This is an example diagram of a digital twin online simulation device for construction processes applied in the online simulation phase, according to an embodiment of this application;

[0062] Figure 24 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0063] Among them, 10-the construction process digital twin online simulation device applied to the offline training stage, 20-the construction process digital twin online simulation device applied to the online simulation stage; 101-map construction module, 102-modeling module, 103-training module; 201-monitoring module, 202-input module, 203-simulation module; 2401-memory, 2402-processor, 2403-communication interface. Detailed Implementation

[0064] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0065] The following describes the online digital twin simulation method and apparatus for construction process based on embodiments of this application with reference to the accompanying drawings. Addressing the problems mentioned in the background art, this application provides an online digital twin simulation method for construction process. In this method, a safety-quality-process coupling evolution knowledge base is constructed based on the original construction schedule plan, a pre-set safety-process coupling knowledge base, and a quality-process coupling knowledge base. A coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is also established. The coupling relationship knowledge graph is modeled to determine the corresponding state space, action space, state transition mechanism, and disturbance mechanism. A multi-objective reward function corresponding to the original construction schedule plan is constructed. Based on the state space, action space, state transition mechanism, multi-objective reward function, and disturbance mechanism, a process scheduling optimization model is constructed. Through the process scheduling optimization model, a hierarchical deep reinforcement learning network model is determined. A construction simulation environment is constructed using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained. During the online simulation phase, the original construction schedule plan is simulated online using a digital twin. This application enables the modeling of the multi-element coupled evolution mechanism of construction safety, quality, and processes, constructs a process scheduling optimization model based on Markov chains, and ultimately achieves online digital twin simulation, thereby assisting in construction decision-making. This addresses the problem that existing technologies primarily consider dynamic changes in construction sites such as resource fluctuations and weather changes, failing to respond to events such as safety issues or quality hazards at the construction site.

[0066] Specifically, Figure 1 This is a flowchart illustrating an online digital twin simulation method for construction processes applied during the offline training phase, as provided in an embodiment of this application.

[0067] like Figure 1 As shown, the online digital twin simulation method for the construction process includes the following steps:

[0068] In step S101, based on the original construction schedule plan, the preset safety-process coupling knowledge sub-base and quality-process coupling knowledge sub-base, a safety-quality-process coupling evolution knowledge base is constructed, and a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is established.

[0069] The embodiments of this application can first establish a structured safety-quality-process coupling evolution knowledge base based on the safety-process coupling knowledge sub-base and the quality-process coupling knowledge sub-base. In this way, by constructing a structured knowledge base covering safety risks, quality hazards and dynamic interaction mechanisms of processes, a core knowledge foundation for generating scheduling responses can be provided.

[0070] Optionally, in one embodiment of this application, based on the original construction schedule plan, a pre-set safety-process coupling knowledge sub-base, and a quality-process coupling knowledge sub-base, a safety-quality-process coupling evolution knowledge base is constructed, and a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is established. This includes: acquiring multiple safety risks and multiple safety accidents during the target construction process, determining the risk level and risk category of each safety risk, and determining the corresponding process scheduling impact type based on the risk level and risk category; establishing an accident type mapping relationship between risk category and safety accident type, a level mapping relationship between each safety risk and risk level, a risk category mapping relationship between each safety risk and risk category, and a process mapping relationship between each safety risk and process scheduling impact type based on multiple safety accidents, so as to establish an accident type mapping relationship, a level mapping relationship, a risk category mapping relationship, and a process mapping relationship based on the accident type mapping relationship, the level mapping relationship, and the risk category mapping relationship. A safety-process coupled knowledge sub-base is constructed based on the mapping relationships between safety and process procedures. Multiple quality hazards during the target construction process are identified, and the severity of each hazard is quantified to obtain a corresponding severity level. Based on the severity level, multiple process interaction influence modes are established, and a status indicator corresponding to each quality hazard is determined. Mapping relationships are established between the status indicator and the quality hazard, severity level, and process interaction influence mode, respectively, to construct a quality-process coupled knowledge sub-base based on these mapping relationships. Based on the safety-process coupled knowledge sub-base and the quality-process coupled knowledge sub-base, a safety-quality-process coupled evolution knowledge base is constructed. Using a pre-defined graph database strategy, safety-process knowledge graphs and quality-process knowledge graphs corresponding to the safety-process coupled knowledge sub-base and the quality-process coupled knowledge sub-base are established, respectively, to construct a corresponding visualized coupling relationship knowledge graph based on the safety-process knowledge graph and the quality-process knowledge graph.

[0071] In practical implementation, this application's embodiments involve in-depth research into domestic and international construction standards, relevant literature, and construction accident case data. It systematically analyzes safety risks, safety accidents, and quality hazards during construction, establishing a structured safety-quality-process coupled evolutionary knowledge base. This base aims to systematically store the dynamic scheduling rules between safety issues, quality hazards, and process scheduling during construction, providing core knowledge support for realizing digital twin-driven online simulation of construction processes. This knowledge base consists of two core sub-bases: a safety-process coupled knowledge sub-base and a quality-process coupled knowledge sub-base.

[0072] Among them, the core entities of the safety-process coupling knowledge sub-base are 138 types of safety risks and 12 types of safety accidents. Safety risks cover five major categories: personnel factors, object factors, environmental factors, management factors, and structural factors. Safety accidents include vehicle accidents, object strikes, sinking, lifting accidents, electric shock, poisoning / suffocation, falls from heights, fires, mechanical injuries, collapses, and overturning.

[0073] Based on extensive case analysis, the embodiments of this application can determine the risk level of different safety issues and summarize four typical interaction modes: process parameter adjustment type, process increment type, process substitution type, and process lag type. The interaction modes are described as follows:

[0074] 1) The process parameter adjustment type is mainly for minor safety risks. It only adjusts the time or resource parameters of the process without changing the process logic structure.

[0075] 2) The incremental process type is suitable for situations with significant safety hazards, requiring the addition of new safety control steps to the original process;

[0076] 3) For process substitution to address serious safety hazards, the original process needs to be reorganized or completely modified;

[0077] 4) The delayed process type is used to deal with serious safety accidents that have occurred, resulting in a complete suspension of construction and rectification.

[0078] Therefore, the safety-process coupling knowledge sub-base clearly establishes four types of mapping relationships: safety risk type-possible accident type (i.e., accident type mapping relationship), safety risk-risk level (i.e., level mapping relationship), safety risk-risk category (i.e., risk category mapping relationship), and safety risk-process scheduling impact type (i.e., process mapping relationship).

[0079] The core entities of the quality-process coupled knowledge sub-base are nine types of quality hazards that occur in reinforced concrete structures, including honeycomb pitting, cracks, spalling, voids, exposed reinforcement, dimensional deviations, structural displacement, insufficient formwork stability, and defects in reinforcement binding quality.

[0080] Based on extensive case analysis, the severity of various quality hazards was quantified, resulting in three typical interactive impact patterns: process parameter adjustment type, stage repair embedded type, and emergency defect repair type. These interactive impact patterns are described below:

[0081] 1) The process parameter adjustment type is suitable for minor quality problems, which can be solved by adjusting the process time or resource parameters;

[0082] 2) Stage-based embedded repair is designed for moderate quality defects, and repair procedures are inserted after the current construction stage is completed;

[0083] 3) Emergency defect repair type is for serious quality problems that require immediate interruption of other processes for repair.

[0084] Therefore, the quality-process coupling knowledge sub-base of this application embodiment clearly establishes three types of mapping relationships: quality hazard-status indicator, status indicator-severity level, and status indicator-process scheduling impact type.

[0085] To achieve structured storage of the aforementioned coupling mechanisms, this application embodiment utilizes Neo4j graph database technology to construct a knowledge graph of the safety-quality-process coupling relationship. The safety-process knowledge graph incorporates six core node types, including safety risks, sub-projects, accident types, risk levels, risk categories, and process scheduling types, as well as five types of inter-node relationships. The quality-process knowledge graph comprises five core node types and four types of node relationships. The resulting knowledge graph contains 138 safety risks, 12 accident types, 13 quality hazards, and corresponding 7 process scheduling strategies. This achieves intuitive visualization and computer-readable structured expression of the coupling relationships among multiple construction elements, providing a knowledge foundation for subsequent digital twin-driven process deduction.

[0086] In step S102, the knowledge graph of coupling relationship is modeled to determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and a multi-objective reward function corresponding to the original construction schedule plan is constructed. Based on the state space, action space, state transition mechanism, multi-objective reward function and disturbance mechanism, a process scheduling optimization model is constructed.

[0087] Furthermore, embodiments of this application can design a reinforcement learning network architecture specifically for dynamic scheduling of construction processes, taking into account the unique characteristics of construction process scheduling, in order to construct a reinforcement learning network and model oriented towards process scheduling, thereby achieving intelligent scheduling decisions.

[0088] Optionally, in one embodiment of this application, a knowledge graph of coupling relationships is modeled to determine the corresponding state space, action space, state transition mechanism, and disturbance mechanism, and a multi-objective reward function corresponding to the original construction schedule is constructed. Based on the state space, action space, state transition mechanism, multi-objective reward function, and disturbance mechanism, a process scheduling optimization model is constructed, including: calculating the global environment state and process characteristic state in the construction process corresponding to the original construction schedule, and determining the state space based on a preset two-layer state representation strategy, combined with the global environment state and process characteristic state; constructing the action space based on a preset binary action representation strategy, and determining whether all predecessor processes of the current process have been completed according to the construction progress of the original construction schedule, wherein, if all predecessor processes are completed, the state of the current process is updated to determine the state transition mechanism; obtaining disturbance data in the construction process corresponding to the original construction schedule, and generating corresponding processing measures according to the disturbance data and a preset response strategy to determine the disturbance mechanism.

[0089] It should be noted that the embodiments of this application can be based on Markov decision process theory to construct a process scheduling optimization model, thereby abstracting the process scheduling problem into a mathematical form for representation and solution. The specific process is as follows:

[0090] (1) State-space modeling:

[0091] In the embodiments of this application, state space modeling can adopt a two-layer state representation method to comprehensively capture the construction environment state and process characteristic information.

[0092] The first layer represents the global environment state. The macroscopic characteristics of the construction simulation environment are described by equation (1), which includes five key dimensions: average resource utilization rate, standard deviation of resource utilization rate, process completion rate, standard deviation of completion rate, and estimated delay. Average resource utilization rate reflects the overall efficiency of resource use; standard deviation of resource utilization rate measures the balance of resource use for different types of work; process completion rate reflects the overall progress of the project; standard deviation of completion rate is calculated based on Bernoulli distribution to help the model identify different stages of the project; estimated delay is used to predict possible project delays.

[0093] (1)

[0094] in, Indicates average resource utilization rate; Indicates the standard deviation of resource utilization rate; Indicates the process completion rate; This represents the standard deviation of the completion rate; This indicates an estimated delay.

[0095] The second layer is the process characteristic state, as shown in Equation (2), which describes the micro-characteristics of each specific process, including seven dimensions: planned duration, number of workers required, number of materials required, floor, X-direction partition information, Y-direction partition information, and number of subsequent processes. Each process can be in one of four discrete states at any time: not ready (preceding processes are not all completed), ready (all preceding processes have been completed), in progress (selected for execution), and completed (execution completed).

[0096] (2)

[0097] in, Indicates the duration of the plan; Indicates the number of workers required; Indicates the quantity of materials required; Indicates the floor number; Indicates X-direction partition information; Indicates partition information in the Y direction; Indicates the number of subsequent processes.

[0098] (2) Action space modeling

[0099] Action space modeling defines how the system makes process scheduling decisions. This application's embodiments employ a binary action representation method, independently deciding whether to initiate each ready process. If multiple ready processes exist, the action space can be represented as a multi-dimensional binary vector, where each element takes the value 0 or 1, where 1 indicates initiating the process and 0 indicates not initiating it. This representation method's advantage lies in its ability to flexibly handle the parallel execution of multiple processes, aligning with the parallel operation characteristics in actual construction. Simultaneously, the action selection for a process is subject to various constraints, primarily logical dependency constraints and work time constraints. Logical dependency constraints ensure that a process can only initiate after all its predecessor processes are completed, a fundamental requirement for ensuring the correctness of the construction process. Work time constraints consider the actual working hours in construction (e.g., 9 a.m. - 17 p.m.); if the process duration spans non-working hours, the system will appropriately extend the completion time. The binary action space design offers greater flexibility, ensuring the engineering rationality of the scheduling scheme by combining logical dependencies and work time constraints.

[0100] (3) Process status transfer mechanism

[0101] The process state transition mechanism (i.e., the state transition mechanism) is the core logic of the entire scheduling system, controlling how a process transitions from one state to another, specifically as follows: Figure 2As shown. During system initialization, all processes are in a "not ready" state by default. Processes without predecessor processes automatically transition to a "ready" state, indicating they can begin execution immediately. As construction progresses, when all predecessor processes for a given process are completed, its state changes from "not ready" to "ready," and it enters the schedulable queue. The system selects one or more processes from the schedulable queue for execution. The selected process's state is updated to "in progress," and its planned completion time is recorded. When a process reaches its planned completion time, its state is automatically updated to "completed," simultaneously triggering a status update for its subsequent processes.

[0102] (4) Modeling of disturbance factors

[0103] Disturbance factor modeling abstracts uncertainties during construction into system disturbances, including external variables such as safety issues, quality risks, resource supply fluctuations, and weather changes. Specific modeling mechanisms include... Figure 3 As shown in the diagram. Specifically, when a disturbance is detected in the construction environment, the system first records the type, severity, location information, and scope of the disturbance. Then, it generates handling measures based on a preset response strategy, dynamically modifies the dependency network of the work processes, and finally re-plans the optimal work process execution plan.

[0104] For different types of disturbances, the embodiments of this application have designed differentiated processing mechanisms, as shown in Table 1.

[0105] Table 1

[0106]

[0107] For quality hazards, they are categorized into three levels based on severity: minor, moderate, and severe. This application's embodiments employ different handling strategies for each: minor hazards only require enhanced supervision; moderate hazards are repaired after all processes on the same floor are completed; and severe hazards are repaired immediately, with all unstarted processes waiting for repair completion. Similarly, safety risks are categorized into three levels based on severity, employing methods such as adjusting process parameters, adding preparatory processes, or replacing original processes. For safety accidents, the downtime is determined based on casualties, and all unstarted processes must wait until the downtime ends. This mechanism systematically integrates various unforeseen circumstances at the construction site with the process optimization decision framework, providing crucial environmental interaction capabilities for the agent to learn and adaptively adjust scheduling schemes under disturbances.

[0108] Optionally, in one embodiment of this application, the mathematical expression of the multi-objective reward function is:

[0109]

[0110] in, Status improvement rewards; Parallel process rewards; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

[0111] In addition, the embodiments of this application also need to design a multi-objective reward mechanism in the process of constructing the process scheduling optimization model. The design of this multi-objective reward mechanism comprehensively considers multiple optimization objectives in the actual construction process scheduling, forming a multi-level, multi-objective comprehensive evaluation system, as shown in equation (3):

[0112] (3)

[0113] in, Status improvement rewards; Parallel process rewards; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

[0114] It should be noted that this multi-objective reward mechanism includes five core reward or penalty items: state improvement reward, parallel process reward, idle time penalty, resource overrun penalty, and time delay penalty. Specifically, the state improvement reward assesses the change in the global environment state after each decision; improving average resource utilization, reducing the standard deviation of resource utilization, increasing completion rate, and reducing latency rate all result in positive rewards. The parallel process reward assesses the number of processes running simultaneously in each state, incentivizing decisions to fully utilize resource availability. The idle time penalty incurs a penalty when there are ready processes but no processes have been started, preventing downtime or resource waste. The resource overrun penalty penalizes when resource usage exceeds resource limits, ensuring that scheduling does not over-allocate resources. The time delay penalty penalizes project delays, while early completion results in additional rewards.

[0115] Therefore, the embodiments of this application, by constructing a multi-objective reward function, can effectively guide the agent to learn a scheduling strategy that balances multi-objective conflicts, which is the core driving force for realizing process optimization scheduling.

[0116] In step S103, a hierarchical deep reinforcement learning network model is determined through the process scheduling optimization model, and a construction simulation environment is constructed using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained so that the original construction schedule plan can be digitally twinned online based on the trained hierarchical deep reinforcement learning network model during the online simulation stage.

[0117] Furthermore, embodiments of this application can also address the problem that reinforcement learning models have difficulty effectively learning and responding to sudden construction disturbances by determining an adaptive model training method that considers disturbance events, thereby enabling the model to learn how to dynamically adjust the process network when a disturbance occurs.

[0118] Therefore, this application embodiment can achieve multi-element coupled evolution mechanism modeling of construction safety, quality, and processes by combining the original construction schedule plan (i.e., the original construction schedule plan) and real-time construction quality / safety issues, construct a process scheduling optimization model based on Markov chains, and finally realize digital twin online simulation in the subsequent online simulation stage, thereby assisting construction decision-making. Its overall execution framework is as follows: Figure 4 As shown in the figure. Among them, the process scheduling optimization model based on Markov chains and the online simulation based on digital twins are the core steps and core requirements.

[0119] Optionally, in one embodiment of this application, a hierarchical deep reinforcement learning network model is determined through a process scheduling optimization model, and a construction simulation environment is constructed using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained, including: constructing a hierarchical deep reinforcement learning network model based on a pre-constructed input layer, encoding layer, process attention layer, feature fusion layer, two-stream architecture layer, and output layer; extracting component geometric information and spatial relationships from a preset building information model, and obtaining construction activity data corresponding to component geometric information and spatial relationships based on a preset four-dimensional work decomposition structure; determining various process constraint rules based on the construction activity data, constructing a perturbation simulation generation algorithm based on the various process constraint rules and coupling relationship knowledge graph, and constructing a construction simulation environment through the perturbation simulation generation algorithm, so as to train the hierarchical deep reinforcement learning network model in the construction simulation environment through preset conventional construction paths and perturbation processing paths.

[0120] As one possible approach, embodiments of this application can use a hierarchical deep Q-Network (H-DQN) algorithm as the basic framework for work sequence scheduling to achieve construction sequence scheduling decisions under disturbances. This algorithm utilizes a hierarchical strategy, on the one hand determining the objective or sub-task (such as selecting objectives for optimizing schedule, prioritizing safety, or prioritizing quality) through a meta-strategy (high-level), and on the other hand executing specific scheduling actions (such as selecting specific work sequences or resource allocation) through low-level strategies. The specific process is as follows:

[0121] (1) H-DQN network architecture design

[0122] This application presents a hierarchical deep reinforcement learning network architecture specifically designed for dynamic scheduling problems of high-dimensional, multi-constraint construction processes. This architecture effectively integrates domain knowledge of construction scheduling and, through multi-level feature extraction, attention mechanisms to capture complex dependencies, high-level policy guidance, and two-stream value evaluation, effectively solves the challenges of traditional methods in handling large-scale processes, high-dimensional action spaces (parallel process selection), and sudden disturbances. It comprises six main components: an input layer, an encoding layer, a process attention layer, a feature fusion layer, a two-stream architecture layer, and an output layer. A detailed description of each component is as follows:

[0123] 1) The input layer adopts a dual-channel parallel design to receive global environment status and process feature status information respectively, avoiding information interference and laying the foundation for subsequent layered processing;

[0124] 2) The coding layer uses nonlinear transformations to map the original input to a high-dimensional representation space;

[0125] 3) The process attention layer captures complex dependencies between processes through four attention heads specifically designed for construction process scheduling (resource dependency head, spatial proximity head, temporal dependency head, and duration head), which can improve the model's ability to understand complex process networks.

[0126] 4) The feature fusion layer combines global information with process features to form a comprehensive representation, enabling each process decision to consider both its own characteristics and the global environment.

[0127] 5) The dual-stream architecture layer decomposes the Q-value into two parts: state value and action advantage, thereby improving the model's evaluation capability;

[0128] 6) The output layer generates process selection decisions, deeply integrating high-level strategy guidance and low-level Q-value evaluation to achieve intelligent and adaptive process selection.

[0129] (2) Construction of construction simulation environment

[0130] To establish a virtual environment that accurately maps the actual construction process, this application's embodiment designs a BIM (Building Information Modeling)-based process scheduling simulation environment, mainly consisting of four steps: BIM data extraction and processing, Work Breakdown Structure (WBS) data extraction, process constraint rule definition, and disturbance simulation generation. The specific details of each step are as follows:

[0131] 1) Extract component geometry and spatial relationships from the BIM model using the Revit-Dynamo plugin, including coordinate extraction, vertical zoning, horizontal zoning, and attribute labeling, converting building entities into digital representations, specifically as follows: Figure 5 As shown;

[0132] 2) Construct a four-dimensional work breakdown structure, decomposing construction activities into four dimensions: component type, process type, material type, and worker type;

[0133] 3) Define six types of process constraint rules, including Finish to Start rule, component construction sequence rule, vertical priority rule, construction sequence rule within area, sequential construction rule, and resource constraint rule;

[0134] 4) Based on the knowledge graph of the coupling relationship between safety, quality and process, develop a disturbance simulation generation algorithm to realize the simulation of the whole process from the determination of disturbance type to the generation of structured data.

[0135] (3) Simulation training

[0136] The model training process in this embodiment adopts a parallel architecture of conventional construction path and disturbance processing path, specifically as follows: Figure 6 As shown in Table 2, the conventional construction path handles the normal execution of procedures under undisturbed conditions, including four steps: disturbance detection, execution of the original plan, procedure initiation, and time advancement. The disturbance handling path is activated when a sudden situation is detected, including six steps: disturbance information recording, disturbance impact handling, scheduling system activation, Q-value calculation, procedure selection, and model update. The key optimization parameters during model training are shown in Table 2.

[0137] Table 2

[0138]

[0139] Therefore, the embodiments of this application realize the modeling and analysis of the coupling relationship between construction safety, quality and process, and sort out and present the correlation between 138 safety risks, 12 accident types, 13 quality hazards and 7 process influence modes. It can more accurately reflect the mutual influence relationship between various elements in the construction process and can more completely describe and analyze the complexity of the construction process. In addition, the embodiments of this application address the multi-element coupling problem of safety issues, quality hazards and process logic in the construction process. Through the digital twin dynamic inference method based on deep reinforcement learning and building information model, it can be well applied to specific scenarios such as real-time risk warning, multi-objective scheduling optimization and three-dimensional visualization decision-making in the construction process of building engineering.

[0140] According to the online digital twin simulation method for construction processes proposed in the embodiments of this application, which is applied to the offline training stage, a safety-quality-process coupling evolution knowledge base is constructed based on the original construction schedule plan, a preset safety-process coupling knowledge base, and a quality-process coupling knowledge base. A coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is also established. The coupling relationship knowledge graph is modeled to determine the corresponding state space, action space, state transition mechanism, and disturbance mechanism. A multi-objective reward function corresponding to the original construction schedule plan is constructed. Based on the state space, action space, state transition mechanism, multi-objective reward function, and disturbance mechanism, a process scheduling optimization model is constructed. Through the process scheduling optimization model, a hierarchical deep reinforcement learning network model is determined. A construction simulation environment is constructed using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained. During the online simulation stage, the original construction schedule plan is simulated online using a digital twin based on the trained hierarchical deep reinforcement learning network model. This application enables the modeling of the multi-element coupled evolution mechanism of construction safety, quality and process, constructs a process scheduling optimization model based on Markov chains, and ultimately realizes online digital twin simulation, thereby assisting construction decision-making.

[0141] Figure 7 The flowchart illustrates a digital twin online simulation method for construction processes applied in the online simulation phase, as provided in this application embodiment.

[0142] like Figure 7 As shown, the online digital twin simulation method for the construction process includes the following steps:

[0143] In step S701, the current construction site status corresponding to the target construction site is monitored, and the corresponding disturbance information is obtained based on the current construction site status.

[0144] In step S702, the disturbance information is input into the pre-constructed knowledge graph of coupling relationships to output the corresponding process scheduling scheme and process adjustment suggestions.

[0145] In step S703, the process scheduling scheme and process adjustment suggestions are converted to obtain the corresponding target format data, and the target format data is input into the pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

[0146] It should be noted that after the trained model is deployed to the actual construction environment, the embodiments of this application realize an online simulation process based on digital twins, such as... Figure 8 As shown, the specific process is as follows:

[0147] (1) Disturbance information acquisition

[0148] The embodiments of this application can first monitor the construction site status in real time through digital twin technology. When safety risks, quality hazards or other disturbances are detected, the system automatically records the detailed information of the disturbance, including key parameters such as problem description, location, time of occurrence, and severity. If a safety accident is involved, personnel casualties should also be recorded.

[0149] (2) Knowledge graph processing

[0150] Secondly, in this embodiment, the acquired disturbance information (such as sudden safety issues, quality problems, material supply shortages, weather changes, etc., which affect normal construction) can be input into a pre-built construction safety-quality-process coupling association knowledge graph for processing. By matching the problem with the knowledge graph, the corresponding process scheduling scheme can be obtained, and targeted process adjustment suggestions can be generated.

[0151] (3) Model input

[0152] Furthermore, the embodiments of this application can transform the scheduling suggestions, security risk assessments, quality control measures, and other information output by the knowledge graph into a data format suitable for input to the H-DQN model, including status representations such as process status, executed processes, processes to be executed, and related constraints.

[0153] (4) Model scheduling decision

[0154] Finally, in this embodiment, the trained H-DQN model can be invoked to make intelligent scheduling decisions for processes that have not yet been executed. The model calculates the Q value of each process based on the current construction status and disturbance characteristics, and outputs the optimal process scheduling scheme based on multi-objective optimization (i.e., the final process scheduling scheme).

[0155] Furthermore, embodiments of this application can send the generated optimal process scheduling scheme to on-site management personnel through a construction management platform, providing specific process execution suggestions and resource allocation guidance, thereby achieving fully automated response from disturbance detection to scheduling execution.

[0156] Therefore, the embodiments of this application construct an online process simulation system based on digital twin data with real-time response capabilities, which can reflect the dynamic changes, working status, risk level and other characteristics of the construction site and perform appropriate process optimization scheduling when safety problems or quality hazards occur.

[0157] The following specific embodiment, using a reinforced concrete residential building complex as an example, and in conjunction with the accompanying drawings, illustrates the execution logic and effects of the online digital twin simulation method for the construction process of this application.

[0158] Figure 9 This is a schematic diagram illustrating an online simulation of a reinforced concrete residential building complex project, as shown in a specific embodiment. Figure 9 As shown in the specific embodiment of this application, the reinforced concrete residential building complex includes a total of 5 buildings, each with 4 floors, such as... Figure 9 As shown in (a) of the diagram. The project adopts a traditional reinforced concrete frame structure, with main components including structural columns, structural walls, beams, and floor slabs. Through BIM model creation and grid division, the entire construction area is divided into a grid system of 6 areas in the X direction and 4 areas in the Y direction, forming 24 basic construction units, as shown in (a). Figure 9 As shown in (b) of the diagram.

[0159] This reinforced concrete residential building complex project is divided into 660 construction processes, including 165 processes each for rebar tying, formwork erection, concrete pouring, and concrete pouring, with a total concrete volume of 8491.3 cubic meters. Based on construction constraints, 1575 predecessor-successor relationships were established, with each process having 0-3 predecessor processes. In this specific embodiment, the Minizinc constrained programming method was used to generate the initial static scheduling plan, with the shortest construction period as the optimization objective, resulting in an optimal planned construction period of 789 days.

[0160] Subsequently, the embodiments of this application will select three representative cases of quality hazard disturbances for detailed analysis and verification:

[0161] Case (1) On the 154th day of construction, cracks of moderate severity were found in the wall in the 4th area of ​​X direction and the 1st area of ​​Y direction on the 1st floor; at this time, 353 procedures had been completed, and the progress was 53.6%.

[0162] Case (2) On the 397th day of construction, wall cracks were found in the 4th area of ​​X direction and the 1st area of ​​Y direction on the 1st floor. The severity was moderate. At this time, 572 procedures had been completed, and the completion rate was 86.7%.

[0163] Case (3) On the 154th day of construction, severe cracks were found in the wall in the 4th area of ​​X direction and the 1st area of ​​Y direction on the 1st floor; at this time, 353 procedures had been completed and the progress was 53.6%.

[0164] Through real-time monitoring using digital twin technology, this application embodiment can detect potential quality hazards and automatically record detailed information. For example, for case (1), the recorded information includes key parameters such as the problem description "cracks appear in the concrete wall", the location of occurrence "1st floor X4Y1 area", the severity "medium", and the discovery time "2024-06-01 10:30:00".

[0165] In this embodiment, disturbance information can be input into a pre-constructed quality-process coupling relationship knowledge graph, and relevant nodes and relationships can be matched using the Cypher query language. Based on the mapping relationship of "concrete wall crack - moderate severity - staged repair embedded type" in the knowledge graph, it is determined that a repair process needs to be inserted after all processes on that floor are completed.

[0166] Furthermore, the embodiments of this application can convert the disturbance information into the input format of the H-DQN model. The model solution times are (1) 13.02 seconds (2) 2.95 seconds (3) 7.99 seconds, where Case (1) decides to perform the repair process on the 271st day, increasing the total project duration by 2 days; Case (2) decides to perform the repair process on the 407th day, increasing the total project duration by 1 day; and Case (3) decides to perform the repair process on the 155th day, increasing the total project duration by 3 days.

[0167] The scheduling results are displayed through a visual interface. The Gantt chart shows that the model strictly follows the original plan and maintains the established construction sequence before the disturbance occurs. In Case (1), after the disturbance occurs, the specific embodiment of this application adds one repair process and adjusts the time arrangement of subsequent processes, such as Figure 10 As shown in the cumulative project progress curve, although there was a slight delay in the mid-term progress, the progress deficit was successfully made up in the later stages through reasonable resource allocation. Figure 11 As shown in the analysis of resource allocation and optimization results, the demand curves for various types of workers and materials remained relatively stable, with no significant concentration or idleness of resources. Figure 12 and Figure 13 As shown.

[0168] Similarly, the scheduling change Gantt chart, cumulative progress curve comparison, demand curve comparison for each type of worker, and demand curve comparison for each type of material in Case (2) are shown below. Figure 14 , Figure 15 , Figure 16 and Figure 17 As shown, the Gantt chart of the scheduling change, the comparison of the cumulative progress curve, the comparison of the demand curves for each type of worker, and the comparison of the demand curves for each type of material in Case (3) are as follows: Figure 18 , Figure 19 , Figure 20 , Figure 21 As shown; among them. In case (2), the number of unfinished processes is relatively small, and the adjustments are also relatively small. The red line and the blue line in the cumulative progress curve are highly overlapping; In case (3), since the risk level of the disturbance is severe, the repair process needs to be executed immediately. Therefore, the repair process was executed immediately in the scheduling plan, which is in line with the scheduling logic.

[0169] In summary, the hierarchical deep reinforcement learning network model demonstrates a fast response speed when handling disturbance cases, completing scheduling decisions for over 600 processes within 15 seconds with minimal impact on project time, balanced resource utilization, and no resource overruns. Furthermore, the hierarchical deep reinforcement learning network model employs a reasonable scheduling strategy, arranging repair processes at appropriate times to avoid excessive interference with the critical path.

[0170] The online digital twin simulation method for construction processes proposed in this application involves monitoring the current state of the target construction site and acquiring corresponding disturbance information based on this state. This disturbance information is then input into a pre-constructed knowledge graph of coupled relationships to output corresponding process scheduling schemes and process adjustment suggestions. The process scheduling schemes and process adjustment suggestions are then format-converted to obtain corresponding target format data, which is input into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme. This application enables the modeling of the multi-element coupled evolution mechanism of construction safety, quality, and processes, constructs a process scheduling optimization model based on Markov chains, and ultimately achieves online digital twin simulation, thereby assisting in construction decision-making.

[0171] Secondly, the construction process digital twin online simulation device proposed according to the embodiments of this application is described with reference to the accompanying drawings.

[0172] Figure 22 This is a block diagram of a construction process digital twin online simulation device applied to the offline training phase, according to an embodiment of this application.

[0173] like Figure 22 As shown, the construction process digital twin online simulation device 10 applied to the offline training phase includes: a map construction module 101, a modeling module 102, and a training module 103.

[0174] Among them, the graph construction module 101 is used to construct a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, the preset safety-process coupling knowledge sub-base and quality-process coupling knowledge sub-base, and to establish a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base.

[0175] Modeling module 102 is used to model the knowledge graph of coupling relationships, determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and construct the multi-objective reward function corresponding to the original construction schedule plan, so as to build a process scheduling optimization model based on the state space, action space, state transition mechanism, multi-objective reward function and disturbance mechanism.

[0176] Training module 103 is used to determine the hierarchical deep reinforcement learning network model through the process scheduling optimization model, and to build a construction simulation environment using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained so that the original construction schedule plan can be digitally twinned online based on the trained hierarchical deep reinforcement learning network model during the online simulation phase.

[0177] Optionally, in one embodiment of this application, the map construction module 101 includes: an acquisition unit, a first mapping unit, a determination unit, a second mapping unit, and an establishment unit.

[0178] The acquisition unit is used to acquire various safety risks and accidents during the target construction process, determine the risk level and risk category of each safety risk, and determine the corresponding process scheduling impact type based on the risk level and risk category.

[0179] The first mapping unit is used to establish accident type mapping relationships between risk categories and safety accident types, level mapping relationships between each safety risk and risk level, risk category mapping relationships between each safety risk and risk category, and process mapping relationships between each safety risk and process scheduling impact type based on multiple safety accidents, so as to construct a safety-process coupled knowledge sub-base according to the accident type mapping relationship, level mapping relationship, risk category mapping relationship and process mapping relationship.

[0180] The determination unit is used to acquire various quality hazards during the target construction process, quantify the severity of each quality hazard to obtain the corresponding severity level, and establish multiple process interaction influence modes based on the severity level, and determine the status index corresponding to each quality hazard.

[0181] The second mapping unit is used to establish mapping relationships between status indicators and quality hazards, severity levels, and process interaction patterns, so as to construct a quality-process coupled knowledge sub-base based on the mapping relationships.

[0182] A unit is established to construct a safety-quality-process coupled evolutionary knowledge base based on the safety-process coupled knowledge base and the quality-process coupled knowledge base. Through a preset graph database strategy, safety-process knowledge graphs and quality-process knowledge graphs corresponding to the safety-process coupled knowledge base and the quality-process coupled knowledge base are established respectively, so as to construct corresponding visualized coupling relationship knowledge graphs based on the safety-process knowledge graph and the quality-process knowledge graph.

[0183] Optionally, in one embodiment of this application, the modeling module 102 includes: a calculation unit, a judgment unit, and a generation unit.

[0184] The calculation unit is used to calculate the global environment state and process characteristic state in the construction process corresponding to the original construction schedule plan, and to determine the state space based on the preset two-layer state representation strategy, combining the global environment state and process characteristic state.

[0185] The judgment unit is used to construct an action space based on a preset binary action representation strategy, and to determine whether all the preceding actions of the current process have been completed according to the construction progress of the original construction progress plan. If all the preceding actions have been completed, the state of the current process is updated to determine the state transition mechanism.

[0186] The generation unit is used to acquire disturbance data during the construction process corresponding to the original construction schedule plan, and generate corresponding processing measures based on the disturbance data and the preset response strategy to determine the disturbance mechanism.

[0187] Optionally, in one embodiment of this application, the training module 103 includes: a first construction unit, an extraction unit, and a second construction unit.

[0188] The first building unit is used to construct a hierarchical deep reinforcement learning network model based on a pre-built input layer, encoding layer, process attention layer, feature fusion layer, two-stream architecture layer, and output layer.

[0189] The extraction unit is used to extract component geometric information and spatial relationships from a preset building information model, and to obtain construction activity data corresponding to component geometric information and spatial relationships based on a preset four-dimensional work breakdown structure.

[0190] The second building unit is used to determine multiple process constraint rules based on construction activity data, and to build a disturbance simulation generation algorithm based on the multiple process constraint rules and coupling relationship knowledge graph. The disturbance simulation generation algorithm is used to build a construction simulation environment, and to train a hierarchical deep reinforcement learning network model in the construction simulation environment through preset conventional construction paths and disturbance processing paths.

[0191] Optionally, in one embodiment of this application, the mathematical expression of the multi-objective reward function is:

[0192]

[0193] in, Indicates a reward for improved status; Indicates a reward for parallel processing steps; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

[0194] It should be noted that the foregoing explanation of the embodiment of the construction process digital twin online simulation method applied to the offline training stage also applies to the construction process digital twin online simulation device 10 applied to the offline training stage of this embodiment, and will not be repeated here.

[0195] The construction process digital twin online simulation device 10 proposed in the embodiments of this application for offline training includes a graph construction module 101, used to construct a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, a preset safety-process coupling knowledge base and a quality-process coupling knowledge base, and to establish a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base; a modeling module 102, used to model the coupling relationship knowledge graph, determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and construct a multi-objective reward function corresponding to the original construction schedule plan, so as to construct a process scheduling optimization model based on the state space, action space, state transition mechanism, multi-objective reward function and disturbance mechanism; and a training module 103, used to determine a hierarchical deep reinforcement learning network model through the process scheduling optimization model, and to construct a construction simulation environment using the hierarchical deep reinforcement learning network model, and to train the hierarchical deep reinforcement learning network model based on the construction simulation environment, so as to perform digital twin online simulation of the original construction schedule plan based on the trained hierarchical deep reinforcement learning network model in the online simulation stage. This application enables the modeling of the multi-element coupled evolution mechanism of construction safety, quality and process, constructs a process scheduling optimization model based on Markov chains, and ultimately realizes online digital twin simulation, thereby assisting construction decision-making.

[0196] Figure 23 This is a block diagram of a digital twin online simulation device for the construction process applied in the online simulation phase according to an embodiment of this application.

[0197] like Figure 23 As shown, the construction process digital twin online simulation device 20 applied to the online simulation stage includes: a monitoring module 201, an input module 202, and a simulation module 203.

[0198] The monitoring module 201 is used to monitor the current construction site status corresponding to the target construction site and obtain the corresponding disturbance information based on the current construction site status.

[0199] The input module 202 is used to input disturbance information into a pre-built knowledge graph of coupling relationships to output corresponding process scheduling schemes and process adjustment suggestions.

[0200] The deduction module 203 is used to convert the process scheduling scheme and process adjustment suggestions into the corresponding target format data, and input the target format data into the pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

[0201] It should be noted that the foregoing explanation of the embodiment of the construction process digital twin online simulation method applied to the online simulation stage also applies to the construction process digital twin online simulation device 20 applied to the online simulation stage of this embodiment, and will not be repeated here.

[0202] The construction process digital twin online simulation device 20 proposed in this application, applied to the online simulation stage, includes a monitoring module 201 for monitoring the current construction site status corresponding to the target construction site and obtaining corresponding disturbance information based on the current construction site status; an input module 202 for inputting the disturbance information into a pre-constructed coupling relationship knowledge graph to output corresponding process scheduling schemes and process adjustment suggestions; and a simulation module 203 for converting the process scheduling schemes and process adjustment suggestions into corresponding target format data, and inputting the target format data into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme. This application can realize the multi-element coupling evolution mechanism modeling of construction safety, quality and processes, construct a process scheduling optimization model based on Markov chains, and ultimately realize digital twin online simulation, thereby assisting construction decision-making.

[0203] Figure 24 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0204] The memory 2401, the processor 2402, and the computer program stored on the memory 2401 and capable of running on the processor 2402.

[0205] When the processor 2402 executes the program, it implements the online simulation method for digital twins of the construction process provided in the above embodiments.

[0206] Furthermore, electronic devices also include:

[0207] Communication interface 2403 is used for communication between memory 2401 and processor 2402.

[0208] The memory 2401 is used to store computer programs that can run on the processor 2402.

[0209] The memory 2401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0210] If the memory 2401, processor 2402, and communication interface 2403 are implemented independently, then the communication interface 2403, memory 2401, and processor 2402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 24 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0211] Optionally, in a specific implementation, if the memory 2401, processor 2402, and communication interface 2403 are integrated on a single chip, then the memory 2401, processor 2402, and communication interface 2403 can communicate with each other through an internal interface.

[0212] The processor 2402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0213] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described online digital twin simulation method for the construction process.

[0214] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0215] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0216] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0217] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0218] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0219] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0220] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0221] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for online digital twin simulation of a construction process, applied to the offline training phase, characterized in that, Includes the following steps: Based on the original construction schedule plan, the pre-set safety-process coupling knowledge sub-base and quality-process coupling knowledge sub-base, a safety-quality-process coupling evolution knowledge base is constructed, and a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is established; The knowledge graph of the coupling relationship is modeled to determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and a multi-objective reward function corresponding to the original construction schedule plan is constructed. Based on the state space, the action space, the state transition mechanism, the multi-objective reward function and the disturbance mechanism, a process scheduling optimization model is constructed. The hierarchical deep reinforcement learning network model is determined through the process scheduling optimization model, and a construction simulation environment is constructed using the hierarchical deep reinforcement learning network model. Based on the construction simulation environment, the hierarchical deep reinforcement learning network model is trained so that the original construction schedule plan can be digitally twinned online based on the trained hierarchical deep reinforcement learning network model during the online simulation phase. Specifically, based on the original construction schedule plan, a pre-set safety-process coupling knowledge sub-base, and a quality-process coupling knowledge sub-base, a safety-quality-process coupling evolution knowledge base is constructed, and a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base is established, including: The system acquires various safety risks and accidents during the target construction process, determines the risk level and risk category of each of the various safety risks, and determines the corresponding process scheduling impact type based on the risk level and risk category. Based on the aforementioned multiple safety incidents, an accident type mapping relationship is established between the risk category and the safety incident type, a level mapping relationship is established between each safety risk and the risk level, a risk category mapping relationship is established between each safety risk and the risk category, and a process mapping relationship is established between each safety risk and the process scheduling impact type. The safety-process coupled knowledge sub-base is then constructed based on the accident type mapping relationship, the level mapping relationship, the risk category mapping relationship, and the process mapping relationship. Multiple quality hazards during the target construction process are identified, and the severity of each quality hazard is quantified to obtain a corresponding severity level. Based on the severity level, multiple process interaction and influence modes are established, and the status index corresponding to each quality hazard is determined. Establish mapping relationships between the status indicators and the quality hazards, the severity levels, and the process interaction patterns, respectively, and construct the quality-process coupled knowledge sub-base based on the mapping relationships; Based on the safety-process coupled knowledge sub-base and the quality-process coupled knowledge sub-base, a safety-quality-process coupled evolution knowledge base is constructed. Through a preset graph database strategy, safety-process knowledge graphs and quality-process knowledge graphs corresponding to the safety-process coupled knowledge sub-base and the quality-process coupled knowledge sub-base are established respectively, so as to construct corresponding visualized coupling relationship knowledge graphs based on the safety-process knowledge graph and the quality-process knowledge graph. The process involves modeling the knowledge graph of the coupling relationship, determining the corresponding state space, action space, state transition mechanism, and disturbance mechanism, and constructing a multi-objective reward function corresponding to the original construction schedule plan. Based on the state space, action space, state transition mechanism, multi-objective reward function, and disturbance mechanism, a process scheduling optimization model is constructed, including: Calculate the global environment state and process characteristic state during the construction process corresponding to the original construction schedule plan, and determine the state space by combining the global environment state and the process characteristic state based on a preset two-layer state representation strategy. Based on a preset binary action representation strategy, the action space is constructed, and according to the construction progress of the original construction progress plan, it is determined whether all the preceding processes of the current process have been completed. If all the preceding processes have been completed, the state of the current process is updated to determine the state transition mechanism. Obtain disturbance data during the construction process corresponding to the original construction schedule plan, and generate corresponding processing measures based on the disturbance data and the preset response strategy to determine the disturbance mechanism; The step of determining a hierarchical deep reinforcement learning network model through the process scheduling optimization model, constructing a construction simulation environment using the hierarchical deep reinforcement learning network model, and training the hierarchical deep reinforcement learning network model based on the construction simulation environment includes: The hierarchical deep reinforcement learning network model is constructed based on a pre-built input layer, encoding layer, process attention layer, feature fusion layer, two-stream architecture layer, and output layer. Extract component geometric information and spatial relationships from a preset building information model, and obtain construction activity data corresponding to the component geometric information and spatial relationships based on a preset four-dimensional work decomposition structure; Based on the construction activity data, multiple process constraint rules are determined. Based on the multiple process constraint rules and the coupling relationship knowledge graph, a disturbance simulation generation algorithm is constructed. The construction simulation environment is constructed through the disturbance simulation generation algorithm. In the construction simulation environment, the hierarchical deep reinforcement learning network model is trained through preset conventional construction paths and disturbance processing paths. The mathematical expression for the multi-objective reward function is: in, Indicates a reward for improved status; Indicates a reward for parallel processing steps; This indicates a penalty for idle time; This indicates a penalty for exceeding resource limits; This indicates a time delay penalty.

2. A digital twin online simulation method for a construction process, applied in the online simulation stage, characterized in that, The online simulation method for digital twins of construction processes applied in the offline training phase, as described in claim 1, includes the following steps: Monitor the current construction site status corresponding to the target construction site, and obtain the corresponding disturbance information based on the current construction site status; The disturbance information is input into a pre-constructed knowledge graph of coupling relationships to output corresponding process scheduling schemes and process adjustment suggestions; The process scheduling scheme and the process adjustment suggestions are converted to obtain the corresponding target format data, and the target format data is input into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

3. A digital twin online simulation device for construction processes, applied in the offline training phase, characterized in that, The method for online digital twin simulation of construction processes applied in the offline training phase, as described in claim 1, wherein the apparatus comprises: The knowledge graph construction module is used to construct a safety-quality-process coupling evolution knowledge base based on the original construction schedule plan, the preset safety-process coupling knowledge sub-base and quality-process coupling knowledge sub-base, and to establish a coupling relationship knowledge graph corresponding to the safety-quality-process coupling evolution knowledge base; The modeling module is used to model the coupling relationship knowledge graph, determine the corresponding state space, action space, state transition mechanism and disturbance mechanism, and construct the multi-objective reward function corresponding to the original construction progress plan, so as to construct a process scheduling optimization model based on the state space, the action space, the state transition mechanism, the multi-objective reward function and the disturbance mechanism; The training module is used to determine a hierarchical deep reinforcement learning network model through the process scheduling optimization model, construct a construction simulation environment using the hierarchical deep reinforcement learning network model, and train the hierarchical deep reinforcement learning network model based on the construction simulation environment, so as to perform digital twin online simulation of the original construction schedule plan based on the trained hierarchical deep reinforcement learning network model during the online simulation phase.

4. A digital twin online simulation device for a construction process, applied in the online simulation phase, characterized in that, The method for online simulation of construction processes using digital twins as described in claim 2, wherein the apparatus comprises: The monitoring module is used to monitor the current construction site status corresponding to the target construction site and obtain the corresponding disturbance information based on the current construction site status. The input module is used to input the disturbance information into a pre-constructed knowledge graph of coupling relationships to output corresponding process scheduling schemes and process adjustment suggestions; The deduction module is used to convert the format of the process scheduling scheme and the process adjustment suggestions to obtain the corresponding target format data, and input the target format data into a pre-trained hierarchical deep reinforcement learning network model to output the final process scheduling scheme.

5. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the online digital twin simulation method for the construction process as described in claim 1 or claim 2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the online digital twin simulation method for the construction process as described in claim 1 or claim 2.

Images