Planning and scheduling system for optimizing road and bridge construction progress by utilizing artificial intelligence

Abstract

The invention discloses a planning and scheduling system for optimizing road and bridge construction progress by using artificial intelligence, and relates to the technical field of road and bridge construction. The system comprises a data interaction system module, and the data interaction system module is bidirectionally and electrically connected with a model collaboration mechanism module. Through multi-source data fusion and model collaboration, the construction progress prediction precision is improved by 30%, resource waste is reduced by 20%, in the aspect of risk prevention and control, the geological disaster early warning advance rate reaches 85% through a space-time risk model, the social risk occurrence rate is reduced by 89% through public opinion monitoring, resource scheduling can be adjusted in real time through reinforcement learning and an evolutionary strategy, and the construction progress prediction efficiency is improved. The emergency risk response time is shortened by 50%, the block chain storage ensures that data cannot be tampered, the intelligent contract improves the multi-party collaboration transparency, and the construction management level is comprehensively improved. The problem that a road and bridge construction progress planning and scheduling system in the prior art does not dynamically adjust resource allocation in combination with risk factors and does not cover collaborative analysis of multi-source data is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of road and bridge construction, and particularly to an artificial intelligence-based system for optimizing the construction schedule planning and scheduling of road and bridges. Background Art

[0002] In the context of the booming development of current road and bridge construction, the scale and complexity of construction are constantly increasing, posing stringent requirements for the efficiency, accuracy, and intelligence of the construction schedule planning and scheduling system. However, many problems that urgently need to be solved have emerged in the current road and bridge construction management methods.

[0003] From the data perspective, various types of data generated during the construction process, such as construction progress, resource status, environmental risks, and public opinion feedback, are stored and managed in a decentralized manner. The data involved in different departments and different business processes often act independently, lacking a unified integration and coordination mechanism. This makes it difficult to conduct a comprehensive analysis of the construction process from a global perspective, unable to fully explore the potential correlations between multi-source data, and thus affecting the scientificity and comprehensiveness of construction decisions.

[0004] For example, a patent for an intelligent management method for highway bridge construction (CN118396573A) predicts the progress through construction speed. However, this method only focuses on a single factor of construction speed, does not fully consider the risk factors during the construction process, and is even less able to dynamically adjust resource allocation according to risk changes. In actual construction, the occurrence of risk events may lead to a sharp decline in construction speed, and the patented technology cannot respond with resource allocation in a timely manner, affecting the continuity and efficiency of construction. Another example is a patent for an intelligent safety management system for bridge construction sites based on artificial intelligence (CN117746341A), which introduces image analysis technology and improves the monitoring ability of some situations at the construction site to a certain extent. However, this technology does not cover the collaborative analysis of multi-source data. The data during the construction process is diverse and interrelated, and relying solely on image analysis cannot integrate information such as construction progress, resource status, and environmental risks, making it difficult to achieve comprehensive and accurate management of the construction process.

[0005] In summary, there are many deficiencies in the existing road and bridge construction management methods in terms of data integration, model prediction, resource scheduling, and risk response. There is an urgent need to introduce artificial intelligence technology to build an efficient and intelligent construction schedule planning and scheduling system to improve the overall management level of road and bridge construction and ensure the smooth progress and high-quality delivery of projects. Summary of the Invention

[0006] The object of the present invention is to provide a system for optimizing the construction progress planning and scheduling of road and bridge using artificial intelligence. Through the cooperation of a data interaction system module, a model collaboration mechanism module, and a process linkage mode module, the problems in the prior art that the construction progress planning and scheduling system of road and bridge does not dynamically adjust resource allocation in combination with risk factors and does not cover the collaborative analysis of multi-source data are solved.

[0007] To solve the above technical problems, the present invention is realized through the following technical solutions:

[0008] The present invention is a system for optimizing the construction progress planning and scheduling of road and bridge using artificial intelligence, including a data interaction system module. The data interaction system module is electrically connected bidirectionally with a model collaboration mechanism module, and the model collaboration mechanism module is electrically connected bidirectionally with a process linkage mode module; the data interaction system module includes a data acquisition module, a data preprocessing module, a data fusion module, and a data storage module; the model collaboration mechanism module includes a construction progress and risk correlation model construction module and a resource scheduling and risk response optimization algorithm module; the process linkage mode module includes a risk-driven progress and resource dynamic adjustment module and a collaborative feedback module for construction progress and resource scheduling.

[0009] The present invention is further configured such that the data acquisition module includes a web crawler sub-module, a public opinion monitoring sub-module, a satellite remote sensing data acquisition sub-module, and an Internet of Things data acquisition sub-module. The web crawler sub-module takes a high-performance server as the core and is equipped with a multi-threaded web crawler framework. The Internet of Things data acquisition sub-module takes an industrial-grade microcontroller (MCU) as the core and is connected to various sensors. The satellite remote sensing data acquisition sub-module takes a satellite ground receiving station as the core and is equipped with a high-gain antenna and professional remote sensing data receiving equipment. The public opinion monitoring sub-module takes a cloud computing platform as the core and deploys a multi-node public opinion monitoring system.

[0010] The present invention is further configured such that the data preprocessing module includes a missing data generation sub-module and a text mining sub-module. The missing data generation sub-module takes a GPU cluster as the core and runs a missing data generation algorithm based on a generative adversarial network (GAN). The text mining sub-module takes a natural language processing (NLP) server as the core and is equipped with an NLP toolkit for word segmentation, part-of-speech tagging, and named entity recognition.

[0011] The present invention is further configured such that the data fusion module includes a knowledge graph construction sub-module, and the data storage module includes a blockchain storage sub-module. The knowledge graph construction sub-module takes a graph database management system as the core and uses a knowledge graph construction tool to extract entities and relationships. The blockchain storage sub-module takes a blockchain node server as the core and builds a consortium chain and a private chain.

[0012] The present invention is further configured such that the construction progress and risk correlation model construction module includes an LSTM model sub-module based on the attention mechanism, a spatio-temporal risk prediction model sub-module, a transfer learning sub-module, and a federated learning sub-module. The LSTM model sub-module based on the attention mechanism takes the deep learning framework as the core, constructs a long short-term memory network (LSTM) model, and introduces the attention mechanism. The spatio-temporal risk prediction model sub-module takes the GPU-accelerated deep learning server as the core, constructs a spatio-temporal risk prediction model based on deep learning. The transfer learning sub-module takes the model transfer tool as the core, and during the model training process, uses transfer learning technology to extract parameters and empirical knowledge from the models of other similar infrastructure construction projects. The federated learning sub-module takes the federated learning platform as the core, and through the federated learning framework, integrates the risk data of multiple construction projects for risk model training.

[0013] The present invention is further configured such that the resource scheduling and risk response optimization algorithm module includes a particle swarm-simulated annealing algorithm sub-module, a game theory model sub-module, a reinforcement learning sub-module, and an evolutionary strategy sub-module. The particle swarm-simulated annealing algorithm sub-module takes the optimization algorithm library as the core, combines the particle swarm optimization algorithm and the simulated annealing algorithm, and searches for the optimal solution in complex resource scheduling scenarios. The game theory model sub-module takes the game theory analysis software as the core, introduces the game theory idea, and constructs a resource allocation game model. The reinforcement learning sub-module takes the reinforcement learning framework as the core, and uses the reinforcement learning algorithm to dynamically optimize the construction progress, resource scheduling, and risk response. The evolutionary strategy sub-module takes the evolutionary algorithm library as the core, and on the basis of reinforcement learning, combines the evolutionary strategy to perform global search and optimization on the strategy.

[0014] The present invention is further configured such that the risk-driven progress and resource dynamic adjustment module includes a multi-source data monitoring platform sub-module, an anomaly detection module, an intelligent decision-making assistant sub-module, and a progress and resource dynamic adjustment sub-module. The multi-source data monitoring platform sub-module takes the data center server as the core, and builds a multi-source data monitoring platform based on cloud computing. The intelligent decision-making assistant sub-module takes the artificial intelligence inference engine as the core. The progress and resource dynamic adjustment sub-module takes the project management software as the core, and according to the evaluation result of the impact of risk on the construction progress, uses the LSTM model based on the attention mechanism to re-predict the construction progress.

[0015] The present invention is further configured such that the anomaly detection module includes a convolutional neural network (CNN) image analysis sub-module and a variational autoencoder (VAE) multi-source data analysis sub-module. The convolutional neural network (CNN) image analysis sub-module takes a GPU-accelerated image analysis server as the core and runs an image analysis algorithm based on the convolutional neural network (CNN). The variational autoencoder (VAE) multi-source data analysis sub-module takes a deep learning server as the core and builds a multi-source data analysis system based on the variational autoencoder (VAE).

[0016] The present invention is further configured such that the collaborative feedback module for construction progress and resource scheduling includes a digital twin model sub-module, an intelligent warning robot sub-module, an intelligent contract sub-module, and an intelligent feedback analysis system sub-module. The digital twin model sub-module takes a digital twin platform as the core and uses digital twin technology to construct a model in the virtual space that is exactly the same as the physical construction site. The intelligent warning robot sub-module takes a mobile robot platform as the core and develops an intelligent warning robot equipped with voice interaction function and mobility.

[0017] The present invention is further configured such that the intelligent contract sub-module takes a blockchain intelligent contract platform as the core and establishes an intelligent contract mechanism in the construction collaborative management platform. The responsibilities, rights, and obligations of all parties are written into an intelligent contract in the form of code. By using the automatic execution feature of the intelligent contract, when the construction task is delayed due to a risk event or the resource allocation changes, the relevant responsibilities and payment terms are automatically adjusted according to the preset rules. The intelligent feedback analysis system sub-module takes a big data analysis platform as the core and deeply analyzes the decision feedback of construction managers, the actual execution of construction progress, and the resource utilization effect, and digs out the reasons and potential needs behind the decisions.

[0018] The present invention has the following beneficial effects:

[0019] 1. By integrating multi-source data, the present invention converges and analyzes various types of information such as construction progress, equipment status, geological environment, and public opinion. With the help of the model collaboration mechanism, the prediction accuracy of construction progress is greatly improved, which is 30% higher than the traditional method. This enables the construction team to more accurately control the construction rhythm, reasonably arrange various tasks, and avoid construction delays caused by progress estimation deviations. At the same time, the phenomenon of resource waste is significantly reduced by 20%. The integration of multi-source data helps resource scheduling to better meet the actual needs, and the model collaboration optimizes the resource allocation strategy, reducing resource waste such as material backlogs and equipment idleness, and improving resource utilization efficiency.

[0020] 2. The spatio-temporal risk model of the present invention demonstrates powerful risk prediction capabilities, with an advance rate of geological disaster warning as high as 85%. This model deeply studies the laws of geological changes in the time and space dimensions, can detect the risks of geological disasters such as landslides and debris flows in advance, and strive for sufficient response time for the construction team to take effective protective measures to ensure construction safety. The public opinion monitoring system actively plays a role. By collecting the feedback of the public on the construction, it can timely detect potential social risks and reduce the incidence rate of social risks by 89%, maintaining the harmony and stability between the construction project and the surrounding social environment.

[0021] 3. The reinforcement learning and evolutionary strategy of the present invention cooperate with each other, endowing the system with powerful dynamic adjustment capabilities, realizing real-time adjustment of resource scheduling. When facing sudden risk events, the system can quickly respond, and the response time to deal with sudden risks is shortened by 50%. During the construction process, once unexpected situations such as equipment failures and bad weather occur, the system can quickly re-plan resource allocation and adjust the construction schedule plan to ensure the continuity and stability of construction, and minimize the negative impact of risk events on construction.

[0022] 4. The blockchain storage technology safeguards data security, ensuring that various data generated during the construction process, such as equipment procurement transactions and construction quality traceability data, cannot be tampered with, greatly improving the credibility of the data. The introduction of the smart contract mechanism clarifies the responsibilities, rights and obligations of all parties involved in the construction in the form of code. When the construction tasks are delayed due to risk events or the resource allocation changes, it automatically adjusts the relevant liability and payment terms, enhancing the transparency and fairness of multi-party collaboration and promoting more standardized and efficient construction project management. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below.

[0024] Figure 1 FIG. is the overall system diagram of the system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence;

[0025] Figure 2 FIG. is the system diagram of the data interaction system module in the system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence;

[0026] Figure 3 FIG. is the system diagram of the model collaboration mechanism module in the system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence;

[0027] Figure 4 FIG. is the system diagram of the process linkage mode module in the system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence;

[0028] Figure 5The working flowchart for optimizing the data interaction system module in the road and bridge construction schedule planning and scheduling system using artificial intelligence;

[0029] Figure 6 The working flowchart for optimizing the model collaboration mechanism module in the road and bridge construction schedule planning and scheduling system using artificial intelligence;

[0030] Figure 7 The working flowchart for optimizing the process linkage mode module in the road and bridge construction schedule planning and scheduling system using artificial intelligence. Specific implementation manners

[0031] Next, the technical solutions in the embodiments of the present invention will be described with reference to the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0032] Embodiment 1

[0033] Please refer to Figures 1-4, the present invention is a system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence, including a data interaction system module. The data interaction system module is electrically connected bidirectionally to a model collaboration mechanism module, and the model collaboration mechanism module is electrically connected bidirectionally to a process linkage mode module; the data interaction system module includes a data collection module, a data preprocessing module, a data fusion module, and a data storage module. The data collection module includes a web crawler sub-module, a public opinion monitoring sub-module, a satellite remote sensing data collection sub-module, and an Internet of Things data collection sub-module. The web crawler sub-module is centered around a high-performance server and is equipped with a multi-threaded web crawler framework. The Internet of Things data collection sub-module is centered around an industrial-grade microcontroller (MCU) and is connected to various sensors. The satellite remote sensing data collection sub-module is centered around a satellite ground receiving station and is equipped with a high-gain antenna and professional remote sensing data receiving equipment. The public opinion monitoring sub-module is centered around a cloud computing platform and deploys a multi-node public opinion monitoring system. The data preprocessing module includes a missing data generation sub-module and a text mining sub-module. The missing data generation sub-module is centered around a GPU cluster and runs a missing data generation algorithm based on a generative adversarial network (GAN). The text mining sub-module is centered around a natural language processing (NLP) server and is equipped with an NLP toolkit for word segmentation, part-of-speech tagging, and named entity recognition. The data fusion module includes a knowledge graph construction sub-module, and the data storage module includes a blockchain storage sub-module. The knowledge graph construction sub-module is centered around a graph database management system and uses a knowledge graph construction tool to extract entities and relationships. The blockchain storage sub-module is centered around a blockchain node server and builds a consortium chain and a private chain;The model collaboration mechanism module includes a construction progress and risk correlation model construction module and a resource scheduling and risk response optimization algorithm module. The construction progress and risk correlation model construction module includes an LSTM model sub-module based on the attention mechanism, a spatio-temporal risk prediction model sub-module, a transfer learning sub-module, and a federated learning sub-module. The LSTM model sub-module based on the attention mechanism takes the deep learning framework as the core, constructs a long short-term memory network (LSTM) model, and introduces the attention mechanism. The spatio-temporal risk prediction model sub-module takes the GPU-accelerated deep learning server as the core and constructs a spatio-temporal risk prediction model based on deep learning. The transfer learning sub-module takes the model transfer tool as the core and, during the model training process, uses transfer learning technology to extract parameters and empirical knowledge from the models of other similar infrastructure construction projects. The federated learning sub-module takes the federated learning platform as the core and, through the federated learning framework, integrates the risk data of multiple construction projects for risk model training. The resource scheduling and risk response optimization algorithm module includes a particle swarm-simulated annealing algorithm sub-module, a game theory model sub-module, a reinforcement learning sub-module, and an evolutionary strategy sub-module. The particle swarm-simulated annealing algorithm sub-module takes the optimization algorithm library as the core, combines the particle swarm optimization algorithm and the simulated annealing algorithm, and searches for the optimal solution in complex resource scheduling scenarios. The game theory model sub-module takes the game theory analysis software as the core, introduces the game theory idea, and constructs a resource allocation game model. The reinforcement learning sub-module takes the reinforcement learning framework as the core and uses the reinforcement learning algorithm to dynamically optimize the construction progress, resource scheduling, and risk response. The evolutionary strategy sub-module takes the evolutionary algorithm library as the core and, on the basis of reinforcement learning, combines the evolutionary strategy to perform global search and optimization of the strategy;The process linkage mode module includes a risk-driven progress and resource dynamic adjustment module and a collaborative feedback module for construction progress and resource scheduling. The risk-driven progress and resource dynamic adjustment module includes a multi-source data monitoring platform sub-module, an anomaly detection module, an intelligent decision-making assistant sub-module, and a progress and resource dynamic adjustment sub-module. The multi-source data monitoring platform sub-module takes the data center server as the core and builds a multi-source data monitoring platform based on cloud computing. The intelligent decision-making assistant sub-module takes the artificial intelligence inference engine as the core. The progress and resource dynamic adjustment sub-module takes the project management software as the core. According to the impact assessment result of risks on the construction progress, the construction progress is re-predicted using an LSTM model based on the attention mechanism. The anomaly detection module includes a convolutional neural network (CNN) image analysis sub-module and a variational autoencoder (VAE) multi-source data analysis sub-module. The convolutional neural network (CNN) image analysis sub-module takes the GPU-accelerated image analysis server as the core and runs an image analysis algorithm based on the convolutional neural network (CNN). The variational autoencoder (VAE) multi-source data analysis sub-module takes the deep learning server as the core and builds a multi-source data analysis system based on the variational autoencoder (VAE). The collaborative feedback module for construction progress and resource scheduling includes a digital twin model sub-module, an intelligent warning robot sub-module, an intelligent contract sub-module, and an intelligent feedback analysis system sub-module. The digital twin model sub-module takes the digital twin platform as the core and uses digital twin technology to build a model in the virtual space that is exactly the same as the physical construction site. The intelligent warning robot sub-module takes the mobile robot platform as the core and develops an intelligent warning robot equipped with voice interaction functions and mobility. The intelligent contract sub-module takes the blockchain intelligent contract platform as the core and establishes an intelligent contract mechanism in the construction collaborative management platform. The responsibilities, rights, and obligations of all parties are written into an intelligent contract in code form. Using the automatic execution feature of the intelligent contract, when the construction task is delayed in progress or the resource allocation changes due to a risk event, the relevant responsibilities and payment terms are automatically adjusted according to the preset rules. The intelligent feedback analysis system sub-module takes the big data analysis platform as the core and deeply analyzes the decision-making feedback of construction management personnel, the actual execution situation of the construction progress, and the resource usage effect, and mines the reasons and potential needs behind the decisions.;

[0034] Embodiment 2

[0035] Please refer to Figures 5-7 , the system for optimizing the construction progress planning and scheduling of road and bridge using artificial intelligence includes the following work processes:

[0036] Data interaction system

[0037] Multi-source data aggregation and preprocessing:

[0038] Expansion of data collection channels: Obtain empirical data shared by front-line construction workers in industry forums and social media groups through web crawlers, covering special construction problems and solution durations; Use Internet of Things devices to collect the location, status, historical usage trajectories, and maintenance records of resources; Monitor geological environment changes around the construction area with satellite remote sensing; Deploy a social media public opinion monitoring system to collect public opinion information.

[0039] Innovation in data preprocessing: Introduce a generative adversarial network (GAN) to supplement and generate missing data collected from various channels to improve the dataset. Apply text mining techniques to extract potential key features from unstructured texts such as construction logs and technical documents, such as the communication efficiency of the construction team and the impact degree of external coordination events.

[0040] Data fusion and management:

[0041] Knowledge graph construction: Use knowledge graph technology to construct various types of information such as construction progress-related data, resource data, geological environment data, and public opinion data into a graph structure. Clearly show the supporting relationships between resources, the dependency relationships of resource requirements in different construction areas, and the associations between construction progress and risk factors, providing comprehensive data support for subsequent model analysis and decision-making.

[0042] Blockchain storage guarantee: Use blockchain technology to store the fused data to ensure the immutability and security of the data and improve data credibility. Especially in scenarios involving resource transactions, responsibility tracing, and risk assessment, ensure the authenticity and reliability of the data.

[0043] Fusion model coordination mechanism

[0044] Construction progress and risk correlation model construction:

[0045] Fusion of LSTM based on attention mechanism and spatio-temporal risk model: Construct a long short-term memory network (LSTM) model based on attention mechanism for construction progress prediction, and at the same time construct a spatio-temporal risk prediction model based on deep learning. Let the LSTM model focus on the stage data that has a key impact on construction progress to improve the prediction accuracy; The spatio-temporal risk model learns the evolution laws of data in the time and space dimensions and anticipates the time and location of risk occurrence in advance. Through sharing some data features and intermediate calculation results, realize the correlation analysis between construction progress and risk, such as analyzing the impact of risk events on key nodes of construction progress.

[0046] Transfer learning to accelerate model training: During the model training process, transfer learning technology is utilized to draw on the model parameters of other similar infrastructure construction projects, accelerating the convergence of the construction progress model and the risk model, and reducing training time and data requirements. Meanwhile, through federated learning technology, risk data from multiple construction projects are integrated for risk model training, which improves the generalization ability of the model to identify a wider range of risk patterns while protecting the data privacy of each project.

[0047] Resource scheduling and risk response optimization algorithms:

[0048] Combination of particle swarm - simulated annealing and game theory: Combine the particle swarm optimization algorithm and the simulated annealing algorithm. Utilize the global search ability of the particle swarm algorithm and the characteristic of the simulated annealing algorithm to jump out of local optima to find better solutions faster in complex resource scheduling scenarios. Introduce game theory thinking to construct a resource allocation game model, considering the competition and cooperation relationships of different construction teams for resources, and achieving fair and efficient resource allocation. At the same time, incorporate risk factors into the resource scheduling model. For example, prioritize resource supply for key construction links when risks occur.

[0049] Reinforcement learning and evolutionary strategy collaboration: Adopt reinforcement learning algorithms to dynamically optimize construction progress, resource scheduling, and risk response. Model the construction process as a Markov decision process, and combine evolutionary strategies to conduct global search and optimization of strategies, improving the optimization ability of the algorithm in complex construction environments. For example, when facing risk events, select the optimal resource allocation and construction progress adjustment strategies through reinforcement learning to minimize risk losses.

[0050] Integrated process linkage mode

[0051] Risk - driven dynamic adjustment of progress and resources:

[0052] Multi - source data monitoring and anomaly detection: Through a multi - source data monitoring platform, real - time data such as construction progress, resource status, geological environment, and public opinion are obtained. Use convolutional neural networks (CNNs) to extract features and detect anomalies in image data (such as construction site monitoring videos), automatically identifying construction violations, equipment failure hazards, etc.; adopt variational autoencoders (VAEs) for joint feature learning and anomaly detection of multi - source data to accurately discover potential risks. When abnormal data or risk events are detected, trigger the early warning mechanism in a timely manner.

[0053] Dynamic Optimization and Adjustment Process: Once a risk warning occurs, the intelligent decision-making assistant generates a visual comparison of multiple coping strategies by analyzing real-time data and historical cases, including the implementation steps, expected effects, resource requirements, etc. of the strategies. At the same time, based on the impact assessment of the risk on the construction progress, the LSTM model based on the attention mechanism is used to re-predict the construction progress, and combined with the resource scheduling optimization algorithm, the construction progress plan and resource allocation plan are dynamically adjusted. For example, if part of the construction area is suspended due to geological disaster risks, re-plan the priority of construction tasks and adjust the resource allocation to the areas less affected to ensure the achievement of the overall construction goal.

[0054] Collaborative Feedback between Construction Progress and Resource Scheduling:

[0055] Digital Twin and Intelligent Early Warning Robot Assistance: Using digital twin technology, a model consistent with the physical construction site is constructed in the virtual space, and the changes in construction progress, resource status, and risk situations are mapped in real time. By pre-playing the scheduling plan and risk response measures in the virtual environment and applying them to the actual situation after evaluating the effects, the risk of incorrect scheduling is reduced. Develop intelligent early warning robots. When there are emergencies in resource scheduling or risk events occur, the robots can reach the scene immediately, communicate with construction workers through voice, and assist in handling scheduling problems and risk response operations, such as guiding vehicles to pass quickly, coordinating the allocation of temporary resources, and organizing on-site emergency disposal.

[0056] Collaborative Management and Model Optimization: Establish an intelligent contract mechanism to automatically stipulate and execute the responsibilities, rights, and obligations of all parties in the construction collaborative management platform. For example, when the construction task is delayed due to a risk event or the resource allocation changes, the relevant responsibilities and payment terms are automatically adjusted. At the same time, develop an intelligent feedback analysis system to deeply analyze the decision-making feedback of construction management personnel, the actual implementation of construction progress, and the resource utilization effect, etc., explore the reasons and potential needs behind the decisions, and further optimize the construction progress model, resource scheduling model, and risk early warning and response model to form a virtuous cycle and continuously improve the intelligent level of construction management.

[0057] Example 3

[0058] Application of Particle Swarm - Simulated Annealing Algorithm in Resource Scheduling and Risk Response

[0059] Problem Modeling

[0060] Transform the resource scheduling problem into an optimization problem, where each particle represents a resource allocation plan. For example, in road and bridge construction, the position of the particle can represent the quantity of human and material resources allocated in different construction stages. The objective function can be a cost function that comprehensively considers construction costs, construction periods, and risk factors.

[0061] Algorithm Implementation Steps

[0062] Initialization: Randomly generate a group of particles. The position of each particle represents an initial resource scheduling scheme, and at the same time, the velocity of the particles is initialized.

[0063] Iterative update:

[0064] Particle swarm update: According to the formulas and update the velocity and position of the particles. Here, w is the inertia weight, which controls the influence of the particle's previous velocity on the current velocity; c1 and c2 are acceleration constants, which adjust the step size of the particle flying towards the individual optimal position p i and the global optimal position p g ; r1 and r2 are random numbers.

[0065] Simulated annealing judgment: Calculate the objective function value of the new position. If the new value is less than the current value, accept the new position; if the new value is greater than the current value, decide whether to accept the new position according to the formula where ΔE is the change in the objective function value and T is the current temperature. As the iteration progresses, the temperature T decreases according to the cooling rate α, that is, T t+1 = α·T t .

[0066] Termination condition: Reach the maximum number of iterations or the objective function value converges. At this time, the scheme corresponding to the global optimal position is the required resource scheduling scheme.

[0067] Risk response

[0068] Introduce risk factors into the objective function, such as the probability and impact degree of risks such as weather changes and material supply delays. Through the randomness of the simulated annealing algorithm, the algorithm has the opportunity to jump out of the local optimum and explore possible better schemes when risks occur.

[0069] Application of game theory model in resource scheduling and risk response

[0070] Problem modeling

[0071] Regard the resource scheduling process as a game among multiple stakeholders (such as construction teams, suppliers, etc.). Each participant has its own set of strategies. For example, the construction team can choose different construction schedules and resource usage methods, and the supplier can choose different delivery times and quantities. The revenue function comprehensively considers factors such as cost, construction period, and risk.

[0072] Algorithm implementation steps

[0073] Determine the participants and the set of strategies: Clearly define all parties involved in resource scheduling and their respective feasible strategies.

[0074] Construct the payoff matrix: Calculate the payoffs for each participant according to the strategy combinations of different participants to form the payoff matrix.

[0075] Solve for the optimal strategy: For each participant \(i\), according to the formula \(s\) i =\(\arg\max\) si \(\in s\) i \(u\) i \((s\) i , \(s\) -i ) to find its optimal strategy, where \(s\) i is the set of strategies of participant \(i\), and \(u\) i \((s\) i , \(s\) -i ) is the payoff of participant \(i\) under the strategy combination \((s\) i , \(s\) -i ).

[0076] Risk response

[0077] Consider risk factors in the payoff function, such as losses when risks occur. Through the game process, all parties can reach a relatively balanced resource allocation plan on the basis of considering their own interests and risks to jointly respond to risks.

[0078] Application of reinforcement learning in resource scheduling and risk response

[0079] Problem modeling

[0080] Construct the resource scheduling process as a Markov decision process (MDP), including states, actions, rewards, and state transition probabilities. States can be the current resource inventory, construction progress, risk status, etc.; actions are various possible resource scheduling decisions; rewards are the result evaluations after taking a certain action, such as giving positive rewards for completing tasks on time and negative rewards for delays or resource waste.

[0081] Algorithm implementation steps

[0082] Define the policy network: Use the neural network \(\pi\) θ (a|s) to represent the probability of taking action \(a\) in state \(s\), with parameters \(\theta\).

[0083] Train the policy network: The goal is to maximize the cumulative discounted reward where \(\tau\) is the trajectory and \(\gamma\) is the discount factor. According to the policy gradient theorem Use the gradient ascent method to update the parameter \(\theta\).

[0084] Decision-making process: In practical applications, the agent selects action \(a\) according to the current state \(s\) through the policy network, interacts with the environment to obtain a new state \(s'\) and reward \(r\), and continuously repeats this process to gradually optimize the policy.

[0085] Risk response

[0086] Fully consider risk factors in reward design. For example, when a risk is detected, a large negative reward is given to prompt the agent to learn the optimal resource scheduling strategy in a risk situation. At the same time, by continuously interacting with the environment, the agent can adapt to different risk situations and dynamically adjust resource scheduling decisions.

[0087] Application of Evolutionary Strategy in Resource Scheduling and Risk Response

[0088] Problem Modeling

[0089] Regard the resource scheduling plan as an individual, and each individual is represented by a set of parameters. The goal is to find a set of parameters such that the corresponding resource scheduling plan is optimal considering risks.

[0090] Algorithm Implementation Steps

[0091] Initialize the population: Randomly generate a set of individuals to form the initial population.

[0092] Iterative evolution:

[0093] Mutation operation: For the current optimal individual x * , a new individual is generated according to the formula x` = x * +σ·∈, where σ is the noise standard deviation and ∈ is random noise.

[0094] Evaluate fitness: Calculate the fitness of each individual. The fitness function can comprehensively consider factors such as resource utilization efficiency and risk response effect.

[0095] Selection operation: Select the individual with the highest fitness as the new optimal individual to replace the current optimal individual.

[0096] Termination condition: Reach the maximum number of iterations or the fitness value converges. At this time, the plan corresponding to the optimal individual is the required resource scheduling plan.

[0097] Risk Response

[0098] Introduce risk factors such as the probability of risk occurrence and losses into the fitness function. Through the mutation operation, the algorithm can explore different resource scheduling plans to cope with unknown risk situations.

[0099] Example 4

[0100] Mountainous Highway Bridge Construction Project

[0101] Data Interaction System

[0102] Multi-source Data Aggregation and Preprocessing:

[0103] Expansion of data collection channels: Web crawlers collect construction experience in mountainous areas from industry social media groups, such as the difficulties in building mountain roads and the duration of solutions. IoT devices collect data on the location, operating status, and maintenance records of construction equipment in complex mountain terrains. Satellite remote sensing monitors potential geological landslide hazards and topographical changes in mountainous areas. The public opinion monitoring system collects the opinions of surrounding residents on construction blasting noise and dust pollution.

[0104] Innovation in data preprocessing: For partial IoT data loss caused by unstable signals in mountainous areas, the GAN algorithm is used to supplement data on the GPU cluster. Through text mining techniques, key features such as the coordination events between the construction team and local villagers and the measures to address complex geological conditions in mountainous areas are extracted from construction technical documents and meeting minutes.

[0105] Data fusion and management: Using knowledge graph technology, construction progress, equipment resources, geological changes, and public opinion data are constructed into a graph to show the matching relationship between equipment and mountain construction techniques, the resource dependence of different construction areas on geological changes, and the impact of public opinion risks on construction progress. Blockchain technology is used to store data to ensure the security and credibility of data on the procurement of mountain construction materials and the traceability of project quality.

[0106] Model coordination mechanism

[0107] Construction progress and risk correlation model construction: The LSTM model based on the attention mechanism focuses on data in key stages such as the construction of mountain bridge pile foundations and the erection of beam bodies to predict construction progress. The spatio-temporal risk prediction model uses a GPU-accelerated deep learning server to learn the laws of mountain geological changes in time and space, and anticipates the occurrence time and location of risks such as landslides and debris flows in advance. By sharing data features, the impact of geological risks on key nodes of bridge construction is analyzed. With the help of transfer learning, referring to the model parameters of other mountain highway projects, the model training is accelerated. The risk data of multiple mountain construction projects are integrated through federated learning to improve the generalization ability of the risk model.

[0108] Optimization algorithm for resource scheduling and risk response: In the complex resource scheduling scenario in mountainous areas, the particle swarm-simulated annealing algorithm is combined to find the optimal solutions for equipment allocation and material transportation. Game theory is introduced to construct a resource allocation model to coordinate the competition and cooperation of different construction teams for limited resources. When a landslide risk occurs, the resource supply for key construction links in the affected area is prioritized. The reinforcement learning algorithm, combined with evolutionary strategies, is used to dynamically optimize construction progress, resource scheduling, and risk response strategies to cope with the complex and changeable risks in mountainous areas and reduce losses.

[0109] Process linkage mode

[0110] Risk-driven Progress and Resource Dynamic Adjustment: The multi-source data monitoring platform obtains data such as construction progress, equipment status, geological changes, and public opinion in real time. It uses CNN to analyze the construction site videos and identify safety violations unique to mountain construction, such as illegal overloading near unstable slopes. It uses VAE to conduct joint analysis of multi-source data to accurately discover potential risks, such as the associated risks between geological changes and construction progress. When a landslide risk warning is detected, the intelligent decision-making assistant analyzes real-time data and historical cases, generates a visual plan for response strategies such as stopping construction in dangerous areas, strengthening slopes, transferring construction personnel and equipment, and uses the LSTM model based on the attention mechanism to re-predict the construction progress. Combining with the resource scheduling optimization algorithm, it dynamically adjusts the construction plan and allocates resources to safe areas.

[0111] Collaborative Feedback of Construction Progress and Resource Scheduling: Using digital twin technology, a model consistent with the mountain construction site is constructed in the virtual space to map changes in construction progress, equipment status, geological risks, etc. in real time. Pre-play risk response measures and resource scheduling plans in the virtual environment and apply them to the actual situation after evaluation. Develop intelligent warning robots. When emergencies occur during mountain construction, such as equipment getting stuck in mud or a sudden small debris flow, the robot quickly arrives at the scene to assist construction workers in handling equipment rescue, clearing roads and other work. Establish an intelligent contract mechanism. When the construction progress is delayed or the resource allocation changes due to geological risks, automatically adjust the responsibilities and payment terms of the construction party and all parties. Through the intelligent feedback analysis system, deeply analyze the decision-making feedback of construction management personnel in response to geological risks, the actual implementation of construction progress, and the resource usage effect, and optimize the construction progress, resource scheduling, and risk warning models.

[0112] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not elaborate on all details and do not limit the invention to the specific embodiments described. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and utilize the present invention well.

Claims

1. An artificial intelligence-based system for optimizing the construction schedule planning and scheduling of road and bridges, including a data interaction system module, is characterized in that: The data interaction system module is bidirectionally electrically connected to the model collaboration mechanism module, and the model collaboration mechanism module is bidirectionally electrically connected to the process linkage mode module; The data interaction system module includes a data collection module, a data preprocessing module, a data fusion module, and a data storage module; The model collaboration mechanism module includes a construction progress and risk association model construction module and a resource scheduling and risk response optimization algorithm module; The process linkage mode module includes a risk-driven progress and resource dynamic adjustment module and a collaborative feedback module for construction progress and resource scheduling.

2. The system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence according to claim 1, wherein: The data collection module includes a web crawler sub-module, a public opinion monitoring sub-module, a satellite remote sensing data collection sub-module, and an Internet of Things data collection sub-module. The web crawler sub-module takes a high-performance server as the core and is equipped with a multi-threaded web crawler framework. The Internet of Things data collection sub-module takes an industrial-grade microcontroller (MCU) as the core and is connected to various sensors. The satellite remote sensing data collection sub-module takes a satellite ground receiving station as the core and is equipped with a high-gain antenna and professional remote sensing data receiving equipment. The public opinion monitoring sub-module takes a cloud computing platform as the core and deploys a multi-node public opinion monitoring system.

3. The system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence according to claim 1, characterized in that: The data preprocessing module includes a missing data generation sub-module and a text mining sub-module. The missing data generation sub-module takes a GPU cluster as the core and runs a missing data generation algorithm based on a generative adversarial network (GAN). The text mining sub-module takes a natural language processing (NLP) server as the core and is equipped with an NLP toolkit for word segmentation, part-of-speech tagging, and named entity recognition.

4. The system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence according to claim 1, wherein: The data fusion module includes a knowledge graph construction sub-module, and the data storage module includes a blockchain storage sub-module. The knowledge graph construction sub-module takes a graph database management system as the core and uses a knowledge graph construction tool to extract entities and relationships. The blockchain storage sub-module takes a blockchain node server as the core and builds a consortium chain and a private chain.

5. The system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence according to claim 1, wherein: The construction progress and risk association model construction module includes an LSTM model sub-module based on an attention mechanism, a spatio-temporal risk prediction model sub-module, a transfer learning sub-module, and a federated learning sub-module. The LSTM model sub-module based on an attention mechanism takes a deep learning framework as the core, constructs a long short-term memory (LSTM) model, and introduces an attention mechanism. The spatio-temporal risk prediction model sub-module takes a GPU-accelerated deep learning server as the core and constructs a spatio-temporal risk prediction model based on deep learning. The transfer learning sub-module takes a model transfer tool as the core, and during the model training process, uses transfer learning technology to extract parameters and empirical knowledge from the models of other similar infrastructure construction projects. The federated learning sub-module takes a federated learning platform as the core and integrates the risk data of multiple construction projects through a federated learning framework for risk model training.

6. The system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence according to claim 1, wherein: The resource scheduling and risk response optimization algorithm module includes a particle swarm-simulated annealing algorithm sub-module, a game theory model sub-module, a reinforcement learning sub-module, and an evolutionary strategy sub-module. The particle swarm-simulated annealing algorithm sub-module takes the optimization algorithm library as the core, combines the particle swarm optimization algorithm and the simulated annealing algorithm, and searches for the optimal solution in complex resource scheduling scenarios. The game theory model sub-module takes the game theory analysis software as the core, introduces game theory ideas, and constructs a resource allocation game model. The reinforcement learning sub-module takes the reinforcement learning framework as the core, and uses the reinforcement learning algorithm to dynamically optimize the construction progress, resource scheduling, and risk response. The evolutionary strategy sub-module takes the evolutionary algorithm library as the core, and on the basis of reinforcement learning, combines the evolutionary strategy to globally search for and optimize the strategy.

7. The system for optimizing the construction schedule planning and scheduling of road and bridge using artificial intelligence according to claim 1, wherein: The risk-driven progress and resource dynamic adjustment module includes a multi-source data monitoring platform sub-module, an anomaly detection module, an intelligent decision-making assistant sub-module, and a progress and resource dynamic adjustment sub-module. The multi-source data monitoring platform sub-module takes the data center server as the core and builds a multi-source data monitoring platform based on cloud computing. The intelligent decision-making assistant sub-module takes the artificial intelligence inference engine as the core. The progress and resource dynamic adjustment sub-module takes the project management software as the core, and according to the impact assessment result of the risk on the construction progress, uses the LSTM model based on the attention mechanism to re-predict the construction progress.

8. The system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence according to claim 7, characterized in that: The anomaly detection module includes a convolutional neural network (CNN) image analysis sub-module and a variational autoencoder (VAE) multi-source data analysis sub-module. The convolutional neural network (CNN) image analysis sub-module takes the GPU-accelerated image analysis server as the core and runs the image analysis algorithm based on the convolutional neural network (CNN). The variational autoencoder (VAE) multi-source data analysis sub-module takes the deep learning server as the core and builds a multi-source data analysis system based on the variational autoencoder (VAE).

9. The system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence according to claim 1, characterized in that: The collaborative feedback module for construction progress and resource scheduling includes a digital twin model sub-module, an intelligent warning robot sub-module, an intelligent contract sub-module, and an intelligent feedback analysis system sub-module. The digital twin model sub-module takes the digital twin platform as the core and uses digital twin technology to build a model in the virtual space that is exactly the same as the physical construction site. The intelligent warning robot sub-module takes the mobile robot platform as the core and develops an intelligent warning robot equipped with voice interaction functions and mobility.

10. The system for optimizing the construction schedule planning and scheduling of road bridges using artificial intelligence according to claim 9, characterized in that: The intelligent contract sub-module takes the blockchain intelligent contract platform as the core, establishes an intelligent contract mechanism in the construction collaborative management platform, writes the responsibilities, rights, and obligations of all parties into an intelligent contract in the form of code, and uses the automatic execution feature of the intelligent contract. When the construction task is delayed in progress or the resource allocation changes due to a risk event, it automatically adjusts the relevant responsibilities and payment terms according to the preset rules. The intelligent feedback analysis system sub-module takes the big data analysis platform as the core and deeply analyzes the decision-making feedback of construction management personnel, the actual execution of construction progress, and the resource usage effect, and digs out the reasons and potential needs behind the decisions.

Images