BIM-based traffic engineering construction progress and resource management optimization system

Abstract

The invention discloses a BIM-based traffic engineering construction progress and resource management optimization system, and the system comprises a three-dimensional cognitive interaction body which adjusts the construction progress and resource configuration in real time, and eliminates the physical difference between a virtual model and an actual environment; the self-adaptive calculation engine is used for realizing butt joint between the virtual model and an actual construction site by utilizing real-time sensing data and a cognitive correction mechanism; the progress prediction simulation module is used for realizing dynamic prediction and optimization of the construction progress through deep learning and a self-adaptive feedback mechanism; the virtual-reality synchronous perception grid is constructed based on the obtained progress dynamic prediction and optimization result, and a virtual model and a construction site are synchronized and dynamically corrected; and the self-optimization module is used for realizing self-adaptive optimization and long-period evolution through a genetic algorithm and multi-objective optimization by utilizing a self-optimization and ecological optimization mechanism. According to the method, the problem of virtual reality distortion is solved, and the difference between virtuality and reality in the construction process is accurately corrected.

Description

Technical Field

[0001] The present invention relates to the technical field of BIM optimization, and in particular to a BIM-based transportation engineering construction progress and resource management optimization system. Background Art

[0002] With the rapid development of information technology, Building Information Modeling (BIM) has become one of the key technologies in construction and infrastructure project management. BIM not only provides a three-dimensional visual model of each stage of the project, but also integrates a variety of data in the project, including time schedule, cost, resource allocation, etc., to help the project team make more efficient and accurate decisions in design, construction, operation and other links. In the field of transportation engineering, the application of BIM technology is gradually changing the traditional construction management model, especially in the optimization of construction schedule and resource management, showing great potential.

[0003] However, transportation engineering construction has complex task characteristics and a dynamic external environment. Traditional BIM technology faces many challenges in dealing with construction progress and resource management. For example, transportation engineering involves many construction tasks with long cycles, a complex and changeable construction environment, and a wide variety of resources with close dependencies. In addition, the construction process may be disturbed by factors such as weather changes and emergencies, which may lead to deviations in construction progress and resources, bringing additional difficulties to project management. In order to meet these challenges, more and more research has begun to focus on how to improve the accuracy and efficiency of construction and reduce construction delays and cost overruns through BIM-based construction progress prediction and resource optimization management.

[0004] In this process, the combination of BIM and other advanced technologies, such as the Internet of Things (IoT), big data analysis, artificial intelligence (AI) and virtual reality (VR), provides a more intelligent and data-driven solution. For example, BIM models based on real-time data can help managers accurately understand the real-time status of the construction site and automatically adjust resource allocation and schedule. With the support of virtual reality technology, the construction team can rehearse the construction process in a virtual environment, conduct risk assessment and optimization adjustments, and reduce errors and uncontrollable factors in actual construction.

[0005] However, in the process of combining virtual reality technology with BIM, especially in the construction of large-scale transportation projects, there is a prominent problem, namely the virtual reality distortion problem. Specifically, with the gradual deepening of BIM models, more and more physical behaviors (such as mechanics, fluid mechanics, thermodynamics, etc.) are incorporated into the simulation system in order to more accurately simulate the real conditions of the construction site. However, in this process, there may be significant differences between the physical behaviors in the virtual model and the physical conditions in the actual construction. For example, the construction process may be affected by a variety of factors such as ambient temperature and humidity, soil characteristics, and construction equipment performance. These factors have complex effects on construction progress and resource allocation, and simple geometric models cannot effectively simulate these factors.

[0006] Current simulation engines mainly rely on geometric data and engineering design parameters for calculations, often ignoring the deep integration of physical states and dynamic environmental changes, resulting in deviations between simulation results and the actual construction environment. This difference is particularly evident in large-scale, multi-level, and multi-factor construction environments. This problem not only affects the accurate prediction of construction progress, but also makes it difficult to fully verify and optimize resource management solutions in actual applications. Therefore, how to accurately simulate and correct these differences without sacrificing computing performance has become a core issue that needs to be solved in the current deep integration of BIM technology and virtual reality technology.

[0007] This problem challenges the limits of existing simulation engines, requiring that in complex construction environments, virtual models should not only accurately present the geometric form of buildings, but also simulate the changes in various physical behaviors during the construction process, and highly align these changes with the actual conditions on the construction site. This places extremely high demands on the system's computing performance, data processing capabilities, and real-time update mechanisms. Therefore, exploring a new optimization method to solve the physical differences between virtual and reality has become a major technical challenge in the application of BIM in transportation engineering. Summary of the invention

[0008] In order to solve the problem of virtual reality distortion, the present invention provides a BIM-based transportation engineering construction progress and resource management optimization system.

[0009] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0010] The three-dimensional cognitive interactive body adjusts the construction progress and resource allocation in real time through the adaptive feedback of the perception layer, cognitive decision-making layer, collaborative decision-making mechanism and resource scheduling, eliminating the physical differences between the virtual model and the actual environment;

[0011] Adaptive computing engine, using real-time perception data and cognitive correction mechanism to achieve the connection between virtual model and actual construction site;

[0012] The progress prediction simulation module realizes dynamic prediction and optimization of construction progress through deep learning and adaptive feedback mechanism, providing decision support for progress management;

[0013] Virtual-reality synchronous perception grid, based on the obtained progress dynamic prediction and optimization results, the internship virtual model and construction site synchronization and dynamic correction;

[0014] Self-optimization module: Utilize self-optimization and ecological optimization mechanisms, through genetic algorithms and multi-objective optimization, to achieve adaptive optimization and long-term evolution.

[0015] Further: the three-dimensional cognitive interaction body includes:

[0016] The perception layer combines the environmental data collected by multiple sensors to build an instant three-dimensional state model of the construction site. Through sensor input, the actual situation of the construction site is fed back to the three-dimensional cognitive interaction body in real time to form a real-time perception data stream.

[0017] The cognitive decision-making layer introduces nonlinear optimization and feedback control to dynamically adjust the execution path of the construction task. When the construction progress deviates from the expectation, the nonlinear least squares algorithm is used to make corrections and calculate the optimal path adjustment. The optimization goal is to minimize the time and resource consumption of the construction task:

[0018]

[0019] Among them, P represents the optimization path, f i (P) is the task execution function on the path, d i is the desired target state;

[0020] Through multiple rounds of cognitive feedback on the construction process, the three-dimensional cognitive interaction body can adjust itself. Each time a construction task is executed, a feedback signal is returned to the cognitive decision-making layer. Then reinforcement learning is introduced. Reinforcement learning enables each correction to further optimize its decision-making ability based on historical feedback. Finally, the probability of each decision is updated through the Bayesian update rule, and the optimal action plan is derived in real time.

[0021] The collaborative decision-making mechanism introduces a game theory model to simulate the strategy selection of multiple three-dimensional cognitive interactions in resource competition and coordinate the behaviors among the three-dimensional cognitive interactions.

[0022] Adaptive feedback of resource scheduling is expressed as a multi-objective optimization problem, where the optimization goal is to minimize resource consumption and construction time:

[0023]

[0024] Among them, r is the resource allocation scheme, T i(r) is the time consumption of the ith task, C i (r) is the consumption of the jth resource, and λ1 and λ2 are weighting factors.

[0025] Further: the adaptive computing engine includes:

[0026] Real-time perception data processing module: extracts meaningful environmental features and change patterns from the data obtained from the perception layer to provide input for subsequent cognitive correction;

[0027] Cognitive correction decision module: Based on perception data and historical behavior patterns, an adaptive model correction method is used to correct the physical behavior in the virtual model to ensure a better match with the actual environment of the construction site;

[0028] Correction feedback and optimization module: Provides real-time feedback on the effect of model correction and uses optimization algorithms to update correction strategies to gradually improve the consistency between virtual and reality.

[0029] Further: the progress prediction simulation module includes:

[0030] The neural network model inspired by the brain has an input layer that includes real-time construction data and historical construction progress data, an output layer that predicts future construction progress, and an intermediate layer that processes and calculates progress decisions through a multi-layer neural network.

[0031] Input layer: real-time data D(t) = {d1(t), d2(t), ..., d n (t)}, where d i (t) represents the i-th data collected at time t;

[0032] Middle layer: Calculates the dependencies between tasks through the synaptic connections of neurons and adjusts the progress prediction based on historical progress data and real-time feedback;

[0033] Output layer: Calculate the predicted value of future construction progress through the activation function of the neural network;

[0034]

[0035] Among them, P pred (t) is the construction progress predicted at time t, ω i is the weight of the input data, σ is the activation function;

[0036] By strengthening the relevant signal, the connection weights between neurons are updated. Assuming that there is a certain synergistic relationship between the input and output of neurons i and j, that is, when the output of neuron i activates neuron j, the synaptic connection weight should increase. The weight update formula is as follows:

[0037] Δωij (t) = η·(D i (t)·P j (t))

[0038] Among them, ω ij is the synaptic connection weight from neuron i to neuron j, η is the learning rate, and D i (t) is the i-th input data P j (t) is the activation value of the jth output;

[0039] In order to further improve the accuracy of progress prediction, Q learning method is combined for dynamic adjustment to maximize the expected reward in the future:

[0040]

[0041] Among them, Q t (a,s) is the value function for an action a and state s at time t, α is the learning rate, r t is the immediate reward, γ is the discount factor;

[0042] Through the progress forecast error of each time period and the feedback from the construction site, the specific correction process is as follows:

[0043] Prediction error calculation:

[0044] ∈(t)=|P actual (t)-P pred (t)|

[0045] Among them, P actual (t) is the actual construction progress, P pred (t) is the predicted construction progress, ∈(t) is the prediction error;

[0046] Based on the prediction error ∈(t), the weight parameters in the neural network are updated by the gradient descent method, thereby adjusting the progress prediction model;

[0047]

[0048] Among them, Δω i (t) is the weight adjusted at time t, η is the learning rate, is the gradient of the prediction error with respect to the weight;

[0049] Through the multi-task learning mechanism, the network can simultaneously predict the progress of multiple construction tasks and learn the interdependence between tasks.

[0050] Further: the virtual-reality synchronous perception grid includes:

[0051] Virtual-reality data mapping module: responsible for mapping and synchronously updating the physical parameters in the virtual model with the perception data in the actual construction site. It needs to dynamically adjust the state of the virtual model according to the on-site sensor data and historical construction progress data.

[0052] Real-time data acquisition and processing module: Based on the sensor network at the construction site, it collects multi-modal data at the site in real time, and converts the sensor network data into digital signals that can be connected to the virtual model through data fusion algorithms;

[0053] Synchronous control and feedback module: controls the update of the virtual model according to the difference between the virtual model and the real environment, and provides dynamic feedback on the behavior of the construction site based on the corrected virtual state.

[0054] Further: the self-optimization module includes:

[0055] The introduced self-optimization mechanism simulates the principles of natural selection and gene mutation in biological evolution, and combines the real-time feedback and learning mechanism that occurs during system operation to achieve continuous optimization of models and algorithms; the self-optimization mechanism includes:

[0056] Genetic algorithm: simulates the natural selection, crossover, and mutation processes in biological evolution, and automatically optimizes resource allocation, schedule prediction, and risk management during the construction process;

[0057] Reinforcement learning and adaptive optimization: Through interactive learning with the environment, progress prediction and resource scheduling decisions are optimized to achieve long-term performance optimization;

[0058] A self-optimization cycle based on evolutionary selection, mutation, and adaptive feedback mechanisms. After each cycle, the optimization direction is adjusted based on real-time data and historical feedback.

[0059] Each evolution step includes:

[0060] Initialize the population: Based on the existing progress prediction and resource scheduling strategy, initialize a population S(t), which contains different evolutionary entities;

[0061] Fitness evaluation: By comparing the construction target with the actual result, the fitness of each individual is evaluated. The fitness evaluation function is:

[0062]

[0063] Among them, A i (t) is the fitness of individual i at time t, P real (t) is the actual progress, P pred (t) is the predicted progress, ∈(t) is the error;

[0064] Select individuals with higher fitness for crossover operation to form the next generation solution. The crossover operation is implemented by the following formula:

[0065] S offspring (t) = β·S1(t) + (1-β)·S2(t)

[0066] Among them, S offspring (t) is the new individual generated by crossover, S1(t) and S2(t) are the two selected parent individuals, and β is the crossover coefficient;

[0067] Perform mutation operations on some individuals, randomly change their resource allocation or progress prediction strategy, introduce new mutations and find potential optimization space. The mutation process is as follows:

[0068] S mutated (t) = S(t) + δ·N(0,σ)

[0069] Among them, δ is the mutation amplitude, N(0,σ) is Gaussian noise, and the generated mutation term causes random changes in individuals;

[0070] Adjust the parameters in the genetic algorithm based on real-time construction environment feedback and historical data to ensure that individuals in each generation are more adaptable to complex construction environments;

[0071] Ecological optimization draws on the coordination and mutual assistance mechanisms of ecology in nature, and through multi-task collaborative optimization and resource sharing mechanisms, achieves overall performance optimization in large-scale and complex construction environments; through ecological optimization algorithms, finds the best collaborative solution among multi-dimensional optimization objectives.

[0072] Further: The ecological optimization algorithm is implemented by the following steps:

[0073] Assume there are m optimization goals, such as progress P progress , cost P cost , Quality P quality , each target corresponds to a cost function, and the overall ecological fitness A eco for:

[0074] A eco (t) = ω1·P progress (t)+ω2·P cost (t)+ω3·P quality (t)

[0075] Among them, ω1, ω2, ω3 are the weight coefficients of each target;

[0076] In the process of ecological optimization, the optimal compromise solution between different objectives is found through the Pareto optimal solution principle. Assume that an individual has the following performance on multiple objectives: {f1,f2,...,fk}, this embodiment finds the Pareto frontier solution through non-dominated sorting:

[0077]

[0078] In the construction task, the resource dependency and time constraints between tasks are considered so that the tasks can be collaboratively optimized. The resource allocation mechanism is modeled by the following formula:

[0079]

[0080] Among them, R task (t) is the resource allocation of task j at time t, R max It is the biggest limitation of resources;

[0081] Through multiple rounds of optimization iterations, the resource allocation and progress scheduling between tasks are continuously updated to achieve the overall optimal state:

[0082] ΔR(t)=γ·(A eco (t)-A eco,old (t))

[0083] Among them, ΔR(t) is the adjustment amount of resource allocation, A eco,old (t) is the ecological fitness of the previous iteration, and γ is the adjustment coefficient.

[0084] Compared with the prior art, the present invention has the following technical advances:

[0085] First, the present invention ensures that the multi-dimensional data of the construction site can be updated and accurately fed back in real time through the deep integration of the three-dimensional cognitive interaction body and the adaptive computing engine of perception and cognitive correction, thereby minimizing the physical difference between the virtual model and the actual construction environment. Unlike the traditional BIM model that only relies on geometric data, the present invention introduces real-time physical environment perception, which can dynamically adjust factors such as environmental temperature and humidity, mechanics, and equipment load, ensuring the precise synchronization of mechanics, fluid mechanics, and thermodynamics in simulation and real construction.

[0086] Secondly, in the process of optimizing progress and resource management, the method uses a brain-inspired simulation progress prediction algorithm and a virtual-reality synchronous intelligent perception grid to model the complex physical behavior of the construction process in a simulation environment and make real-time progress adjustments and resource allocations. Through multiple adaptive mechanisms, the deviation between the physical behavior of the virtual model and the actual construction state can be quickly discovered and corrected in time, avoiding physical distortion between virtual and reality and ensuring efficient computing performance, without causing slow system operation or bottlenecks in data processing capabilities due to overly sophisticated simulations.

[0087] Finally, by combining ecological optimization and self-evolution mechanisms, the present invention can automatically optimize the progress forecast and resource allocation during the construction process, find the optimal solution for construction management through genetic algorithms and multi-objective optimization, and continuously correct the virtual model based on real-time feedback. This can not only optimize the geometric errors in the traditional BIM model, but also effectively solve the virtual reality distortion problem caused by the dynamic changes in the construction environment, thereby achieving more accurate construction progress forecasting and resource management, reducing project costs, shortening construction periods, and improving construction quality.

[0088] In summary, the present invention not only solves the problem of virtual reality distortion, but also accurately corrects the difference between virtual and reality during the construction process through an adaptive correction mechanism and an innovative optimization algorithm, thereby greatly improving the intelligence and accuracy of construction management and ensuring efficient coordination of the entire construction process and rational allocation of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0090] In the attached picture:

[0091] Figure 1 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0092] The following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0093] like Figure 1 As shown, the present invention discloses a BIM-based transportation engineering construction progress and resource management optimization system, including: a three-dimensional cognitive interactive body, which adjusts the construction progress and resource allocation in real time through the perception layer, cognitive decision layer, collaborative decision mechanism and adaptive feedback of resource scheduling, and eliminates the physical difference between the virtual model and the actual environment;

[0094] Adaptive computing engine, using real-time perception data and cognitive correction mechanism to achieve the connection between virtual model and actual construction site;

[0095] The progress prediction simulation module realizes dynamic prediction and optimization of construction progress through deep learning and adaptive feedback mechanism, providing decision support for progress management;

[0096] Virtual-reality synchronous perception grid, based on the obtained progress dynamic prediction and optimization results, the internship virtual model and construction site synchronization and dynamic correction;

[0097] Self-optimization module: Utilize self-optimization and ecological optimization mechanisms, through genetic algorithms and multi-objective optimization, to achieve adaptive optimization and long-term evolution.

[0098] Specifically, by constructing a three-dimensional cognitive interactive body, the multi-dimensional information of the transportation engineering construction site (such as environment, resources, personnel behavior, etc.) is integrated into an intelligent perception and adaptive decision-making system. The three-dimensional cognitive interactive body no longer relies on traditional hard-coded control, but dynamically adjusts the construction progress and resource allocation through real-time environmental perception, autonomous learning and collaborative decision-making, thereby solving the distortion problem between the virtual model and the actual physical state.

[0099] 1.1. Perception Layer

[0100] The 3D cognitive interactive body combines the environmental data collected by multiple sensors (such as temperature sensors, pressure sensors, displacement sensors, visual sensors, etc.) to build an instant 3D state model of the construction site. These sensors do not work alone, but work together in the perception layer through multimodal data fusion to ensure the accuracy and comprehensiveness of the perceived information. The actual situation of the construction site is fed back to the 3D cognitive interactive body in real time through sensor input, forming a real-time perception data stream.

[0101] 1.2. Conversion and preprocessing of sensory data

[0102] The raw data stream received by the perception layer needs to be processed through adaptive filtering algorithms and data fusion algorithms. The key task of this layer is to eliminate noise and extract effective information to ensure that environmental data can be accurately converted into a three-dimensional cognitive model with minimal delay.

[0103] Specifically, it includes data filtering and fusion: the Kalman filter algorithm is used to suppress sensor noise, and the weighted average method is used to combine the data of multiple sensors to obtain a more accurate three-dimensional state model.

[0104] 1.3. Cognitive Decision-Making Layer

[0105] Once the perception layer provides enough data flow, the three-dimensional cognitive interactive body corrects the virtual model according to the nonlinear changes in the real-time construction environment through an adaptive physical correction algorithm. Specifically, this part introduces nonlinear optimization and feedback control to dynamically adjust the execution path of the construction task. When the construction progress deviates from the expected progress, the nonlinear least squares algorithm is used to correct it and calculate the optimal path adjustment. The optimization goal is to minimize the time and resource consumption of the construction task:

[0106]

[0107] Among them, P represents the optimization path, f i (P) is the task execution function on the path, di is the desired target state.

[0108] Through multiple rounds of cognitive feedback on the construction process, the three-dimensional cognitive interactive body can adjust itself. Each time a construction task is executed, a feedback signal will be returned to the cognitive decision-making layer, and then reinforcement learning will be introduced. Reinforcement learning enables each correction to further optimize its decision-making ability based on historical feedback. Finally, the probability of each decision is updated through the Bayesian update rule, and the optimal action plan is derived in real time.

[0109] 1.4. Collaborative decision-making mechanism

[0110] The three-dimensional cognitive interactive body is not only responsible for the decision-making of a single task, but also needs to collaborate with other three-dimensional cognitive interactive bodies to complete the overall construction goal. In this embodiment, a game theory model is introduced to simulate the strategy selection of multiple three-dimensional cognitive interactive bodies in resource competition and coordinate the behavior between each three-dimensional cognitive interactive body.

[0111] At every point in time, each three-dimensional cognitive interaction entity needs to choose its optimal strategy to ensure efficient use of resources while avoiding conflicts with other three-dimensional cognitive interactions. The game theory model determines the optimal strategy through the following Nash equilibrium.

[0112] 1.5. Adaptive Feedback for Resource Scheduling

[0113] How to effectively schedule limited resources between multiple 3D cognitive interactive bodies is a key issue. Through multi-objective optimization algorithms, 3D cognitive interactive bodies can maximize resource utilization efficiency while ensuring construction quality. The adaptive feedback of resource scheduling can be expressed as a multi-objective optimization problem, and the optimization goal is to minimize resource consumption and construction time:

[0114]

[0115] Among them, r is the resource allocation scheme, T i (r) is the time consumption of the ith task, C i (r) is the consumption of the jth resource, and λ1 and λ2 are weighting factors.

[0116] The three-dimensional cognitive interactive body can adjust the construction progress and resource allocation in real time through the perception layer, cognitive decision layer, collaborative decision-making mechanism and adaptive feedback of resource scheduling, eliminating the physical differences between the virtual model and the actual environment. Through reinforcement learning, nonlinear optimization, and multimodal data fusion, the three-dimensional cognitive interactive body can not only deal with the common virtual distortion problems in traditional solutions, but also provide highly flexible and adaptive solutions in dynamic environments, improving the level of intelligent management in the construction process.

[0117] The perception and real-time feedback processing of on-site data have been completed through the three-dimensional cognitive interactive body. The next step is to build an adaptive computing engine for perception and cognitive correction based on these perception data, so that the three-dimensional cognitive interactive body can optimize the difference between the virtual model and the physical environment in real time according to the dynamic changes of the construction site, avoiding the distortion between virtual and real physical behaviors that occurs in traditional simulation.

[0118] The adaptive computing engine is based on the following key modules:

[0119] Real-time perception data processing module: The data obtained from the perception layer (the perception results from sensors and three-dimensional cognitive interactions) are filtered and preprocessed to extract meaningful environmental features and change patterns to provide input for subsequent cognitive corrections.

[0120] Cognitive correction decision module: Based on perception data and historical behavior patterns, an adaptive model correction method is used to correct the physical behavior in the virtual model to ensure that it better matches the actual environment of the construction site.

[0121] Correction feedback and optimization module: Provides real-time feedback on the effect of model correction and uses optimization algorithms to update correction strategies to gradually improve the consistency between virtual and reality.

[0122] The implementation of the cognitive correction decision module includes:

[0123] 2.1. Dynamic physical correction algorithm. The core idea of ​​this algorithm is to use nonlinear optimization methods to automatically adjust the physical parameters (such as mechanics, fluid mechanics, etc.) in the virtual model under the dynamic environmental changes of the construction site, so as to eliminate the differences between the virtual model and the actual environment.

[0124] 2.2. Adaptive Correction Model

[0125] Based on the field data provided by the perception layer, the correction algorithm adjusts the parameters in the above physical field model through adaptive optimization. To this end, this embodiment designs an adaptive correction model, which generates correction factors based on historical data and real-time feedback to minimize the error between the virtual model and the actual construction environment, where the error definition refers to the error E(t) between the virtual and actual physical behaviors:

[0126]

[0127] Among them, F * (x i ,t) is the physical force in the virtual model, F real (x i ,t) is the physical force measured by the field sensor, and the error E(t) represents the difference between the model and the field data.

[0128] Then define the correction factor. To reduce the error, the gradient descent method is used to calculate the correction factor:

[0129]

[0130] Among them, Δθ t The correction amount of the physical field model parameter θ at time t, η is the learning rate, is the gradient of the error E(t) to the model parameters. During the optimization process, the algorithm continuously adjusts θ (i.e., the physical parameters of the model) to minimize the error E(t), thereby achieving physical correction of the virtual model.

[0131] 2.3. Feedback and Correction Loop

[0132] In order to achieve real-time correction, the above-mentioned physical correction algorithm needs to run continuously during the construction process to update the physical parameters at each time point. The corrected virtual model will be fed back to the cognitive correction decision module as the input of the next decision cycle. This process can be expressed as a feedback optimization loop, that is, the state after each correction will enter the correction model again and be further optimized in the next round of calculation.

[0133] 2.4. Dynamic Feedback Optimization Algorithm

[0134] In order to ensure the adaptability and accuracy of the system, the optimization algorithm of the correction feedback not only relies on the physical correction error, but also needs to consider the construction resource constraints and progress goals. Therefore, this embodiment designs a multi-objective optimization algorithm so that the correction process not only optimizes the matching of physical behaviors, but also optimizes progress delays and resource waste at the same time.

[0135] Multi-objective optimization: The optimization goal is set to minimize the weighted sum of the physical correction error E(t), the construction progress deviation P(t), and the resource consumption R(t), and the following optimization function is obtained:

[0136]

[0137] Among them, λ1, λ2, λ3 are weighting factors, E(t) is the physical correction error, P(t) is the progress deviation, and R(t) is the resource consumption.

[0138] 2.5. Joint Optimization of Resources and Schedule Scheduling

[0139] While making physical corrections, the balance between resource scheduling and progress management must be considered. To this end, this embodiment introduces a resource constraint optimization algorithm to dynamically schedule resources and adjust the progress based on the corrected construction status (such as adjusted physical behavior, progress deviation, etc.).

[0140] Assuming that the construction resources r(t) are subject to some constraints (such as the number of equipment, personnel availability, etc.), the resource scheduling optimization objective can be expressed as:

[0141]

[0142] Among them, r k (t) is the scheduling amount of the kth resource at time t, r ideal (t) is the ideal resource scheduling quantity.

[0143] The implementation of the schedule forecast simulation module includes:

[0144] In the cognitive correction decision-making module, by building an adaptive computing engine for perception and cognitive correction, it is possible to correct physical behavior based on real-time data and optimize construction resources and progress allocation. The goal of the progress prediction simulation module is to further enhance the construction progress prediction capability by introducing a brain-inspired simulation progress prediction algorithm, so that it can perform accurate progress simulation and optimization in a highly dynamic construction environment.

[0145] This algorithm draws on neuroscience and brain-inspired mechanisms, and simulates the working method of the construction brain based on real-time feedback data from the construction site and historical construction progress data, so as to efficiently and accurately predict the construction progress in nonlinear and complex construction environments. It can not only handle various uncertainties in construction, but also dynamically adjust the progress prediction model to ensure its high adaptability and high accuracy.

[0146] First, a model framework of brain-inspired mechanisms is introduced.

[0147] 3.1. Core concepts of brain-inspired mechanisms

[0148] The brain-inspired mechanism is based on the way neural networks and the brain process information, especially the synaptic connections and synaptic weight updates between neurons, simulating the information flow and optimizing the decision-making process in construction progress forecasting. This mechanism simulates the correlation between various factors in the construction site (such as resource scheduling, weather changes, construction personnel efficiency, etc.) and adjusts the progress forecasting model through synaptic weight updates. The core idea is that the progress forecast of the construction process can be represented by a brain-inspired neural network. Each task and each construction stage in the construction can be regarded as a neuron, and the construction progress and resource allocation are the synaptic connections between neurons, represented by the activation value of the neuron.

[0149] 3.2. Neural network structure for simulation progress prediction

[0150] This embodiment designs a brain-inspired neural network model for the construction progress prediction problem. Its input layer consists of real-time construction data and historical construction progress data, the output layer is the future construction progress prediction, and the middle layer processes and calculates the progress decision through a multi-layer neural network.

[0151] Input layer: It consists of real-time data D(t), which includes construction resource status, site environment, weather conditions, etc.:

[0152] D(t)={d1(t),d2(t),...,d n (t)}

[0153] Among them, d i (t) represents the i-th data collected at time t, such as equipment status, personnel efficiency, etc.

[0154] Middle layer (hidden layer): The dependencies between tasks are calculated through the synaptic connections of neurons, and the progress prediction is adjusted based on historical progress data and real-time feedback.

[0155] Output layer: The future construction progress forecast is calculated through the activation function of the neural network.

[0156]

[0157] Among them, P pred (t) is the construction progress predicted at time t, ω i is the weight of the input data, and σ is the activation function (such as Sigmoid or ReLU).

[0158] 3.4. Synaptic Weight Update Rules

[0159] The core of the brain-inspired model is to simulate the update of synaptic weights between neurons, that is, how to update the various weights in the model according to the input real-time data and historical data. In order to apply this mechanism in construction progress prediction, this embodiment adopts a combination of learning rules and reinforcement learning algorithms, so that the progress prediction at each time point can be adaptively adjusted.

[0160] Learning rule: Update the connection weights between neurons by strengthening the relevant signals. Specifically, assume that there is a certain synergistic relationship between the input and output of neurons i and j, that is, when the output of neuron i activates neuron j, the synaptic connection weight should increase. The weight update formula is as follows:

[0161] Δω ij (t) = η·(D i (t)·P j (t))

[0162] Among them, ω ijis the synaptic connection weight from neuron i to neuron j, η is the learning rate, and D i (t) is the i-th input data P j (t) is the activation value of the j-th output.

[0163] Reinforcement learning update: In order to further improve the accuracy of progress prediction, this embodiment combines the Q learning method for dynamic adjustment. Q learning can automatically update the model according to the prediction error and reward mechanism, thereby maximizing the expected reward in the future (i.e., improving the accuracy of the prediction):

[0164]

[0165] Among them, Q t (a,s) is the value function for an action a and state s at time t, α is the learning rate, r t is the immediate reward and γ is the discount factor.

[0166] 3.5. Model Adaptation

[0167] The synaptic weights in the brain-inspired model will be adjusted as the construction site changes dynamically. Through the progress prediction error of each time period and the feedback from the construction site, the model can perform real-time self-correction. The specific correction process is as follows:

[0168] Prediction error calculation:

[0169] ∈(t)=|P actual (t)-P pred (t)|

[0170] Among them, P actual (t) is the actual construction progress, P red (t) is the predicted construction progress, ∈(t) is the prediction error.

[0171] Adaptive adjustment: Based on the prediction error ∈(t), the weight parameters in the neural network are updated through the gradient descent method to adjust the progress prediction model.

[0172]

[0173] Among them, Δω i (t) is the weight adjusted at time t, η is the learning rate, is the gradient of the prediction error with respect to the weights.

[0174] 3.6. Multi-task progress prediction

[0175] Since there are often multiple tasks going on simultaneously at a construction site, the optimization of progress prediction is not only the prediction of a single task, but also includes the mutual influence between tasks. This embodiment uses a multi-task learning mechanism to enable the network to perform progress prediction for multiple construction tasks at the same time and learn the interdependence between tasks.

[0176] Multi-task loss function: Set the total loss function as the weighted sum of all task losses:

[0177]

[0178] Among them, L i (t) is the loss function of the i-th task at time t, λ i is the weight coefficient of the task.

[0179] After each prediction, the progress error will adjust the network parameters through the feedback mechanism and be applied to subsequent progress predictions. Through this feedback mechanism, the accuracy of progress prediction can be gradually improved as the construction progress changes.

[0180] By introducing a brain-inspired simulation progress prediction algorithm, this embodiment can not only cope with the nonlinearity and complexity in the construction progress, but also dynamically adjust the prediction model and correct the error between the virtual model and the actual progress in real time. The brain-inspired mechanism enables the system to self-learn and self-correct like the human brain, and achieve efficient decision-making capabilities in construction progress prediction. This method provides a new and highly intelligent solution for BIM and transportation engineering construction progress management.

[0181] The implementation of the virtual-reality synchronized perception grid includes:

[0182] In the progress prediction simulation module, by introducing a simulated progress prediction algorithm based on brain-like inspiration, this embodiment realizes the dynamic prediction of the construction progress and can adapt to changes in the on-site environment. The goal is to build a virtual-reality synchronous perception grid based on the above-mentioned progress prediction and correction results. The grid can seamlessly connect the virtual model with the actual construction site in real time, and perform synchronous updates, corrections and dynamic feedback. Specifically, various parameters in the virtual model (such as progress, resource allocation, equipment status, etc.) need to be synchronized with the actual situation of the construction site (through perception data collection) to ensure the accuracy and real-time nature of construction management.

[0183] The grid must not only achieve synchronization between virtual and reality, but also ensure low latency, high accuracy of data transmission and efficient computing in large-scale construction environments. Therefore, how to process and optimize the data mapping and synchronization between virtual and reality through an efficient computing framework and intelligent perception algorithms has become the core challenge of this part.

[0184] First, the basic framework of the virtual-reality synchronized intelligent perception grid is constructed.

[0185] 4.1. Architecture design of virtual-reality synchronization

[0186] The virtual-reality synchronous perception grid consists of the following key modules:

[0187] Virtual-reality data mapping module: responsible for mapping and synchronously updating various physical parameters in the virtual model (such as construction progress, resource consumption, equipment status, etc.) with the perception data in the actual construction site. This module needs to dynamically adjust the status of the virtual model based on on-site sensor data and historical construction progress data.

[0188] Real-time data acquisition and processing module: Based on the sensor network at the construction site, including GPS, cameras, temperature and humidity sensors, equipment sensors, etc., it collects multimodal data at the site in real time, and converts these data into digital signals that can be connected to the virtual model through data fusion algorithms.

[0189] Synchronous control and feedback module: controls the update of the virtual model according to the difference between the virtual model and the real environment, and provides dynamic feedback on the behavior of the construction site based on the corrected virtual state.

[0190] 4.2. Key technologies for data synchronization and mapping

[0191] The core of synchronization and mapping technology is to make the virtual progress and the real progress consistent through accurate mapping between the virtual model and the perceived data of the construction site. The following is the specific implementation method of this process.

[0192] Then multi-scale data mapping and synchronization are performed.

[0193] 4.3. Multi-scale mapping between construction site and virtual model

[0194] To ensure data synchronization between the construction site and the virtual model, it is necessary to hierarchically map the various parameters in the virtual model with the multi-scale data on site (such as building area, equipment working status, environmental parameters, etc.). Specifically, the physical space of the virtual-reality synchronization grid is divided into multiple virtual-reality unit grids, each of which corresponds to a physical unit in the actual construction (such as a section of road, bridge, equipment area, etc.), and each data item in these physical units is synchronized through data interpolation and mapping functions.

[0195] Assume that in this embodiment there is a building area A in the virtual model virtual , the region is divided into n sub-unit grids {G1,G2,...,G n}. In the actual area A of the construction site realIn the example, the real-time data set obtained through the sensor network is D real (t), the dataset is connected to each virtual subunit grid via a multi-scale data map.

[0196] The mapping relationship can be expressed as follows:

[0197] D virtual (t) = f mapping (D real (t),G1,G2,...,G n )

[0198] Among them, f mapping is the mapping function between virtual and real data, D virtual (t) is the virtual model data updated at time t, D real (t) is the sensor data collected from the real scene at time t.

[0199] 4.5. Spatial interpolation and dynamic correction

[0200] In order to improve the mapping accuracy, spatial interpolation technology is used to align the sensor data on site with the node data in the virtual model point by point. Assume that each grid cell G i Corresponding to a virtual coordinate space and a real coordinate space, this embodiment uses three-dimensional linear interpolation or Gaussian process regression to perform space correction, so that the virtual model and the real scene are highly consistent in multi-dimensional data such as position and state.

[0201] The spatial interpolation formula is as follows:

[0202]

[0203] Among them, D i (x, y, z, t) is the data of sensor node i at time t, ω i (t) is the weight coefficient corresponding to node i, and (x, y, z) is the spatial coordinate.

[0204] Finally, dynamic synchronization and feedback mechanism are carried out.

[0205] 4.6. Virtual-Reality Progress Feedback Mechanism

[0206] There may be differences between the actual progress and the virtual progress at the construction site, so a feedback mechanism is needed to achieve dynamic correction of the progress. virtual (t) and actual progress P reak (t) When deviations occur, the system will automatically adjust the relevant parameters in the virtual model.

[0207] The progress feedback formula is:

[0208] ΔP(t)=α·(P real (t)-P virtual (t))

[0209] Wherein, ΔP(t) is the correction amount of the virtual model progress at time t, and α is the correction coefficient.

[0210] 4.7. Intelligent perception correction and synchronization control

[0211] In the sensor network at the construction site, the real-time environmental data collected (such as temperature, humidity, personnel status, etc.) may affect the physical behavior of the virtual model. Therefore, through the intelligent perception correction mechanism, the system will use deep learning algorithms to model the environmental impact during each synchronization, thereby performing dynamic correction.

[0212] By establishing a virtual-reality synchronization intelligent perception grid, this embodiment can achieve efficient synchronization and dynamic correction of the virtual model and the construction site in multiple dimensions and at all times. The grid is based on multi-scale data mapping, spatial interpolation, dynamic progress feedback and intelligent perception correction technology, which effectively eliminates the difference between virtual and reality and ensures real-time monitoring and intelligent optimization during the construction process.

[0213] The specific implementation of the self-optimization module includes:

[0214] In the virtual-reality synchronous perception grid, by establishing a virtual-reality synchronous perception grid, this embodiment realizes the precise synchronization of the virtual model and the real construction environment, and adjusts the virtual progress and resource status in real time. The goal of the self-optimization module is to further promote the self-optimization and ecological optimization of the system. Through the adaptive learning mechanism and ecological network optimization algorithm, the system can continuously self-optimize in a long-term, multi-task and complex construction environment, and intelligently evolve in a dynamic environment. The core challenge of this stage is how to enable the system to not only adapt to environmental changes in a large-scale and complex construction environment, but also automatically improve its own performance, such as optimization in resource allocation, progress scheduling and risk management.

[0215] 5.1. Self-optimization

[0216] At this stage, the self-optimization mechanism introduced in this embodiment simulates the principles of natural selection and gene mutation in biological evolution, and combines the real-time feedback and learning mechanism that occurs during system operation to achieve continuous optimization of models and algorithms. The self-optimization mechanism of the system consists of the following two parts:

[0217] Genetic algorithm: simulates the processes of natural selection, crossover, mutation, etc. in biological evolution, and automatically optimizes resource allocation, progress prediction, risk management, etc. during the construction process.

[0218] Reinforcement learning and adaptive optimization: The system learns through environmental interaction, thereby optimizing decisions such as progress prediction and resource scheduling to achieve long-term performance optimization.

[0219] 5.2. Self-optimization cycle

[0220] The process of self-optimization is based on mechanisms such as evolutionary selection, mutation, and adaptive feedback. After each cycle, the system adjusts the optimization direction based on real-time data and historical feedback.

[0221] Each evolution step includes:

[0222] 5.2.1. Initialize the population: Based on the existing progress prediction, resource scheduling and other strategies, initialize a population S(t), which contains different evolutions (i.e. different construction progress and resource allocation schemes).

[0223] 5.2.2. Fitness evaluation: By comparing the construction goals (such as time, cost, quality, etc.) with the actual results, the fitness of each individual is evaluated. The fitness evaluation function is:

[0224]

[0225] Among them, A i (t) is the fitness of individual i at time t, P real (t) is the actual progress, P pred (t) is the predicted progress and ∈(t) is the error.

[0226] 5.23. Selection and crossover: Select individuals with higher fitness for crossover operation to form the next generation of solutions. The crossover operation is implemented by the following formula:

[0227] S offspring (t) = β·S1(t) + (1-β)·S2(t)

[0228] Among them, S offspring (t) is the new individual produced by crossover, S1(t) and S2(t) are the two selected parent individuals, and β is the crossover coefficient.

[0229] 5.2.4. Mutation and adjustment: Perform mutation operations on some individuals, randomly change their resource allocation or progress prediction strategy, introduce new variations and find potential optimization space. The mutation process is as follows:

[0230] S mutated (t)=t(t)+δ·N(0,σ)

[0231] Among them, δ is the mutation amplitude, N(0,σ) is Gaussian noise, and the generated mutation term causes random changes in individuals.

[0232] 5.2.5. Adaptive feedback: According to the real-time construction environment feedback and historical data, adjust the parameters in the genetic algorithm to ensure that each generation of individuals is more adaptable to the complex construction environment.

[0233] 5.3. Ecosystem Optimization

[0234] Ecological optimization draws on the coordination and mutual assistance mechanism of the ecosystem in nature, and achieves the optimization of the overall performance of the system in a large-scale and complex construction environment through multi-task collaborative optimization and resource sharing mechanism. Through the ecological optimization algorithm, this embodiment can find the best collaborative solution in multi-dimensional optimization goals (such as progress, cost, quality, safety, etc.).

[0235] At the construction site, there is a complex resource competition and sharing relationship between multiple tasks (such as roadbed construction, bridge construction, equipment installation, etc.). How to optimize the resource allocation between different tasks so that the entire construction process can evolve co-evolved is the key to ecological optimization.

[0236] 5.4. Design of ecological optimization algorithm

[0237] This embodiment designs a multi-objective ecological optimization algorithm for this purpose, which is implemented through the following steps:

[0238] 5.4.1. Objective function definition: Assume there are m optimization objectives, such as progress P progress , cost P cost , Quality P quality , each target corresponds to a cost function, and the ecological fitness of the entire system is A eco for:

[0239] A eco (t) = ω1·P progress (t)+ω2·P cost (t)+ω3·P qualoty (t)

[0240] Among them, ω1, ω2, ω3 are the weight coefficients of each target.

[0241] 5.4.2. Multi-objective optimization: In the process of ecological optimization, the optimal compromise solution between different objectives is found through the Pareto optimal solution principle. Suppose an individual has the following performance on multiple objectives: {f1,f2,...,f k}, this embodiment finds the Pareto frontier solution through non-dominated sorting:

[0242]

[0243] 5.4.3. Task coordination mechanism: In the construction task, the resource dependency and time constraints between tasks are considered so that the tasks can be coordinated and optimized. The resource allocation mechanism is modeled by the following formula:

[0244]

[0245] Among them, R task (t) is the resource allocation of task j at time t, R max It is the biggest limitation of resources.

[0246] 5.4.4. Ecological feedback and optimization: Through multiple rounds of optimization iterations, the resource allocation and progress scheduling between tasks are continuously updated to achieve the overall optimal state:

[0247] ΔR(t)=γ·(A eco (t)-A eco,old (t))

[0248] Among them, ΔR(t) is the adjustment amount of resource allocation, A eco,old (t) is the ecological fitness of the previous iteration, and γ is the adjustment coefficient.

[0249] Through self-optimization and ecological optimization mechanisms, the system can continuously optimize decisions on construction progress, resource allocation, cost control, and other aspects according to the dynamic changes of the construction site, and promote the intelligent evolution of the system in a complex construction environment. By combining genetic algorithms and multi-objective ecological optimization, the system can not only optimize a single goal, but also find the global optimal construction management solution in the context of multi-task collaboration and resource sharing.

[0250] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of protection of the claims of the present invention.

Claims

1. The BIM-based traffic engineering construction progress and resource management optimization system is characterized by: include: The three-dimensional cognitive interactive body adjusts the construction progress and resource allocation in real time through the adaptive feedback of the perception layer, cognitive decision-making layer, collaborative decision-making mechanism and resource scheduling, eliminating the physical differences between the virtual model and the actual environment; Adaptive computing engine, using real-time perception data and cognitive correction mechanism to achieve the connection between virtual model and actual construction site; The progress prediction simulation module realizes dynamic prediction and optimization of construction progress through deep learning and adaptive feedback mechanism, providing decision support for progress management; Virtual-reality synchronous perception grid, based on the obtained progress dynamic prediction and optimization results, the internship virtual model and construction site synchronization and dynamic correction; Self-optimization module: Utilize self-optimization and ecological optimization mechanisms, through genetic algorithms and multi-objective optimization, to achieve adaptive optimization and long-term evolution.

2. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 1 is characterized in that: The three-dimensional cognitive interaction body comprises: The perception layer combines the environmental data collected by multiple sensors to build an instant three-dimensional state model of the construction site. Through sensor input, the actual situation of the construction site is fed back to the three-dimensional cognitive interaction body in real time to form a real-time perception data stream. The cognitive decision-making layer introduces nonlinear optimization and feedback control to dynamically adjust the execution path of the construction task. When the construction progress deviates from the expectation, the nonlinear least squares algorithm is used to make corrections and calculate the optimal path adjustment. The optimization goal is to minimize the time and resource consumption of the construction task: Among them, P represents the optimization path, f i (P) is the task execution function on the path, d i is the desired target state; Through multiple rounds of cognitive feedback on the construction process, the three-dimensional cognitive interaction body can adjust itself. Each time a construction task is executed, a feedback signal is returned to the cognitive decision-making layer. Then reinforcement learning is introduced. Reinforcement learning enables each correction to further optimize its decision-making ability based on historical feedback. Finally, the probability of each decision is updated through the Bayesian update rule, and the optimal action plan is derived in real time. The collaborative decision-making mechanism introduces a game theory model to simulate the strategy selection of multiple three-dimensional cognitive interactions in resource competition and coordinate the behaviors among the three-dimensional cognitive interactions. Adaptive feedback of resource scheduling is expressed as a multi-objective optimization problem, where the optimization goal is to minimize resource consumption and construction time: Among them, r is the resource allocation scheme, T i (r) is the time consumption of the ith task, C i (r) is the consumption of the jth resource, and λ1 and λ2 are weighting factors.

3. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 2 is characterized in that: The adaptive computing engine comprises: Real-time perception data processing module: extracts meaningful environmental features and change patterns from the data obtained from the perception layer to provide input for subsequent cognitive correction; Cognitive correction decision module: Based on perception data and historical behavior patterns, an adaptive model correction method is used to correct the physical behavior in the virtual model to ensure a better match with the actual environment of the construction site; Correction feedback and optimization module: Provides real-time feedback on the effect of model correction and uses optimization algorithms to update correction strategies to gradually improve the consistency between virtual and reality.

4. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 3 is characterized in that: The progress prediction simulation module includes: The neural network model inspired by the brain has an input layer that includes real-time construction data and historical construction progress data, an output layer that predicts future construction progress, and an intermediate layer that processes and calculates progress decisions through a multi-layer neural network. Input layer: real-time data D(t) = {d1(t), d2(t), ..., d n (t)}, where d i (t) represents the i-th data collected at time t; Middle layer: Calculates the dependencies between tasks through the synaptic connections of neurons and adjusts the progress prediction based on historical progress data and real-time feedback; Output layer: Calculate the predicted value of future construction progress through the activation function of the neural network; Among them, P pred (t) is the construction progress predicted at time t, ω i is the weight of the input data, σ is the activation function; By strengthening the relevant signal, the connection weights between neurons are updated. Assuming that there is a certain synergistic relationship between the input and output of neurons i and j, that is, when the output of neuron i activates neuron j, the synaptic connection weight should increase. The weight update formula is as follows: Give ij (t)=η·(D i (t)·P j (t)) Among them, ω ij is the synaptic connection weight from neuron i to neuron j, η is the learning rate, and D i (t) is the i-th input data P j (t) is the activation value of the jth output; In order to further improve the accuracy of progress prediction, Q learning method is combined for dynamic adjustment to maximize the expected reward in the future: Among them, Q t (a,s) is the value function for an action a and state s at time t, α is the learning rate, r t is the immediate reward, γ is the discount factor; Through the progress forecast error of each time period and the feedback from the construction site, the specific correction process is as follows: Prediction error calculation: ∈(t)=|P actual (t)-P pred (t)| Among them, P actual (t) is the actual construction progress, P pred (t) is the predicted construction progress, ∈(t) is the prediction error; Based on the prediction error ∈(t), the weight parameters in the neural network are updated by the gradient descent method, thereby adjusting the progress prediction model; Among them, Δω i (t) is the weight adjusted at time t, η is the learning rate, is the gradient of the prediction error with respect to the weight; Through the multi-task learning mechanism, the network can simultaneously predict the progress of multiple construction tasks and learn the interdependence between tasks.

5. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 4 is characterized in that: The virtual-reality synchronous perception grid includes: Virtual-reality data mapping module: responsible for mapping and synchronously updating the physical parameters in the virtual model with the perception data in the actual construction site. It needs to dynamically adjust the state of the virtual model according to the on-site sensor data and historical construction progress data. Real-time data acquisition and processing module: Based on the sensor network at the construction site, it collects multi-modal data at the site in real time, and converts the sensor network data into digital signals that can be connected to the virtual model through data fusion algorithms; Synchronous control and feedback module: controls the update of the virtual model according to the difference between the virtual model and the real environment, and provides dynamic feedback on the behavior of the construction site based on the corrected virtual state.

6. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 5 is characterized in that: The self-optimization module comprises: The introduced self-optimization mechanism simulates the principles of natural selection and gene mutation in biological evolution, and combines the real-time feedback and learning mechanism that occurs during system operation to achieve continuous optimization of models and algorithms; the self-optimization mechanism includes: Genetic algorithm: simulates the natural selection, crossover, and mutation processes in biological evolution, and automatically optimizes resource allocation, schedule prediction, and risk management during the construction process; Reinforcement learning and adaptive optimization: Through interactive learning with the environment, progress prediction and resource scheduling decisions are optimized to achieve long-term performance optimization; A self-optimization cycle based on evolutionary selection, mutation, and adaptive feedback mechanisms. After each cycle, the optimization direction is adjusted based on real-time data and historical feedback. Each evolution step includes: Initialize the population: Based on the existing progress prediction and resource scheduling strategy, initialize a population S(t), which contains different evolutionary entities; Fitness evaluation: By comparing the construction target with the actual result, the fitness of each individual is evaluated. The fitness evaluation function is: Among them, A i (t) is the fitness of individual i at time t, P real (t) is the actual progress, P pred (t) is the predicted progress, ∈(t) is the error; Select individuals with higher fitness for crossover operation to form the next generation solution. The crossover operation is implemented by the following formula: S offspring (t)=β·S1(t)+(1-β)·S2(t) Among them, S offspring (t) is the new individual generated by crossover, S1(t) and S2(t) are the two selected parent individuals, and β is the crossover coefficient; Perform mutation operations on some individuals, randomly change their resource allocation or progress prediction strategy, introduce new mutations and find potential optimization space. The mutation process is as follows: S mutated (t)=S(t)+δ·N(0,σ) Among them, δ is the mutation amplitude, N(0,σ) is Gaussian noise, and the generated mutation term causes random changes in individuals; Adjust the parameters in the genetic algorithm based on real-time construction environment feedback and historical data to ensure that individuals in each generation are more adaptable to complex construction environments; Ecological optimization draws on the coordination and mutual assistance mechanisms of ecology in nature, and through multi-task collaborative optimization and resource sharing mechanisms, achieves overall performance optimization in large-scale and complex construction environments; through ecological optimization algorithms, finds the best collaborative solution among multi-dimensional optimization objectives.

7. The BIM-based traffic engineering construction progress and resource management optimization system according to claim 6 is characterized in that: The ecological optimization algorithm is implemented by the following steps: Assume there are m optimization goals, such as progress P progress , cost P cost , Quality P quality , each target corresponds to a cost function, and the overall ecological fitness A eco for: A eco (t)=ω1·P progress (t)+ω2·P cost (t)+ω3·P quality (t) Among them, ω1, ω2, ω3 are the weight coefficients of each target; In the process of ecological optimization, the optimal compromise solution between different objectives is found through the Pareto optimal solution principle. Assume that an individual has the following performance on multiple objectives: {f1,f2,...,f k }, this embodiment finds the Pareto frontier solution through non-dominated sorting: In the construction task, the resource dependency and time constraints between tasks are considered so that the tasks can be collaboratively optimized. The resource allocation mechanism is modeled by the following formula: Among them, R task (t) is the resource allocation of task j at time t, R max It is the biggest limitation of resources; Through multiple rounds of optimization iterations, the resource allocation and progress scheduling between tasks are continuously updated to achieve the overall optimal state: ΔR(t)=γ·(A eco (t)-A eco,old (t)) Among them, ΔR(t) is the adjustment amount of resource allocation, A eco,old (t) is the ecological fitness of the previous iteration, and γ is the adjustment coefficient.

Images