Intelligent construction dynamic management and control system and method based on digital twinning

Abstract

This invention relates to a digital twin-based intelligent construction dynamic management and control system and method, belonging to the field of intelligent construction technology. It includes a data acquisition module, a digital twin model module, a state fusion and deviation analysis module, a predictive simulation and multi-objective optimization decision-making module, and a closed-loop control execution module. It can not only map the physical state of the construction site in real time, but also proactively predict potential construction deviations by integrating multi-dimensional data such as quality, safety, schedule, and cost during the construction process. When a deviation is predicted or detected in real time, the system automatically runs a multi-objective collaborative optimization algorithm that considers the balance of the four elements of QSPC (Quality, Safety, Schedule, and Process) to generate the optimal adjustment strategy. Finally, the closed-loop control execution module translates the optimization strategy into specific instructions and issues them to the construction site, achieving proactive, predictive, and closed-loop dynamic management and control of the construction process, significantly improving the intelligence level of construction management and the overall benefits of engineering projects.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction, specifically to a dynamic control system and method for intelligent construction based on digital twins. Background Technology

[0002] With the development of information technology, the application of digital twin technology in the construction industry (AEC) is increasing. Existing intelligent construction systems typically use Building Information Modeling (BIM) as a foundation, combined with Internet of Things (IoT) devices to collect real-time data from the construction site (such as component locations, environmental temperature and humidity, personnel activities, etc.), to build a digital model in virtual space that corresponds to the physical construction site, enabling visualized monitoring of construction progress, material tracking, and simple "design-actual" discrepancy comparison.

[0003] However, existing technologies have the following significant shortcomings:

[0004] Reactive management: Most systems can only detect and alert on deviations (such as schedule delays or quality problems) after they have already occurred, lacking foresight and resulting in delayed management response.

[0005] Weak decision support capabilities: When deviations are detected, the system usually only provides deviation information, while the complex decision-making process of how to adjust resources and modify plans still relies heavily on the project manager's personal experience and lacks scientific and quantitative decision support.

[0006] Single-goal orientation: Although some systems can provide simple optimization suggestions, they often only focus on a single goal (such as "how to meet the deadline"), ignoring the cost surge and safety risks that may result from rushing the work, and lacking a comprehensive balance of multiple core elements such as quality, safety, schedule, and cost.

[0007] One-way data flow: Data flow is mostly a one-way mapping from the physical world to the virtual world. The analysis and decision-making results in the virtual model are difficult to efficiently and accurately affect the construction execution in the physical world.

[0008] Therefore, there is an urgent need for an intelligent construction dynamic management and control system that can achieve proactive prediction, intelligent decision-making, and closed-loop control to overcome the limitations of existing technologies. Summary of the Invention

[0009] Purpose of the invention: To provide a dynamic control system for intelligent construction based on digital twins, and further to provide a control method based on the above-mentioned dynamic control system for intelligent construction, so as to solve the above-mentioned problems existing in the prior art.

[0010] Technical solution: A dynamic control system for intelligent construction based on digital twins, comprising:

[0011] The data acquisition module is used to acquire multi-source heterogeneous data from physical construction sites and virtual project plans;

[0012] The digital twin model module is used to build a digital twin that is synchronized with the physical construction site in real time.

[0013] The state fusion and deviation analysis module is used to fuse the multi-source heterogeneous data to form a four-dimensional real-time state reflecting the quality-safety-schedule-cost (QSPC) of the construction site, and compare it with the preset benchmark state to identify immediate or potential construction deviations.

[0014] The predictive simulation and multi-objective optimization decision module is characterized in that it is configured to: when a deviation is identified, based on historical data and real-time status, use a predictive algorithm to deduce the chain effect of the deviation on the future four-dimensional state of QSPC; and start a multi-objective collaborative optimization algorithm to seek the Pareto optimal solution among quality, safety, schedule and cost, and automatically generate a dynamic optimization control strategy that includes resource reallocation, process reordering or process adjustment.

[0015] The closed-loop control execution module is characterized in that it is used to parse the dynamic optimization control strategy into executable digital instructions and send them to the intelligent terminals of automated equipment, managers or operators at the physical construction site to drive physical construction activities to be executed according to the optimized strategy, thereby forming a control closed loop from data perception to physical execution.

[0016] In a further embodiment, the state fusion and deviation analysis module further includes a model self-evolution unit, which is used to record the characteristics of each deviation, the generated optimization control strategy, and the final execution effect, and use these records as training data to continuously iterate and optimize the prediction algorithm and optimization algorithm model in the predictive simulation and multi-objective optimization decision module.

[0017] In a further embodiment, the predictive simulation and multi-objective optimization decision-making module employs prediction algorithms including, but not limited to, time series analysis, Markov chains, or long short-term memory networks (LSTM) to predict the probability of construction schedule delays, cost overruns, or the incidence of safety accidents.

[0018] In a further embodiment, the predictive simulation and multi-objective optimization decision-making module employs a multi-objective collaborative optimization algorithm such as a genetic algorithm (GA), a particle swarm optimization algorithm (PSO), or a non-dominated sorting genetic algorithm (NSGA-II), whose objective function integrates quantitative assessments of quality scores, safety risk indices, total project duration, and total cost.

[0019] In a further embodiment, the executable digital instructions issued by the closed-loop control execution module include: component hoisting sequence adjustment instructions sent to the tower crane automatic scheduling system, path and time adjustment instructions sent to the material delivery robot, or updated 3D construction briefings and process inspection points pushed to workers wearing AR glasses.

[0020] In a further embodiment, the digital twin constructed by the digital twin model module not only includes geometric information (3D), progress information (4D), and cost information (5D), but also deeply integrates the resource status (personnel, equipment, materials), environmental data (temperature, humidity, wind speed), and process flow logic of the construction site, forming a high-fidelity, multi-dimensional virtual-real symbiosis.

[0021] A management and control method for a digital twin-based intelligent construction dynamic management and control system, characterized by the following steps:

[0022] Real-time data collection of physical construction site data and virtual project plan data is achieved through multi-source sensors and information systems.

[0023] Construct and synchronize in real time a digital twin model that reflects all elements of the construction site, including people, machinery, materials, methods, and environment;

[0024] By integrating the data, the four-dimensional real-time status of the construction process (Quality-Safety-Schedule-Cost, QSPC) is calculated and evaluated, and compared with a preset dynamic benchmark to identify construction deviations.

[0025] The implementation of closed-loop control is characterized by the step of converting the optimal dynamic adjustment scheme into specific, executable electronic instructions and issuing them to the corresponding execution units at the construction site to drive the adjustment of the physical construction process in reverse.

[0026] In a further embodiment, the predictive analysis and optimization decision-making specifically includes the following steps:

[0027] Once the deviation is identified, the predictive model is invoked to deduce the future development trend of the deviation without intervention and its comprehensive impact on various dimensions of QSPC.

[0028] Trigger a multi-objective collaborative optimization process that aims to minimize quality deductions, safety risks, schedule delays, and cost overruns, and searches for and generates the optimal dynamic adjustment scheme under constraints.

[0029] In a further embodiment, after performing predictive analysis and optimization decision-making, the method further includes:

[0030] Knowledge accumulation and model iteration involve the structured storage of identified construction deviations, adjustment schemes generated in predictive analysis and optimization decisions, and the results executed in closed-loop control steps, forming a case knowledge base. This knowledge base is then used to periodically or triggerively retrain the predictive model initiated after the deviation is identified and the optimization model that triggers a multi-objective collaborative optimization process, thereby improving their accuracy and effectiveness.

[0031] In a further embodiment, the dynamic benchmark is dynamically adjusted based on the initial plan and combined with the actual data of the completed construction portion, making the deviation analysis more timely and accurate.

[0032] After an electronic command is issued, the system will continuously monitor the execution status and effect of the command, and use the feedback data as the input for the next round of data, forming a dynamic control cycle of continuous iteration and self-correction.

[0033] Beneficial Effects: This invention relates to a dynamic control system and method for intelligent construction based on digital twins, which has the following beneficial effects:

[0034] 1. Proactive forecasting and forward-looking management

[0035] Traditional intelligent construction systems are mostly reactive, only detecting and alerting after deviations occur. This system, however, proactively predicts the cascading effects of construction deviations on the future four-dimensional state of Quality-Safety-Schedule-Cost (QSPC) through predictive simulation and multi-objective optimization decision-making modules. Based on historical data and real-time status, it utilizes predictive algorithms such as time series analysis, Markov chains, or Long Short-Term Memory (LSTM) networks. This allows the system to anticipate potential problems, enabling proactive management, preventing problems from escalating, and reducing unnecessary losses.

[0036] For example, by accurately predicting the probability of construction delays, the risk of cost overruns, or the incidence of safety accidents, the system can issue early warnings, giving managers enough time to take countermeasures.

[0037] 2. Scientifically Quantitative Decision Support

[0038] Existing technologies rely heavily on the project manager's personal experience in the decision-making process when deviations are detected, lacking scientific and quantitative support. This system, upon identifying a deviation, will initiate a multi-objective collaborative optimization algorithm, such as a genetic algorithm (GA), particle swarm optimization (PSO), or non-dominated sorting genetic algorithm (NSGA-II), to seek a Pareto optimal solution among quality, safety, schedule, and cost.

[0039] These algorithms integrate quantitative assessments of quality scores, safety risk indices, total project duration, and total cost into their objective functions. They can automatically generate dynamic optimization control strategies that include resource reallocation, process reordering, or process adjustments, providing a scientific basis for decision-making and improving the accuracy and reliability of decisions.

[0040] 3. Comprehensive trade-offs among multiple objectives

[0041] Some existing systems focus only on a single objective, neglecting the comprehensive trade-off between multiple core elements. This system fully considers multiple aspects such as quality, safety, schedule, and cost, and achieves balanced optimization of multiple objectives while satisfying various constraints through a multi-objective collaborative optimization algorithm.

[0042] For example, while rushing to meet deadlines, we will comprehensively consider the increased costs and safety risks to avoid serious problems in other aspects due to the pursuit of a single goal, thereby improving the overall benefits of the project.

[0043] 4. Closed-loop dynamic control

[0044] Traditional systems often involve unidirectional data flow, making it difficult for analysis and decision-making results from virtual models to effectively apply to physical construction execution. This system, through a closed-loop control execution module, parses dynamic optimization control strategies into executable digital instructions and distributes them to automated equipment, management personnel, or workers' smart terminals at the physical construction site, forming a closed-loop control system from data perception to physical execution.

[0045] The system continuously monitors the execution status and effects of instructions, and uses the feedback data as the input for the next round of data, forming a dynamic control cycle of continuous iteration and self-correction. This ensures that the construction process is always executed according to the optimized strategy, significantly improving the level of intelligence in construction management.

[0046] 5. Model self-evolution and knowledge accumulation

[0047] The model self-evolution unit in the state fusion and deviation analysis module records the characteristics of each deviation, the generated optimization control strategy, and the final execution effect. These records are used as training data to continuously iterate and optimize the prediction algorithm and optimization algorithm model in the predictive simulation and multi-objective optimization decision module.

[0048] Meanwhile, through knowledge accumulation and model iteration, construction deviations, adjustment plans, and execution results are stored in a structured manner to form a case knowledge base. The prediction and optimization models are retrained regularly or triggered periodically to continuously improve the accuracy and effectiveness of the models. Attached Figure Description

[0049] Figure 1This is a schematic diagram of the composition framework of the intelligent construction dynamic control system based on digital twins described in this invention.

[0050] Figure 2 This is a flowchart illustrating the intelligent construction dynamic control method based on digital twins described in this invention. Detailed Implementation

[0051] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0052] The purpose of this invention is to provide a dynamic control system and method for intelligent construction based on digital twins, which aims to solve the problems of passive control, subjective decision-making, and one-sided optimization in the existing technology.

[0053] To achieve the above objectives, the present invention adopts the following technical solution:

[0054] 1. Data Acquisition Module

[0055] Data sources: Multi-source heterogeneous data are obtained from physical construction sites and virtual project plans. Data from physical construction sites is collected through multi-source sensors, including but not limited to component position sensors, environmental temperature and humidity sensors, and personnel activity monitoring equipment; virtual project plan data is obtained from project management information systems, such as construction schedule plans and cost budgets.

[0056] Data types: The collected data includes the status of resources at the construction site (personnel, equipment, materials), environmental data (temperature, humidity, wind speed), construction progress information, quality inspection data, safety monitoring data, etc.

[0057] 2. Digital Twin Model Module

[0058] Model building: Construct a digital twin that is synchronized with the physical construction site in real time. This digital twin not only includes geometric information (3D), schedule information (4D), and cost information (5D), but also deeply integrates the resource status, environmental data, and process logic of the construction site, forming a high-fidelity, multi-dimensional virtual-real symbiosis.

[0059] Real-time synchronization: The digital twin model is continuously updated with real-time data acquired through the data acquisition module to ensure that it remains consistent with the physical construction site.

[0060] 3. State Fusion and Deviation Analysis Module

[0061] Data fusion: Integrating heterogeneous data from multiple sources to form a four-dimensional real-time status reflecting the quality, safety, schedule, and cost (QSPC) of the construction site. Data fusion algorithms, such as weighted average and Kalman filtering, can be used to integrate data from different sources.

[0062] Deviation identification: The fused 4D real-time status is compared with a preset baseline status to identify immediate or potential construction deviations. The preset baseline status is dynamically adjusted based on the initial plan and combined with actual data from completed construction sections, making deviation analysis more timely and accurate.

[0063] Model self-evolution: The model self-evolution unit records the characteristics of each deviation, the generated optimized control strategy, and the final execution effect, and uses these records as training data to continuously iterate and optimize the prediction algorithm and optimization algorithm model in the predictive simulation and multi-objective optimization decision module.

[0064] 4. Predictive Simulation and Multi-Objective Optimization Decision Module

[0065] Prediction Algorithm: Using prediction algorithms such as time series analysis, Markov chains, or Long Short-Term Memory (LSTM) networks, based on historical data and real-time states, the algorithm infers the chain effect of deviations on the future four-dimensional state of QSPC.

[0066] Time series analysis: By analyzing historical data, time series models are established to predict future construction status. A commonly used time series model is the Autoregressive Integral Moving Average (ARIMA) model, whose formula is:

[0067]

[0068] in, This is time series data, where B is the lag operator, ϕ(B) is the autoregressive polynomial, θ(B) is the moving average polynomial, d is the difference order, and ϵ... t It's white noise.

[0069] Markov chains: Assuming that the transitions between construction states satisfy the Markov property, meaning that future states depend only on the current state. Future construction states are predicted by calculating the state transition probability matrix.

[0070] Long Short-Term Memory (LSTM) networks are a special type of recurrent neural network capable of processing long sequences of data. Their input gates... Forgotten Gate Output gate and cell state The calculation formula is as follows:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077] Where, x t It is the input vector, h t -1 is the hidden state at the previous time step, W is the weight matrix, b is the bias vector, σ is the sigmoid function, tanh is the hyperbolic tangent function, and ⊙ is the element-wise multiplication.

[0078] Multi-objective collaborative optimization algorithms: Genetic algorithms (GA), particle swarm optimization (PSO), or non-dominated sorting genetic algorithms (NSGA-II) are used to seek Pareto optimal solutions among quality, safety, schedule, and cost.

[0079] Genetic Algorithm (GA): This algorithm performs optimization searches by simulating the biological evolutionary process. Its basic steps include encoding, population initialization, selection, crossover, and mutation. The fitness function F can be defined as:

[0080] Where Q is the quality score, S is the safety risk index, P is the schedule indicator, and C is the cost indicator. It is a weighting coefficient, and .

[0081] Particle Swarm Optimization (PSO) algorithm: This algorithm searches for an optimal solution by simulating the collective behavior of flocks of birds or schools of fish. Each particle represents a solution, and the optimal solution is found by updating the particle's velocity and position. The particle velocity update formula is:

[0082]

[0083] The particle position update formula is:

[0084]

[0085] in, It is the velocity of particle i at time t. It is the position of particle i at time t. It is the optimal position of particle i. It is the globally optimal position, and w is the inertia weight. It is a learning factor. It is a random number.

[0086] Non-dominated sorting genetic algorithm (NSGA-II): It sorts and selects individuals in the population through fast non-dominated sorting and crowding distance calculation to find the Pareto optimal solution set.

[0087] 5. Closed-loop control execution module

[0088] Command parsing and issuance: The dynamic optimization control strategy is parsed into executable digital commands, such as the command to adjust the hoisting sequence of components sent to the tower crane automatic scheduling system, the command to adjust the path and time sent to the material delivery robot, or the updated 3D construction briefing and process checkpoints pushed to workers wearing AR glasses, and these commands are issued to the automated equipment, management personnel or operators' smart terminals at the physical construction site.

[0089] Execution monitoring and feedback: The system continuously monitors the execution status and effect of instructions, and uses the feedback data as the input for the next round of data, forming a dynamic control cycle of continuous iteration and self-correction.

[0090] 6. Detailed control methods and procedures

[0091] Data Acquisition: Real-time acquisition of physical construction site data and virtual project plan data through multi-source sensors and information systems.

[0092] Digital twin model construction and synchronization: Construct and synchronize in real time a digital twin model that reflects all elements of the construction site, including people, machines, materials, methods, and environment.

[0093] Deviation identification: By integrating the collected data, the system calculates and evaluates the real-time four-dimensional status of the construction process in terms of quality, safety, schedule, and cost (QSPC), and compares it with a preset dynamic benchmark to identify construction deviations.

[0094] Predictive analytics and optimization decision making:

[0095] Once the deviation is identified, the predictive model is invoked to deduce the future development trend of the deviation without intervention and its comprehensive impact on various dimensions of QSPC.

[0096] Trigger a multi-objective collaborative optimization process that aims to minimize quality deductions, safety risks, schedule delays, and cost overruns, and searches for and generates the optimal dynamic adjustment scheme under constraints.

[0097] Closed-loop control: The optimal dynamic adjustment scheme is transformed into specific, executable electronic instructions and sent to the corresponding execution units at the construction site to drive the adjustment of the physical construction process in reverse.

[0098] Knowledge Accumulation and Model Iteration: The identified construction deviations, adjustment schemes generated in predictive analysis and optimization decision-making, and the results executed in closed-loop control steps are structured and stored to form a case knowledge base. This knowledge base is then used to periodically or triggerically retrain the predictive models initiated after the deviations are identified and the optimization models in the multi-objective collaborative optimization process to improve their accuracy and effectiveness.

[0099] Furthermore, the detailed description of the intelligent construction dynamic control system based on digital twins described in this application is as follows:

[0100] 1. Multi-module collaborative operation

[0101] The system consists of a data acquisition module, a digital twin model module, a state fusion and deviation analysis module, a predictive simulation and multi-objective optimization decision-making module, and a closed-loop control execution module. These modules work closely together to form an organic whole.

[0102] The data acquisition module obtains heterogeneous data from multiple sources, including the physical construction site and the virtual project plan, providing foundational data support for subsequent modules. For example, it collects real-time data from the construction site using component position sensors and environmental temperature and humidity sensors, and retrieves virtual project plan data, such as construction schedules and cost budgets, from the project management information system.

[0103] The digital twin model module constructs a digital twin that is synchronized with the physical construction site in real time. It not only includes geometric, schedule, and cost information, but also deeply integrates resource status, environmental data, and process logic, forming a high-fidelity, multi-dimensional virtual-physical symbiosis. By updating the model in real time, it ensures consistency with the physical construction site.

[0104] The state fusion and deviation analysis module integrates multi-source heterogeneous data to form a four-dimensional real-time state of Quality-Safety-Schedule-Cost (QSPC), which is then compared with a preset baseline state to identify construction deviations. Simultaneously, the model's self-evolutionary unit records deviation characteristics, optimization strategies, and execution effects, used for iterative optimization of the algorithm model in subsequent modules.

[0105] After identifying the deviation, the predictive simulation and multi-objective optimization decision module uses a predictive algorithm to deduce the chain effect of the deviation on the future four-dimensional state of QSPC, and then initiates a multi-objective collaborative optimization algorithm to generate a dynamic optimization control strategy.

[0106] The closed-loop control execution module parses the optimization strategy into executable digital instructions, sends them to the execution units at the construction site, and monitors the execution status and effects to form a closed-loop control system.

[0107] 2. Closed-loop dynamic control

[0108] Traditional systems often involve unidirectional data flow, making it difficult to apply virtual model analysis and decision-making results to physical construction execution. This system, however, utilizes a closed-loop control execution module to achieve a closed-loop management system from data perception to physical execution. The system continuously monitors the execution status and effects of instructions, using feedback data as input for the next round of data processing, forming a continuously iterating and self-correcting dynamic management cycle. This ensures that the construction process is always executed according to optimized strategies, thereby improving the level of intelligent construction management.

[0109] Furthermore, the predictive simulation and multi-objective optimization decision module employs various prediction algorithms, such as time series analysis, Markov chains, and Long Short-Term Memory (LSTM) networks.

[0110] Time series analysis uses historical data to build time series models, such as the Autoregressive Integrated Moving Average (ARIMA) model, to predict future construction status. The formula is: , among which, This is time series data, where B is the lag operator, ϕ(B) is the autoregressive polynomial, θ(B) is the moving average polynomial, d is the difference order, and ϵ... t It's white noise.

[0111] Markov chains assume that the construction state transitions satisfy the Markov property, and predict future construction states by calculating the state transition probability matrix.

[0112] LSTM is a special type of recurrent neural network capable of processing long sequences of data. While the calculation formulas for its input gate, forget gate, output gate, and cell state are complex, they effectively capture long-term dependencies in the data.

[0113] Furthermore, multi-objective collaborative optimization can be performed using genetic algorithms (GA), particle swarm optimization (PSO), or non-dominated sorting genetic algorithms (NSGA-II).

[0114] Genetic algorithms perform optimization searches by simulating the process of biological evolution. Fitness function.

[0115] Where Q is the quality score, S is the safety risk index, P is the schedule indicator, and C is the cost indicator. It is a weighting coefficient, and .

[0116] Particle swarm optimization (PSO) algorithms find the optimal solution by simulating the collective behavior of flocks of birds or schools of fish, updating particle velocities and positions. The velocity update formula is as follows:

[0117] The position update formula is: .

[0118] The non-dominated sorting genetic algorithm sorts and selects individuals in the population through fast non-dominated sorting and crowding distance calculation to find the Pareto optimal solution set.

[0119] Furthermore, the specific details of the intelligent construction dynamic control method based on digital twins described in this application are as follows:

[0120] 1. Proactive forecasting and forward-looking management

[0121] Traditional intelligent construction systems are mostly reactive in their management. This system, through predictive simulation and multi-objective optimization decision-making modules, proactively extrapolates the cascading effects of construction deviations on the future four-dimensional state of QSPC (Quality, Schedule, and Construction). For example, it can predict in advance the probability of construction delays, cost overruns, or the incidence of safety accidents, issuing early warnings to give managers time to take countermeasures, achieving proactive management and preventing problems from escalating and causing losses.

[0122] 2. Scientifically Quantitative Decision Support

[0123] Current technology decisions rely on the personal experience of project managers and lack scientific quantitative support. This system, after identifying deviations, initiates a multi-objective collaborative optimization algorithm to seek a Pareto optimal solution among quality, safety, schedule, and cost. The algorithm's objective function integrates quantitative assessments of quality scores, safety risk indices, total project duration, and total cost, automatically generating dynamic optimization control strategies that include resource reallocation, process reordering, or technological adjustments. This provides a scientific basis for decision-making, improving its accuracy and reliability.

[0124] 3. Comprehensive trade-offs among multiple objectives

[0125] Some existing systems focus only on a single objective, neglecting the comprehensive trade-off between multiple core elements. This system fully considers multiple aspects such as quality, safety, schedule, and cost, and achieves balanced optimization of multiple objectives while satisfying various constraints through a multi-objective collaborative optimization algorithm. For example, when rushing to meet deadlines, it comprehensively considers both increased costs and enhanced safety risks, avoiding serious problems in other aspects caused by pursuing a single objective, and improving the overall efficiency of the project.

[0126] Furthermore, the model self-evolution unit in the state fusion and deviation analysis module of this application records the characteristics of each deviation, the generated optimized control strategy, and the final execution effect. These records are used as training data to continuously iterate and optimize the prediction algorithm and optimization algorithm model in the predictive simulation and multi-objective optimization decision module, so that the model can continuously adapt to new situations and improve the accuracy of prediction and optimization.

[0127] Furthermore, this application, through knowledge accumulation and model iteration, structurally stores construction deviations, adjustment plans, and execution results to form a case knowledge base. The prediction and optimization models are retrained periodically or triggered regularly, utilizing data from the knowledge base to continuously improve the accuracy and effectiveness of the models, enabling the system to continuously optimize over long-term operation.

[0128] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A dynamic control system for intelligent construction based on digital twins, characterized by: include: The data acquisition module is used to acquire multi-source heterogeneous data from physical construction sites and virtual project plans; The digital twin model module is used to build a digital twin that is synchronized with the physical construction site in real time. The state fusion and deviation analysis module is used to fuse the multi-source heterogeneous data to form a four-dimensional real-time state reflecting the quality, safety, progress and cost of the construction site, and compare it with the preset benchmark state to identify immediate or potential construction deviations. The predictive simulation and multi-objective optimization decision module is characterized in that the module is configured to: when a deviation is identified, based on historical data and real-time status, use a predictive algorithm to deduce the chain effect of the deviation on the future four-dimensional status; and start a multi-objective collaborative optimization algorithm to seek the Pareto optimal solution among quality, safety, schedule and cost, and automatically generate a dynamic optimization control strategy that includes resource reallocation, process reordering or process adjustment. The closed-loop control execution module is characterized in that it is used to parse the dynamic optimization control strategy into executable digital instructions and send them to the intelligent terminals of automated equipment, managers or operators at the physical construction site to drive physical construction activities to be executed according to the optimized strategy, thereby forming a control closed loop from data perception to physical execution.

2. The intelligent construction dynamic control system based on digital twins according to claim 1, characterized in that: The state fusion and deviation analysis module further includes a model self-evolution unit, which records the characteristics of each deviation, the generated optimization control strategy, and the final execution effect, and uses these records as training data to continuously iterate and optimize the prediction algorithm and optimization algorithm model in the predictive simulation and multi-objective optimization decision module.

3. The intelligent construction dynamic control system based on digital twin according to claim 1, characterized in that: The predictive simulation and multi-objective optimization decision-making module employs prediction algorithms including, but not limited to, time series analysis, Markov chains, or long short-term memory networks to predict the probability of construction delays, cost overruns, or the incidence of safety accidents.

4. The intelligent construction dynamic control system based on digital twin according to claim 1, characterized in that: The predictive simulation and multi-objective optimization decision-making module employs a multi-objective collaborative optimization algorithm, such as a genetic algorithm, a particle swarm optimization algorithm, or a non-dominated sorting genetic algorithm. Its objective function integrates quantitative assessments of quality scores, safety risk indices, total project duration, and total cost.

5. The intelligent construction dynamic control system based on digital twin according to claim 1, characterized in that: The executable digital instructions issued by the closed-loop control execution module include: component hoisting sequence adjustment instructions sent to the tower crane automatic scheduling system, path and time adjustment instructions sent to the material delivery robot, or updated 3D construction briefings and process inspection points pushed to workers wearing AR glasses.

6. The intelligent construction dynamic control system based on digital twin according to claim 1, characterized in that: The digital twin model module constructs a digital twin that not only includes geometric information, schedule information, and cost information, but also deeply integrates the resource status, environmental data, and process logic of the construction site, forming a high-fidelity, multi-dimensional virtual-real symbiosis.

7. A control method for a digital twin-based intelligent construction dynamic control system according to any one of claims 1 to 6, characterized in that... Includes the following steps: Real-time data collection of physical construction site data and virtual project plan data is achieved through multi-source sensors and information systems. Construct and synchronize in real time a digital twin model that reflects all elements of the construction site, including people, machinery, materials, methods, and environment; By integrating the data, the real-time status of the construction process in four dimensions—quality, safety, schedule, and cost—is calculated and evaluated, and compared with a preset dynamic benchmark to identify construction deviations. The implementation of closed-loop control is characterized by the step of converting the optimal dynamic adjustment scheme into specific, executable electronic instructions and issuing them to the corresponding execution units at the construction site to drive the adjustment of the physical construction process in reverse.

8. The control method of the intelligent construction dynamic control system based on digital twins according to claim 7, characterized in that, The predictive analysis and optimization decision-making process specifically includes the following steps: Once the deviation is identified, the predictive model is invoked to deduce the future development trend of the deviation without intervention and its comprehensive impact on various dimensions of QSPC. Trigger a multi-objective collaborative optimization process that aims to minimize quality deductions, safety risks, schedule delays, and cost overruns, and searches for and generates the optimal dynamic adjustment scheme under constraints.

9. The control method of the intelligent construction dynamic control system based on digital twins according to claim 8, characterized in that, Following predictive analysis and optimization decision-making, the process also includes: Knowledge accumulation and model iteration involve the structured storage of identified construction deviations, adjustment schemes generated in predictive analysis and optimization decisions, and the results executed in closed-loop control steps, forming a case knowledge base. This knowledge base is then used to periodically or triggerically retrain the predictive model initiated after the deviation is identified and the optimization model that triggers a multi-objective collaborative optimization process, thereby improving their accuracy and effectiveness.

10. The control method of the intelligent construction dynamic control system based on digital twins according to claim 7, characterized in that: The dynamic benchmark is generated dynamically based on the initial plan and combined with the actual data of the completed construction parts, making the deviation analysis more timely and accurate. After an electronic command is issued, the system will continuously monitor the execution status and effect of the command, and use the feedback data as the input for the next round of data, forming a dynamic control cycle of continuous iteration and self-correction.

Images