Municipal road construction section scheduling optimization method and system based on artificial intelligence

Abstract

The invention relates to the technical field of municipal road construction, and discloses a municipal road construction section scheduling optimization method and system based on artificial intelligence, and the method comprises the steps: collecting the multi-source dynamic data of a construction region through Internet of Things equipment and a satellite remote sensing technology, and constructing a time sequence database; a multi-objective optimization model is established, and an initial scheduling scheme is generated by adopting a non-dominated sorting genetic algorithm with the goal of minimizing construction period, resource waste and traffic jam influence. Dynamically adjusting the priority of the construction section by using a reinforcement learning algorithm, and updating a construction task sequence; a graph neural network is utilized to detect space-time conflicts of construction sections, and conflict-free scheduling constraint conditions are generated; and optimizing resource allocation based on mixed integer linear programming, simulating a construction process in combination with a digital twinborn technology, correcting a construction progress deviation by using a Kalman filtering algorithm, and outputting a final scheduling instruction. According to the invention, optimization of municipal road construction section scheduling is realized, the construction period is effectively shortened, resource waste is reduced, and traffic jam is relieved.

Description

Technical Field

[0001] The present invention relates to the technical field of municipal road construction, and specifically to an optimization method and system for the scheduling of municipal road construction sections based on artificial intelligence. Background Art

[0002] With the acceleration of the urbanization process, the scale of municipal road construction has been continuously expanding. Municipal road construction not only relates to the improvement of urban infrastructure, but also has an important impact on urban traffic, residents' lives, and urban economic development. However, the current scheduling of municipal road construction sections faces many problems, seriously restricting the construction efficiency and quality.

[0003] In terms of the construction period, traditional construction scheduling methods lack comprehensive consideration and accurate prediction of multiple factors. The construction process is easily interfered by external factors such as weather and traffic, and the connection between each construction section is unreasonable. When encountering bad weather, if the construction plan cannot be adjusted in time, it will lead to construction stagnation and project delays; improper arrangement of the construction sequence between different construction sections will also cause resource idleness and excessive waiting time, resulting in an extended overall construction period and increased construction costs and time costs.

[0004] The problem of resource waste is prominent. On the one hand, due to inaccurate inventory management of construction materials, there are often situations of material backlog or shortage. Material backlog will occupy a large amount of funds and storage space, and may cause material deterioration and damage due to long-term storage; material shortage will interrupt the construction, reduce the construction efficiency, and increase the emergency procurement cost at the same time. On the other hand, the allocation of construction machinery is unreasonable. Some construction machinery is overused, resulting in increased wear and tear and rising maintenance costs, while some machinery is idle and wasted, resulting in low resource utilization efficiency.

[0005] Traffic congestion is another major challenge faced by municipal road construction. During the construction process, road closures, the passage of construction vehicles, etc. will have a serious impact on the surrounding traffic. Especially in the busy urban center area, traffic congestion caused by construction not only increases the travel time and cost of citizens, but may also lead to traffic accidents and affect the normal operation of the city. Traditional traffic guidance methods mainly rely on manual command and simple traffic control measures, and cannot be dynamically adjusted according to the real-time traffic flow and construction progress, making it difficult to effectively alleviate the negative impact of construction on traffic.

[0006] Most of the existing construction scheduling technologies are based on experience or simple mathematical models and cannot adapt to the complex and changeable situations in municipal road construction. In terms of data acquisition, there is a lack of comprehensive and real-time data collection means, making it difficult to grasp the dynamic information of the construction area; in terms of scheduling decision-making, it is impossible to comprehensively balance multiple objectives such as construction period, resource utilization, and traffic impact. With the rapid development of artificial intelligence technologies, such as machine learning, deep learning, and reinforcement learning, which have achieved remarkable results in many fields, applying them to the scheduling optimization of municipal road construction sections is expected to solve the above problems and achieve the intelligentization and high efficiency of construction scheduling. Summary of the Invention

[0007] The purpose of the present invention is to provide a scheduling optimization method and system for municipal road construction sections based on artificial intelligence to solve the problems raised in the above background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solutions: A building design optimization method based on artificial intelligence, the method includes: Step 1: Collect multi-source dynamic data of the municipal road construction area, including real-time traffic flow, geographical location of construction sections, status of construction machinery, inventory of construction materials, and weather information, obtain data through Internet of Things devices and satellite remote sensing technology, and use historical construction logs to build a time series database; Step 2: Establish a multi-objective optimization model based on the collected data, with the optimization objectives of minimizing construction period, resource waste, and traffic congestion impact, and use the non-dominated sorting genetic algorithm to generate an initial scheduling plan; Step 3: Dynamically adjust the priorities of construction sections through the reinforcement learning algorithm, and update the construction task sequence according to the real-time traffic flow prediction results and sudden events; Step 4: Use a graph neural network to construct a spatio-temporal conflict detection model for construction sections, identify resource occupancy conflicts and spatio-temporal overlapping areas between construction sections, and generate conflict-free scheduling constraint conditions; Step 5: Optimize resource allocation based on the mixed integer linear programming method, combine the movement path planning of construction machinery and the material transportation route, and generate a global resource scheduling plan; Step 6: Integrate digital twin technology to simulate the construction process, combine the Kalman filter algorithm to correct the construction progress deviation in real time, and output the final scheduling instructions after dynamic adjustment.

[0009] Preferably, in the step 1, the multi-source dynamic data is preprocessed by edge computing nodes, outliers are removed, and the long short-term memory network is used to predict the traffic flow change trend in the next 24 hours, and it is superimposed with the satellite remote sensing geographic information data to generate a three-dimensional situation map of the construction area.

[0010] Preferably, in the multi-objective optimization model of step 2, the constraint conditions include the maximum load of construction machinery, the material supply cycle, and the traffic control time window, and the candidate solution set is screened through Pareto front analysis.

[0011] Preferably, in step 3, the reinforcement learning algorithm uses a deep Q-network (DQN). The state space includes the remaining construction period of the construction section, the resource occupancy rate, and the traffic congestion index. The reward function is designed as a weighted combination of construction efficiency improvement and traffic impact reduction.

[0012] Preferably, in the spatio-temporal conflict detection model of step 4, the nodes of the graph neural network represent the attributes of the construction section, and the edges represent the spatio-temporal dependence relationship between the construction sections. The conflict pattern is identified through node embedding and graph attention mechanism.

[0013] Preferably, in the resource allocation optimization of step 5, an improved ant colony algorithm is introduced to optimize the mechanical path, and the pheromone evaporation coefficient is dynamically adjusted to balance global search and local convergence, and the Hungarian algorithm is combined to match equipment with tasks.

[0014] Preferably, the digital twin model in step 6 evaluates the construction interruption risk through Monte Carlo tree search and generates an emergency plan.

[0015] Preferably, the node expansion strategy of the Monte Carlo tree search adopts the Bayesian optimization method, preferentially explores high-uncertainty regions, and updates the construction risk assessment value through backpropagation.

[0016] Preferably, it further includes a distributed data collaboration mechanism based on federated learning, and the local models of each construction section update the global scheduling strategy through encrypted gradient aggregation.

[0017] Preferably, the present invention further includes an artificial intelligence-based municipal road construction section scheduling optimization system, and the system includes: A data acquisition module for collecting multi-source dynamic data of the municipal road construction area through Internet of Things devices and satellite remote sensing technology, including real-time traffic flow, construction section geographical location, construction machinery status, construction material inventory, and weather information, and constructing a time series database using historical construction logs; A multi-objective optimization model construction module that establishes a multi-objective optimization model based on the collected data, with the optimization objectives of minimizing the construction period, resource waste, and traffic congestion impact, and uses a non-dominated sorting genetic algorithm to generate an initial scheduling plan; A construction section priority dynamic adjustment module that updates the construction task sequence according to the real-time traffic flow prediction results and sudden events through a reinforcement learning algorithm to realize the dynamic adjustment of the construction section priority; The spatio-temporal conflict detection module uses a graph neural network to construct a spatio-temporal conflict detection model for construction sections, identify resource occupancy conflicts and spatio-temporal overlap areas between construction sections, and generate conflict-free scheduling constraints; The resource allocation optimization module optimizes resource allocation based on the mixed integer linear programming method combined with the moving path planning of construction machinery and the material transportation route, and generates a global resource scheduling plan; The construction process simulation and regulation module integrates digital twin technology to simulate the construction process, combines the Kalman filter algorithm to correct the construction progress deviation in real time, and outputs the final scheduling instructions after dynamic adjustment.

[0018] Compared with the prior art, the beneficial effects of the present invention are: In terms of the construction period, through multi-source dynamic data collection and accurate prediction models, the reasonable formulation and dynamic adjustment of the construction plan are realized. The Internet of Things devices and satellite remote sensing technology are used to collect real-time traffic flow, weather information, etc., a time series database is constructed by combining historical construction logs, and the long short-term memory network is used to predict the future traffic flow change trend to plan construction tasks in advance. When encountering bad weather, the system can adjust the construction sequence in time according to the weather prediction, give priority to arranging indoor or weather-independent construction tasks, and avoid construction delays caused by weather reasons. By using the reinforcement learning algorithm to dynamically adjust the priority of construction sections, the construction task sequence is updated according to the real-time traffic flow prediction results and sudden events, ensuring that the construction process is compact and efficient, greatly shortening the overall construction period, improving the project delivery speed, and reducing the time cost.

[0019] The resource utilization efficiency is significantly improved. In the process of resource allocation optimization, based on the mixed integer linear programming method, combined with the moving path planning of construction machinery and the material transportation route, the reasonable allocation of resources is realized. The improved ant colony algorithm is introduced to optimize the mechanical path, and the pheromone evaporation coefficient is dynamically adjusted, enabling construction machinery to move more efficiently between different construction sections, reducing the empty driving time and energy consumption of the machinery. Precise construction material inventory management, combined with real-time construction progress information, avoids material backlog and shortage, reduces material costs and waste. By reasonably matching equipment with tasks, the utilization rate of construction machinery is improved, the idle and excessive wear of equipment are reduced, the service life of equipment is extended, and the construction cost is reduced as a whole.

[0020] In terms of alleviating traffic congestion, the present invention has achieved remarkable results. The system collects real-time traffic flow data and dynamically adjusts the construction task sequence by combining reinforcement learning algorithms, giving priority to constructing sections with less impact on traffic. During peak traffic hours, the passage of large construction vehicles is reduced to avoid traffic congestion on the roads around the construction area. A spatio-temporal conflict detection model for construction sections is constructed using graph neural networks to identify resource occupancy conflicts and spatio-temporal overlapping areas between construction sections, generating conflict-free scheduling constraints, reasonably planning the construction site and construction time, and reducing the interference of construction on traffic. By simulating the construction process using digital twin technology, the impact of construction on traffic is evaluated in advance, and corresponding traffic diversion plans are formulated, effectively ensuring the normal operation of urban traffic, reducing the travel time and costs of citizens, and lowering the incidence of traffic accidents.

[0021] The present invention integrates digital twin technology and Kalman filtering algorithms to achieve real-time monitoring and precise control of the construction progress. The digital twin model is mapped in real time with the actual construction scenario, and construction personnel and managers can intuitively understand various state changes during the construction process. The Kalman filtering algorithm uses historical data and current observation data of the construction progress to correct the construction progress deviation in real time, ensuring the smooth progress of construction according to the plan. Based on the distributed data collaboration mechanism of federated learning, local models of each construction section update the global scheduling strategy through encrypted gradient aggregation, realizing data collaborative optimization while protecting data privacy, and improving the overall performance and adaptability of the system. The present invention provides an intelligent and efficient solution for the construction scheduling of municipal roads, with broad application prospects and significant social and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is the working principle diagram of the construction section scheduling optimization method described in the present invention; Figure 2 It is the flow schematic diagram of adjusting the priority of construction sections by reinforcement learning; Figure 3 It is the working flow schematic diagram of the spatio-temporal conflict detection model for construction sections. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to Figures 1-3, the present invention provides an optimization method for the scheduling of municipal road construction sections based on artificial intelligence, aiming to improve the quality and efficiency of building design and achieve the optimization of building performance. The overall implementation plan is as follows: Data collection and database construction: With the help of Internet of Things devices, various sensors are deployed in the municipal road construction area. Traffic flow monitoring sensors are installed at key road nodes to obtain real-time traffic flow data; status monitoring sensors are equipped on construction machinery to keep track of the operating status of construction machinery in real time, such as working hours, equipment temperature, fuel consumption, etc.; inventory monitoring devices are set up in the construction material warehouse to accurately record information such as the inventory quantity of construction materials and the time of incoming and outgoing. Satellite remote sensing technology is used to obtain the geographical location information of the construction section, and at the same time, weather information is collected, including temperature, humidity, precipitation probability, etc. In addition, historical construction logs are sorted out, and information such as construction progress, resource usage, and problems encountered are structured to build a time series database. This database provides rich historical data support for subsequent data analysis and model training, facilitating the exploration of potential laws in the construction process.

[0025] Construction section multi-objective optimization model construction and initial scheduling plan generation: Based on the collected data, a multi-objective optimization model is constructed. This model aims to minimize the construction period, resource waste, and the impact of traffic congestion. Shortening the construction period can speed up project delivery and reduce time costs; reducing resource waste can improve resource utilization rate and reduce construction costs; reducing the impact of traffic congestion can ensure the normal operation of urban traffic and reduce interference with citizens' lives. The non-dominated sorting genetic algorithm (NSGA-II) is used to solve the model and generate an initial scheduling plan. The NSGA-II algorithm simulates the natural evolution process, and through genetic operations such as selection, crossover, and mutation, it searches for the optimal solution in the solution space, which can effectively handle multi-objective optimization problems and provide a basic plan for subsequent optimization and adjustment.

[0026] Dynamic adjustment of construction section priorities: Using reinforcement learning algorithms, according to the real-time traffic flow prediction results and sudden events, the priorities of construction sections are dynamically adjusted, and then the construction task sequence is updated. Real-time traffic flow prediction can predict the possibility and degree of traffic congestion in advance, and sudden events such as traffic accidents and bad weather will have an unexpected impact on construction and traffic. By adjusting the priorities of construction sections in a timely manner, construction tasks with less impact on traffic or that can be completed quickly are given priority, ensuring the smooth progress of construction while minimizing the impact on traffic.

[0027] Spatial-temporal conflict detection of construction sections: Use graph neural networks to construct a spatial-temporal conflict detection model for construction sections. This model regards construction sections as nodes, and the node attributes cover information such as the geographical location, construction time, and required resources of the construction sections; the spatial-temporal dependence relationships between construction sections are represented by edges. For example, if two construction sections overlap in time and use the same resources, there is a spatial-temporal dependence edge between them. Through node embedding technology, the complex attributes of construction sections are mapped to a low-dimensional vector space for easy model processing; the graph attention mechanism is used to enable the model to automatically focus on key information, accurately identify resource occupancy conflicts and spatial-temporal overlap areas between construction sections, generate conflict-free scheduling constraints, avoid conflicts during construction, and ensure the orderly progress of construction.

[0028] Resource allocation optimization and generation of global resource scheduling plan: Optimize resource allocation based on the mixed-integer linear programming method. Comprehensively consider the movement path planning of construction machinery and the material transportation route to achieve reasonable allocation and efficient utilization of resources. The movement path of construction machinery and the material transportation route will affect construction efficiency and cost. By optimizing these factors, the transportation cost can be reduced and construction efficiency can be improved. Finally, a global resource scheduling plan is generated to ensure the reasonable allocation of various resources during construction.

[0029] Construction process simulation and regulation: Integrate digital twin technology to simulate the construction process. The digital twin model is mapped in real time with the real construction scenario and can intuitively display various state changes during the construction process. Combine the Kalman filter algorithm to correct the construction progress deviation in real time. The Kalman filter algorithm uses the historical data and current observation data of the construction progress to make an optimal estimate of the construction progress state, and timely discovers and corrects the progress deviation. According to the simulation and correction results, the final scheduling instructions after dynamic adjustment are output to achieve precise regulation of the construction process and ensure the smooth completion of the construction according to the plan.

[0030] The present invention will be further described below in conjunction with Embodiments 1 to 5: Embodiment 1: In terms of data collection, in addition to Internet of Things devices and satellite remote sensing technology, mobile monitoring devices are introduced. For some areas with complex traffic flow changes or insufficient coverage of monitoring devices, portable traffic monitoring devices installed on mobile vehicles are used for dynamic monitoring. These devices can collect data such as traffic flow and vehicle speed during movement in real time, supplement the data blind spots of fixed monitoring points, and make the traffic flow data more comprehensive and accurate. In the data preprocessing process, for the elimination of outliers, the 3σ criterion based on statistical principles is combined with the isolation forest algorithm. The 3σ criterion regards the data that deviates from the mean by more than 3 times the standard deviation as outliers by calculating the mean and standard deviation of the data; the isolation forest algorithm starts from the distribution structure of the data to identify the data points that are in an isolated state in the data space. The combination of the two methods can more accurately eliminate various types of outliers.

[0031] When predicting the changing trend of traffic flow in the next 24 hours, the long short-term memory network (LSTM) is improved. An attention mechanism is added between the input layer and the hidden layer of the LSTM. The calculation formula of the attention mechanism is as follows:

[0032] Among them, is the attention score, indicating the degree of association between the previous moment of the hidden layer and the th element in the input sequence; is the normalized attention weight; is the input weighted by the attention mechanism. Through the attention mechanism, the model can automatically focus on the important information at different time steps in the traffic flow data, such as the flow characteristics during the morning and evening rush hours on weekdays and special periods such as holidays, improving the accuracy of prediction. Finally, when overlaying the predicted traffic flow data with satellite remote sensing geographic information data to generate a three-dimensional situation map of the construction area, virtual reality (VR) technology is used to enable construction workers and managers to view the three-dimensional situation map in an immersive manner, more intuitively understand the traffic conditions and geographical environment of the construction area, and facilitate decision-making.

[0033] Embodiment 2: This embodiment improves the multi-objective optimization model and the generation process of the initial scheduling plan, enhances the accuracy of the model and the rationality of the initial scheduling plan, and better balances the relationship among the construction period, resource waste, and the impact of traffic congestion.

[0034] In setting the constraint conditions of the multi-objective optimization model, in addition to the maximum load of construction machinery, the material supply cycle, and the traffic control time window, the spatial limitation of the construction site is added. Due to the limited space of the municipal road construction site, different construction activities may conflict due to insufficient site space. For example, the parking and operation of large construction machinery require a certain range of space. If multiple construction sections carry out operations simultaneously in a narrow area, it may cause the machinery to malfunction. Therefore, the spatial limitation of the construction site is used as a constraint condition to ensure the spatial feasibility of the construction process. When constructing the multi-objective optimization model, the ε-constraint method is combined with the NSGA-II algorithm. The ε-constraint method transforms the multi-objective optimization problem into a single-objective optimization problem. By setting the value range (ε value) of one objective, the other objectives are treated as constraint conditions. Taking the minimization of the construction period as the main objective, the resource waste and the impact of traffic congestion are limited within a certain range, and then the NSGA-II algorithm is used to solve it. When screening the candidate solution set through Pareto front analysis, the crowding distance sorting is introduced. The crowding distance is used to measure the distance between individuals on the Pareto front. The larger the distance, the more dispersed the individual distribution. When selecting individuals, those with a large crowding distance are preferred, which can ensure that the candidate solution set has better diversity and avoid the solution set being too concentrated in a certain area, thus providing more choices for generating a more reasonable initial scheduling plan in the future.

[0035] Embodiment 3: In this embodiment, the reinforcement learning algorithm is optimized to enable it to more accurately adjust the construction section priority dynamically according to the actual construction situation, improve the construction efficiency and reduce the impact of traffic congestion.

[0036] In the design of the state space of the deep Q-network (DQN), in addition to the remaining construction period of the construction section, the resource occupancy rate, and the traffic congestion index, the congestion situation of the alternative routes around the construction section is added as a state information. When a road is congested due to construction, the congestion situation of the surrounding alternative routes will affect the impact of the construction on the overall traffic. If the alternative routes are also severely congested, then the negative impact of the construction of this construction section on traffic will be greater and should be considered when adjusting the priority.

[0037] In the design of the reward function, the measurement indicators for improving construction efficiency and reducing traffic impact are further refined. For improving construction efficiency, the construction progress completion rate indicator is introduced. The construction progress completion rate , where is the actual completed engineering quantity of the current construction section, is the planned completed engineering quantity. The part of the reward function for improving construction efficiency is , is the construction progress completion rate at the previous moment, is the weight coefficient. For traffic impact reduction, the traffic delay time caused by construction is considered. The traffic delay time , is the actual driving time of the vehicle on the section affected by construction, is the free driving time of the section without construction, is the number of vehicles counted. The part of traffic impact reduction in the reward function is , is the maximum allowable traffic delay time, is the weight coefficient.

[0038] The improved reward function . During the training process of DQN, a prioritized experience replay mechanism is adopted. Prioritized experience replay samples experiences according to the importance of the experiences, and the experiences with higher importance have a greater probability of being sampled. The importance of the experiences is measured by calculating the TD error (temporal difference error) of the experiences. The larger the TD error, the more important the experience. This can accelerate the learning speed of the model and make the model converge to a better strategy faster.

[0039] Example 4: In terms of the node representation of the graph neural network, in addition to the basic attributes of the construction section, the construction process complexity information of the construction section is added. Different construction processes, such as road surface paving, pipeline installation, etc., have different degrees of complexity, and the time, resources, and space requirements for construction are also different. Incorporating the construction process complexity into the node representation enables the model to more comprehensively consider the actual situation of the construction section when detecting conflicts. The construction process complexity can be obtained through quantitative evaluation of factors such as the number of steps and technical difficulty of the construction process. In the implementation of the graph attention mechanism, an adaptive graph attention mechanism is adopted. The adaptive graph attention mechanism can dynamically adjust the attention weights according to the connection strength and feature similarity between nodes. For node pairs with strong connections and similar features, higher attention weights are given; for node pairs with loose connections or large feature differences, the attention weights are reduced. Through the adaptive graph attention mechanism, the model can more accurately capture the spatio-temporal dependence relationship between construction sections and improve the accuracy of conflict detection.

[0040] Example 5: In this example, the resource allocation optimization in step 5 and the construction process simulation link in step 6 are optimized to achieve efficient resource allocation, improve the accuracy of construction process simulation and the ability to respond to risks, and enhance the overall performance of the system through a distributed data collaboration mechanism.

[0041] In the resource allocation optimization, when improving the ant colony algorithm to optimize the mechanical path, an adaptive strategy is adopted to dynamically adjust the pheromone evaporation coefficient. The pheromone evaporation coefficient is adjusted according to the number of iterations of the algorithm and the quality of the current solution 。At the initial stage of the algorithm, to fully explore the solution space, a smaller value is set, such as ; as the number of iterations increases, when the quality of the solution tends to be stable, the value is gradually increased, such as , where is the current number of iterations, is the maximum number of iterations, and are the minimum and maximum values of the preset pheromone evaporation coefficient respectively. This can balance global search and local convergence and improve the effect of path optimization.

[0042] When combining the Hungarian algorithm to match devices with tasks, consider the maintenance cycle and maintenance cost of the devices. For devices whose maintenance cycle is about to arrive, try to assign shorter and simpler tasks so that the work can be efficiently completed before maintenance; at the same time, incorporate the maintenance cost into the matching consideration factors, and preferentially select devices with low maintenance costs to complete tasks to reduce the overall cost.

[0043] In the construction process simulation, when the digital twin model evaluates the construction interruption risk through Monte Carlo tree search, the node expansion strategy of Monte Carlo tree search adopts the Bayesian optimization method. The Bayesian optimization method determines the expanded nodes by constructing a probability model of the objective function (such as a Gaussian process model) and calculating the expected improvement value (EI) of each node. The calculation formula of the expected improvement value is:

[0044] where, is the predicted mean value of the objective function at point , is the currently known optimal objective function value, is the predicted standard deviation of the objective function at point , and are the cumulative distribution function and probability density function of the standard normal distribution respectively. By preferentially exploring high-uncertainty regions (i.e., regions with large EI values), the construction interruption risk can be more effectively evaluated. When updating the construction risk assessment value through backpropagation, not only consider the results of the current simulation, but also combine historical simulation data for weighted update to make the risk assessment more accurate. At the same time, based on the distributed data collaboration mechanism of federated learning, the local models of each construction section update the global scheduling strategy through encrypted gradient aggregation. In the process of encrypted gradient aggregation, homomorphic encryption technology is adopted. Homomorphic encryption allows specific calculations to be performed on encrypted data, and the calculation results are the same as those obtained by performing the same calculations on plaintext data after decryption. For example, for the gradients and calculated by the local models of two construction sections, first perform homomorphic encryption on them to obtain and , the server calculates under the encrypted state , and then sends the encrypted result back to the local models of each construction section. After decryption by the local models, the aggregated gradients are obtained for updating the global scheduling policy, realizing data collaborative optimization while protecting data privacy.

[0045] The present invention further includes a scheduling optimization system for municipal road construction sections based on artificial intelligence, and the system includes: A data acquisition module, configured to collect multi-source dynamic data of the municipal road construction area through Internet of Things devices and satellite remote sensing technology, including real-time traffic flow, construction section geographical location, construction machinery status, construction material inventory, and weather information, and construct a time-series database by using historical construction logs; A multi-objective optimization model construction module, which establishes a multi-objective optimization model based on the collected data, takes minimizing the construction period, resource waste, and traffic congestion impact as the optimization objectives, and uses the non-dominated sorting genetic algorithm to generate an initial scheduling plan; A construction section priority dynamic adjustment module, which updates the construction task sequence according to the real-time traffic flow prediction result and sudden events through a reinforcement learning algorithm, realizing the dynamic adjustment of the construction section priority; A spatio-temporal conflict detection module, which uses a graph neural network to construct a construction section spatio-temporal conflict detection model, identifies resource occupancy conflicts and spatio-temporal overlapping areas between construction sections, and generates conflict-free scheduling constraint conditions; A resource allocation optimization module, which optimizes resource allocation based on the mixed integer linear programming method in combination with the moving path planning of construction machinery and the material transportation route, and generates a global resource scheduling plan; A construction process simulation and regulation module, which integrates digital twin technology to simulate the construction process, combines the Kalman filter algorithm to correct the construction progress deviation in real time, and outputs the final scheduling instructions after dynamic adjustment.

[0046] The implementation manner of this system refers to the above-mentioned embodiments and will not be elaborated in the description.

[0047] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0048] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A municipal road construction section scheduling optimization method based on artificial intelligence, characterized in that: The following steps are involved: Step 1: Collect multi-source dynamic data of municipal road construction areas, including real-time traffic flow, geographical location of construction sections, status of construction machinery, inventory of construction materials, and weather information. Obtain data through IoT devices and satellite remote sensing technology, and use historical construction logs to build a time series database. Step 2: A multi-objective optimization model is established based on the collected data, with the optimization objectives of minimizing the construction period, resource waste and traffic congestion, and a non-dominated sorting genetic algorithm is used to generate an initial scheduling plan; Step 3: Dynamically adjust the construction section priority through reinforcement learning algorithm and update the construction task sequence according to the real-time traffic flow prediction results and emergencies; Step 4: Use graph neural network to build a construction section spatiotemporal conflict detection model to identify resource occupation conflicts and spatiotemporal overlap areas between construction sections, and generate conflict-free scheduling constraints; Step 5: Optimize resource allocation based on the mixed integer linear programming method, combine the mobile path planning of construction machinery and the material transportation route, and generate a global resource scheduling plan; Step 6: Integrate digital twin technology to simulate the construction process, combine the Kalman filter algorithm to correct the construction progress deviation in real time, and output the final scheduling instructions after dynamic adjustment.

2. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: In step 1, multi-source dynamic data is preprocessed by edge computing nodes, and after outliers are removed, a long short-term memory network is used to predict the traffic flow change trend in the next 24 hours, and the data is superimposed with satellite remote sensing geographic information data to generate a three-dimensional situation map of the construction area.

3. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: In the multi-objective optimization model of step 2, the constraints include the maximum load of construction machinery, the material supply cycle, and the traffic control time window, and the candidate solution set is screened through Pareto front analysis.

4. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: In step 3, the reinforcement learning algorithm adopts a deep Q network, the state space includes the remaining construction period of the construction section, resource occupancy rate and traffic congestion index, and the reward function is designed as a weighted combination of improved construction efficiency and reduced traffic impact.

5. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: In the spatiotemporal conflict detection model of step 4, the nodes of the graph neural network represent the attributes of the construction sections, the edges represent the spatiotemporal dependencies between the construction sections, and the conflict patterns are identified through node embedding and graph attention mechanism.

6. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: In the resource allocation optimization of step 5, an improved ant colony algorithm is introduced to optimize the mechanical path, the pheromone volatility coefficient is dynamically adjusted to balance the global search and local convergence, and the Hungarian algorithm is combined to match equipment and tasks.

7. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: The digital twin model in step 6 evaluates the risk of construction interruption through Monte Carlo tree search and generates an emergency plan.

8. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 7 is characterized in that: The node expansion strategy of the Monte Carlo tree search adopts the Bayesian optimization method, preferentially explores high uncertainty areas, and updates the construction risk assessment value through back propagation.

9. The municipal road construction section scheduling optimization method based on artificial intelligence according to claim 1 is characterized in that: It also includes a distributed data collaboration mechanism based on federated learning, and the local model of each construction section updates the global scheduling strategy through encrypted gradient aggregation.

10. A municipal road construction section scheduling optimization system based on artificial intelligence, characterized in that: include: The data collection module is used to collect multi-source dynamic data of municipal road construction areas through IoT devices and satellite remote sensing technology, including real-time traffic flow, geographical location of construction sections, status of construction machinery, inventory of construction materials and weather information, and build a time series database using historical construction logs; The multi-objective optimization model building module builds a multi-objective optimization model based on the collected data, takes minimizing the construction period, resource waste and traffic congestion as the optimization goal, and uses the non-dominated sorting genetic algorithm to generate the initial scheduling plan; The construction section priority dynamic adjustment module uses a reinforcement learning algorithm to update the construction task sequence according to real-time traffic flow prediction results and unexpected events, thereby dynamically adjusting the construction section priority; The spatiotemporal conflict detection module uses graph neural networks to build a spatiotemporal conflict detection model for construction sections, identify resource occupation conflicts and spatiotemporal overlap areas between construction sections, and generate conflict-free scheduling constraints; Resource allocation optimization module, based on mixed integer linear programming method combined with construction machinery movement path planning and material transportation route, optimizes resource allocation and generates a global resource scheduling plan; The construction process simulation and control module integrates digital twin technology to simulate the construction process, combines the Kalman filter algorithm to correct the construction progress deviation in real time, and outputs the final scheduling instructions after dynamic adjustment.

Images