Integrated multi-dimensional monitoring method and system for construction site

Abstract

The invention relates to the technical field of building construction monitoring and intelligent management, in particular to an integrated multi-dimensional monitoring method and system for a construction site, and the method comprises the steps: collecting multi-dimensional data of a construction site environment through a multi-spectral imaging technology, a sound wave detection technology and an LIDAR technology; dynamic analysis and prediction are carried out through a multistage space-time diagram convolutional network and an adaptive Bayesian-Markov model, and comprehensive environment prediction data are generated; resource allocation and strategy execution of construction site monitoring nodes are optimized by using an interactive dynamic game model, and efficient operation of the system in a complex construction site environment is ensured; a monitoring strategy is continuously optimized through a multi-level intelligent feedback network, the response speed and accuracy of the system are enhanced, and the safety and efficiency of construction site operation are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of building construction monitoring and intelligent management, and in particular to an integrated multi-dimensional monitoring method and system for construction sites. Background Art

[0002] With the expansion of construction site scale and the increase of construction complexity, traditional single monitoring methods can no longer meet the needs of all-round and all-weather safety management of construction sites. In order to ensure the safety and efficiency of the construction process, it is necessary to establish a monitoring system that covers all aspects, responds quickly and can perform multi-dimensional data analysis. Through multi-dimensional monitoring, not only can the dynamics of the construction site be monitored in real time, but also potential safety hazards can be predicted and prevented to ensure the smooth operation of the construction site.

[0003] Existing construction site monitoring systems (China invention patent, publication number: CN118055214A, name: An integrated intelligent monitoring system, method and device for a smart construction site) usually use video monitoring, sensor monitoring and other technical means to collect and analyze data. However, these systems have many shortcomings in practical applications: In the existing technology, the data between the video surveillance system and the sensor system are relatively independent and lack effective cross-system integration; video surveillance is mainly used for monitoring personnel and equipment, while the sensor system focuses on collecting environmental data. There is a lack of effective linkage mechanism between the two, resulting in low data utilization efficiency; Although the existing system can collect a large amount of monitoring data, it still relies on simple rules and threshold judgments in data analysis and lacks intelligent analysis capabilities, making it difficult to accurately predict and manage risks in complex construction site environments; When faced with emergencies, existing systems can usually only rely on preset alarm mechanisms. They lack the ability to adapt to dynamically changing environments and are unable to adjust monitoring strategies in a timely manner, resulting in a slow emergency response. Summary of the invention

[0004] In view of the many problems existing in the above-mentioned prior art, the present invention provides an integrated multi-dimensional monitoring method and system for construction sites. The present invention realizes comprehensive monitoring and real-time adjustment of the construction site environment through advanced data acquisition technology, multi-level intelligent analysis model and dynamic optimization algorithm. Through multi-spectral imaging, acoustic wave detection and LIDAR and other technologies, the system can comprehensively collect multi-dimensional data of the construction site, and perform data analysis and prediction through multi-level spatiotemporal graph convolutional networks and adaptive Bayes-Markov models, and finally realize the optimal allocation of resources and strategy adjustment through interactive dynamic game models to ensure the safety and efficiency of construction site operation.

[0005] An integrated multi-dimensional monitoring method for a construction site comprises the following steps: Through multi-spectral imaging technology, acoustic wave detection and analysis technology and LIDAR technology, environmental data, acoustic wave characteristic data and three-dimensional spatial data are collected at key locations on the construction site, and the collected data are pre-processed and formatted to generate a processed comprehensive data set; The processed comprehensive data set is input into a multi-level spatiotemporal graph convolutional network to perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data. The multi-level spatiotemporal correlation data is then subjected to uncertainty analysis and state transition prediction through an adaptive Bayesian-Markov model to generate comprehensive environmental prediction data. Based on comprehensive environmental prediction data, the interactive dynamic game model is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes, generate global collaborative strategy data, and perform monitoring tasks according to the global collaborative strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; During the execution of the monitoring task, real-time monitoring feedback data is collected and processed, and multi-level iterative optimization is performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating the global collaborative strategy data.

[0006] Preferably, the preprocessing and formatting include the following steps: denoising the environmental data through a filtering algorithm to eliminate environmental noise interference; normalizing the sound wave feature data to standardize the signal strength of different sound sources; converting the three-dimensional spatial data format and converting the point cloud data into a raster data format, thereby generating a comprehensive data set suitable for a multi-level spatiotemporal graph convolutional network.

[0007] Preferably, the multi-level spatiotemporal graph convolutional network includes the following steps: by constructing a multi-layer convolutional neural network, multi-level spatiotemporal correlation modeling is performed on the processed comprehensive data set, wherein each layer of the convolutional network processes data of different time scales and spatial resolutions respectively to extract dynamic change characteristics in the construction site environment, and finally generate multi-level spatiotemporal correlation data containing multi-dimensional spatiotemporal relationships.

[0008] Preferably, the adaptive Bayesian-Markov model performs state transition analysis on multi-level spatiotemporal correlation data by the following steps: Based on historical observation data and current observation data, the Bayesian inference method is used to recursively update the state transition probability and uncertainty parameters; Based on the state transition model of Markov chain, the construction site environmental status at future moments is predicted, thereby generating comprehensive environmental prediction data including short-term and medium-term prediction results.

[0009] Preferably, the state transition probability of the adaptive Bayesian-Markov model is calculated by the following formula:

[0010] in, Indicates at time state Transfer to time state probability; Indicates that in observation Down state The conditional probability of Indicates observation probability; Indicates in status Observation The conditional probability of This formula is used to quantify the state transition relationship in multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data.

[0011] Preferably, the interactive dynamic game model models resource allocation and strategy optimization of construction site monitoring nodes through the following steps: Construct a resource allocation game model, in which the resource allocation decision of each monitoring node is based on the principle of maximizing the global utility function to generate the optimal resource allocation vector; Based on the output of the resource allocation game model, a strategy optimization game model is further constructed. The strategy selection vector of each monitoring node is generated through a dynamic game process, and finally the global collaborative strategy data is generated.

[0012] Preferably, the interactive dynamic game model determines the optimal resource allocation in the resource allocation game by the following formula:

[0013] in, Indicates The resource allocation vector of nodes; Indicates The strategy selection vector of nodes; Indicates The utility function of each node; Indicates the total amount of available resources; This formula is used to optimize resource allocation among monitoring nodes and generate global coordination strategy data.

[0014] Preferably, during the execution of the monitoring task, the following iterative optimization steps are performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network: In the first-level feedback mechanism, preliminary feedback optimization data is generated based on the current monitoring results; In the second-level feedback mechanism, the preliminary feedback optimization data is compared and analyzed with the historical monitoring data to generate the second feedback optimization data; In the third-layer feedback mechanism, the abnormal situation handling records and long-term trend analysis are combined to generate multi-level feedback optimization data for adjusting the parameters and strategies of the construction site monitoring nodes.

[0015] Preferably, the multi-level intelligent feedback network includes multiple feedback levels, each level analyzes and optimizes data of different dimensions, wherein the first level is for current monitoring results, the second level is for historical monitoring data, and the third level is for exception handling records, thereby generating comprehensive feedback optimization data for further optimizing global collaborative strategy data.

[0016] A system for implementing the integrated multi-dimensional monitoring method for a construction site, comprising: The multispectral imaging module, the acoustic wave detection and analysis module and the LIDAR module are used to collect environmental data, acoustic wave characteristic data and three-dimensional spatial data at key locations on the construction site, and pre-process and format the environmental data, acoustic wave characteristic data and three-dimensional spatial data to generate a processed comprehensive data set; A multi-level spatiotemporal graph convolutional network module, which is used to receive the processed comprehensive data set and perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data; An adaptive Bayesian-Markov model module, used for performing uncertainty analysis and state transition prediction on the multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data; An interactive dynamic game model module is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes based on the comprehensive environmental prediction data, generate global collaborative strategy data, and perform monitoring tasks according to the global collaborative strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; The multi-level intelligent feedback network module is used to collect and process real-time monitoring feedback data during the execution of monitoring tasks, and to perform multi-level iterative optimization on real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating global collaborative strategy data.

[0017] Compared with the prior art, the advantages and beneficial effects of the present invention are: Through the combination of multi-spectral imaging technology, acoustic wave detection and analysis technology and LIDAR technology, comprehensive collection and pre-processing of multi-dimensional data is achieved, improving data integration and utilization efficiency; Through multi-level spatiotemporal graph convolutional networks, multi-level spatiotemporal correlation modeling and feature extraction are carried out to accurately capture the dynamic change characteristics of the construction site environment. In combination with the adaptive Bayesian-Markov model, high-precision prediction of future environmental conditions is achieved. Through the interactive dynamic game model, the resource allocation and monitoring strategy of the construction site monitoring nodes are dynamically optimized, achieving real-time response capabilities in complex environments and significantly improving the system's emergency response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 A schematic diagram of spatiotemporal correlation modeling and prediction in the present invention; Figure 3 It is a schematic diagram of resource allocation and strategy optimization in the present invention; Figure 4 It is a schematic diagram of multi-level intelligent feedback in the present invention; Figure 5 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0019] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0020] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0021] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0022] like Figure 1 As shown, an integrated multi-dimensional monitoring method for a construction site includes the following steps: Through multi-spectral imaging technology, acoustic wave detection and analysis technology and LIDAR technology, environmental data, acoustic wave characteristic data and three-dimensional spatial data are collected at key locations on the construction site, and the collected data are pre-processed and formatted to generate a processed comprehensive data set; Preferably, the preprocessing and formatting include the following steps: denoising the environmental data through a filtering algorithm to eliminate environmental noise interference; normalizing the sound wave feature data to standardize the signal strength of different sound sources; converting the three-dimensional spatial data format and converting the point cloud data into a raster data format, thereby generating a comprehensive data set suitable for a multi-level spatiotemporal graph convolutional network.

[0023] Multispectral imaging technology is used to capture environmental data on construction sites. This technology can capture not only visible light information, but also other bands such as near-infrared and infrared by imaging in multiple spectral bands, thereby providing richer environmental data. This data can include parameters directly related to the construction site environment, such as temperature, humidity, and vegetation coverage. In this technical application, multispectral imaging can not only identify changes that are invisible to the naked eye, but also help detect potential risks, such as overheating of equipment or abnormal heat sources, so as to provide early warning.

[0024] Acoustic wave detection and analysis technology is mainly used to capture sound signals on the construction site, including mechanical operation sounds, human activity sounds, and other environmental sounds. By collecting acoustic wave feature data, the system can monitor the normal operation of the construction site and identify abnormal sounds, such as precursors to equipment failures or the occurrence of unexpected events. The advantage of this technology is that it can provide a monitoring method without visual contact in complex construction site environments, especially for those situations where visual monitoring blind spots or limited viewing distances are present.

[0025] In addition, LIDAR technology is used to obtain three-dimensional spatial data of the construction site, measuring the distance of objects by emitting lasers and receiving reflected signals, thereby generating high-precision three-dimensional point cloud data. This data is used to build an accurate three-dimensional model of the construction site, providing real-time information on the site layout, equipment location, and dynamic changes. In the construction site environment, LIDAR technology can quickly and accurately capture site changes, such as material accumulation, equipment movement, or structural changes, thereby providing basic data for subsequent monitoring and analysis.

[0026] In the data preprocessing and formatting steps, first, the environmental data is denoised through a filtering algorithm. This step aims to eliminate noise interference caused by environmental factors (such as wind, rain, dust, etc.), thereby improving the accuracy and reliability of the data. The denoised data can more realistically reflect the actual conditions of the construction site and avoid misjudgments caused by noise.

[0027] For the acoustic feature data, the system performs normalization processing to standardize the signal strength of different sound sources. Since the signal strength and frequency of different sound sources on the construction site may vary greatly, normalization processing can unify the sound data of different sound sources on the same scale, which is convenient for subsequent analysis and processing. This standardization not only improves the comparability of the data, but also enables the system to more effectively detect and distinguish different sound features, such as distinguishing normal operating sounds from potential abnormal sounds.

[0028] The three-dimensional spatial data is converted into a raster data format by converting the point cloud data generated by LIDAR. This process converts irregular point cloud data into a regular raster format, making it easier to combine the data with other two-dimensional or three-dimensional data for processing. The rasterized data can better meet the modeling requirements of the multi-level spatiotemporal graph convolutional network, ensuring that these spatial data can be effectively used in spatiotemporal correlation modeling to generate accurate monitoring results.

[0029] like Figure 2 As shown, the processed comprehensive data set is input into a multi-level spatiotemporal graph convolutional network to perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data, and an adaptive Bayesian-Markov model is used to perform uncertainty analysis and state transition prediction on the multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data; The processed comprehensive data set is input into a multi-level spatiotemporal graph convolutional network. The multi-level spatiotemporal graph convolutional network is a deep learning model for processing spatiotemporal data, which can capture the complex correlation of data in time and space dimensions. In the present invention, the network analyzes and extracts spatiotemporal features in the comprehensive data set through multiple convolutional levels. Each layer of the convolutional network processes data of different time scales and spatial resolutions respectively, ensuring that the system can identify various information from short-term changes to long-term trends. For example, in a construction site environment, this spatiotemporal correlation modeling can identify the movement trajectory of equipment, the gradual changes in environmental parameters, and the dynamic adjustment of the construction site layout. This multi-level analysis enables the system to not only grasp the current status of the construction site, but also model and predict potential trends of change, so as to respond in advance.

[0030] On the basis of spatiotemporal correlation modeling, the system further performs uncertainty analysis and state transition prediction on multi-level spatiotemporal correlation data through an adaptive Bayesian-Markov model. The Bayesian-Markov model is a statistical model for processing uncertainty and state transition. It combines the characteristics of Bayesian inference and Markov chain, and can predict future states based on historical data and current observations. In the present invention, the application of the adaptive Bayesian-Markov model enables the system to dynamically predict the state changes of the construction site environment. Especially in the face of complex and changeable environments, the model can capture the uncertainty in the system by recursively updating the state transition probability. For example, on the construction site, factors such as weather changes, equipment failures, or personnel transfers will introduce environmental uncertainties. The model generates comprehensive environmental prediction data by quantifying these uncertainties and predicting their impacts, thereby providing data support for subsequent decision-making and optimization.

[0031] This prediction capability has significant effects in practical applications. For example, when the system identifies abnormal fluctuations in the running trajectory of a certain device through a multi-level spatiotemporal graph convolutional network, the adaptive Bayes-Markov model can further analyze whether the fluctuation will cause equipment failure and predict the possibility and time of failure. This comprehensive environmental prediction data can not only help site managers take preventive measures in advance to avoid accidents, but also optimize the scheduling and allocation of resources to ensure the stability and efficiency of site operations.

[0032] Preferably, the multi-level spatiotemporal graph convolutional network includes the following steps: by constructing a multi-layer convolutional neural network, multi-level spatiotemporal correlation modeling is performed on the processed comprehensive data set, wherein each layer of the convolutional network processes data of different time scales and spatial resolutions respectively to extract dynamic change characteristics in the construction site environment, and finally generate multi-level spatiotemporal correlation data containing multi-dimensional spatiotemporal relationships.

[0033] The multi-level spatiotemporal graph convolutional network processes the spatiotemporal information in the dataset through multiple convolutional layers. Each layer of the convolutional network processes different time scales and spatial resolutions to ensure that the system can capture a variety of dynamic features in the construction site environment, from micro changes to macro trends. For example, in a multi-layer convolutional network, the first layer may focus on capturing high-frequency changes in a short period of time, such as vibration or noise of mechanical equipment, while the second layer may process low-frequency changes in a longer time span, such as the day-to-night changes in the overall temperature of the construction site or the slow increase in material accumulation. This hierarchical processing method enables the system to effectively extract different levels of spatiotemporal features, which often have important decision-making significance in construction site management.

[0034] In the application of multi-level spatiotemporal graph convolutional networks, the basic principles of convolutional neural networks (CNNs) are further extended to the spatiotemporal dimension. Traditional CNNs are usually used to process two-dimensional image data and capture spatial features through local perception domains. In the present invention, the spatiotemporal convolutional network combines time and space for modeling, so that it can handle continuous changes in the construction site environment. For example, when monitoring the safety status of a construction site, the temperature rise and abnormal changes in the sound wave frequency at a certain moment, combined with historical data, can indicate potential risks of equipment overheating or mechanical failure. Through the processing of multi-layer convolutional networks, the system can not only identify this risk, but also analyze its possible development trend, providing a decision-making basis for construction site management.

[0035] The application of multi-level spatiotemporal graph convolutional networks significantly improves the system's depth of understanding and predictive capabilities for complex construction site environments. Since the processing objectives of each layer of the convolutional network are different, the system can analyze the data on the construction site from multiple angles and levels, thereby generating multi-level spatiotemporal correlation data containing rich information. These data can be used for real-time monitoring, risk warning, and resource optimization in practical applications. For example, the system can identify whether the operating mode of a certain equipment is abnormal by analyzing multi-level spatiotemporal correlation data, thereby timely warning of possible failures and avoiding losses caused by downtime.

[0036] In the specific implementation, suppose the operating data of a piece of equipment on a construction site is input into a multi-level spatiotemporal graph convolutional network. The first layer of the network may capture the high-frequency vibration data of the equipment in a short period of time, indicating that the equipment may have abnormal operation; the second layer of the network processes data over a longer time span and finds that the temperature of the equipment is gradually rising. Through these hierarchical analyses, the system can generate detailed multi-level spatiotemporal correlation data to describe the state change trend of the equipment, and make early warning and scheduling decisions based on this to ensure the safety and efficiency of construction site operations.

[0037] Preferably, the adaptive Bayesian-Markov model performs state transition analysis on multi-level spatiotemporal correlation data by the following steps: Based on historical observation data and current observation data, the Bayesian inference method is used to recursively update the state transition probability and uncertainty parameters; Based on the state transition model of Markov chain, the construction site environmental status at future moments is predicted, thereby generating comprehensive environmental prediction data including short-term and medium-term prediction results.

[0038] The application of Bayesian inference in this model aims to continuously and dynamically model the state of the construction site environment by recursively updating the state transition probability and uncertainty parameters. In a complex construction site environment, there is often a certain degree of uncertainty in the observation of various data, such as the noise generated by the sensor data of the equipment due to external interference, or the fluctuation of the observation data caused by environmental factors (such as weather). The Bayesian inference method can use historical observation data and current observation data to quantify these uncertainties and reflect the real changes in the construction site environment by updating the state transition probability. Through this recursive update mechanism, the system can dynamically adjust its understanding of the construction site status, thereby more accurately capturing the changing trends of the environment.

[0039] The state transition model of the Markov chain is used in the present invention to predict the state of the construction site environment at a future moment. The core idea of ​​the Markov chain is that the prediction of the future state depends only on the current state and has nothing to do with the past state. This state transition model has strong adaptability when dealing with dynamic changes in the construction site environment, especially when facing multiple possible states, it can predict the future state based on the current state and the known transition probability. For example, in construction site management, if an abnormal fluctuation is observed in the operating state of a certain equipment, the system can use the Markov chain model to predict the possible state of the equipment in the future and determine whether it will further deteriorate into a fault state or return to normal.

[0040] Combining these two parts, the comprehensive environmental prediction data generated by the system through the adaptive Bayesian-Markov model not only contains the current state information of the construction site environment, but also provides predictions for the future short-term and medium-term state. This kind of prediction data plays an important guiding role in construction site management, and can help managers make decisions in advance to prevent potential risks. For example, the system may predict that due to continuous high temperature environment, some mechanical equipment may overheat and fail in the next few hours. Based on this prediction, the construction site manager can adjust the workload of the equipment in advance, or arrange regular inspections to avoid downtime and safety accidents caused by equipment failure.

[0041] Preferably, the state transition probability of the adaptive Bayesian-Markov model is calculated by the following formula:

[0042] in, Indicates at time state Transfer to time state probability; Indicates that in observation Down state The conditional probability of Indicates observation probability; Indicates in status Observation The conditional probability of This formula is used to quantify the state transition relationship in multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data.

[0043] In practical applications, this formula first uses the principle of Bayesian inference to update the system's current state based on current observation data and historical state information. Through recursive calculation, these conditional probabilities can dynamically reflect changes in the construction site environment. For example, if the temperature in a certain area of ​​the construction site suddenly rises The system will use this observation data to re-evaluate the current operating status of the device , and calculate the future state of the device through the formula Possible changes, such as further warming, stabilization, or cooling.

[0044] This analysis based on the probability of state transition has important practical effects in construction site management. Through accurate prediction of state transition, managers can better understand and grasp the dynamic changes of construction site equipment or environment, so as to take corresponding measures in advance. For example, when the formula calculates that the probability of equipment failure in the future is high, the system can issue an alarm and suggest managers to inspect or maintain the equipment, thereby avoiding potential downtime risks.

[0045] A typical example is to monitor the temperature data collected by sensors during the concrete pouring process at a construction site. The temperature in a certain area rises rapidly. The system analyzes the current status through an adaptive Bayesian-Markov model. , combined with observation data and historical temperature records, predict the future period There is a risk that cracks may occur in the concrete. If it is higher, the system will recommend that site managers take measures, such as adding cooling or covering measures, to prevent material quality problems.

[0046] like Figure 3 As shown, based on the comprehensive environmental prediction data, the interactive dynamic game model is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes, generate global collaborative strategy data, and perform monitoring tasks according to the global collaborative strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; The comprehensive environmental prediction data is obtained through early data collection and analysis (for example, through multi-level spatiotemporal graph convolutional networks and adaptive Bayesian-Markov models). These data provide dynamic state information of the construction site environment, including possible risk points, equipment working status, distribution of personnel activities, etc. This information is the basis for the system to allocate resources and optimize strategies.

[0047] Interactive dynamic game models play a key role in this process. Game models are a mathematical tool that is widely used to analyze the strategic interactions between multiple participants (monitoring nodes in this case). In the construction site monitoring system, each monitoring node, such as a camera, sensor, etc., can be regarded as a "participant". These nodes need to coordinate their work under limited resource conditions, such as power, bandwidth, and computing resources, to achieve the overall optimal monitoring effect. The dynamic game model allows these nodes to adjust their strategies based on comprehensive environmental prediction data to maximize the overall benefit.

[0048] Specifically, the interactive dynamic game model generates global collaborative strategy data through a two-step optimization process. The first step is resource allocation optimization. The model calculates the utility function of each monitoring node (that is, the monitoring effect that can be achieved under different resource allocation conditions) and determines the optimal resource allocation plan while ensuring that the total amount of resources does not exceed the system's allowable range. For example, if the risk of a certain area of ​​a construction site is higher during a certain period of time, the model may allocate more power and bandwidth to the camera responsible for that area to improve its monitoring quality.

[0049] The second step is monitoring strategy optimization. After the resource allocation is determined, the game model further optimizes the specific strategy of each monitoring node, including the camera's viewing angle adjustment, sensor sensitivity setting, data transmission priority, etc. This step ensures that under the current resource configuration, the monitoring behaviors of all nodes can be coordinated with each other, thereby achieving the best overall monitoring effect.

[0050] The generation of global collaborative strategy data is the final result of the game model calculation, and these data directly guide the system to perform monitoring tasks. The key to the execution process is real-time adjustment, that is, when the monitoring task is in progress, the system can dynamically adjust the parameters and task priorities of the monitoring equipment according to the actual situation and feedback data. For example, if a monitoring node finds that the activities of people in a certain area are frequent and abnormal during the execution of the task, the system may immediately adjust the monitoring perspective of the node, or increase the priority of the data it transmits, to ensure timely response to potential security risks.

[0051] Through the interactive dynamic game model, the present invention can realize the intelligent allocation of construction site monitoring system resources and adaptive optimization of strategies, so that the entire system can always maintain efficient and accurate monitoring capabilities when facing complex and changeable construction site environments. The model can not only adjust the monitoring strategy according to real-time data, but also maximize the monitoring effect under limited resources, reduce resource waste, and improve the overall efficiency and safety of construction site management.

[0052] For example, in a large construction site, if the system predicts that a certain construction area is at risk of collapse in the next few hours, the interactive dynamic game model will prioritize allocating more resources to the equipment responsible for monitoring the area, and adjust the monitoring strategies of these equipment, such as improving the resolution and data collection frequency of the camera, and transmitting its data to the central monitoring system in real time for key analysis. Through this optimization process, the system can not only detect and warn of risks in a timely manner, but also effectively respond to emergencies through flexible resource management and strategy adjustment to ensure safe operation of the construction site.

[0053] Preferably, the interactive dynamic game model models resource allocation and strategy optimization of construction site monitoring nodes through the following steps: Construct a resource allocation game model, in which the resource allocation decision of each monitoring node is based on the principle of maximizing the global utility function to generate the optimal resource allocation vector; Based on the output of the resource allocation game model, a strategy optimization game model is further constructed. The strategy selection vector of each monitoring node is generated through a dynamic game process, and finally the global collaborative strategy data is generated.

[0054] The first step of the interactive dynamic game model is to construct a resource allocation game model. In a construction site monitoring system, resources (such as electricity, bandwidth, computing power, etc.) are limited, and each monitoring node (such as cameras, sensors, etc.) needs these resources to perform its monitoring tasks. The resource allocation game model regards each monitoring node as a "player" in the game, allowing these nodes to make resource allocation decisions based on their own needs and the global goals of the system. Specifically, the goal of resource allocation decisions is to maximize the global utility function, which measures the overall monitoring effect and resource utilization of the system. By calculating the utility value of each node under different resource allocation conditions, the model can generate the optimal resource allocation vector to ensure that the system's resource allocation is optimal on a global scale. For example, in a large construction site, if the monitoring demand in a certain area suddenly increases (such as dense personnel or abnormal equipment), the system will prioritize allocating more resources to the monitoring nodes responsible for that area to improve the monitoring quality.

[0055] After completing resource allocation, the interactive dynamic game model enters the second step, which is to build a strategy optimization game model. At this point, each monitoring node has obtained the corresponding resources, and the next step is to decide how to best use these resources for monitoring. The strategy optimization game model generates a strategy selection vector for each node through dynamic gaming. In this process, nodes will consider each other's strategies and behaviors, and gradually adjust their own strategies through mutual gaming to achieve the overall optimal monitoring effect. For example, during the monitoring process after resource allocation, if a camera finds abnormal activity in the area it is responsible for, while another camera in a neighboring area does not find any abnormality, the two may adjust their respective monitoring directions and focal lengths through the gaming process to cover a larger monitoring range, ensuring that any abnormal activity can be captured in a timely manner.

[0056] Through the synergy of these two steps, the interactive dynamic game model finally generates global collaborative strategy data. These data include the resource allocation plan and specific monitoring strategy of each monitoring node, ensuring that the entire system can operate in the best state. The generation of global collaborative strategy data means that the system not only optimizes the performance of a single node, but more importantly, realizes the collaborative work of all nodes on a global scale, thereby maximizing the overall effectiveness of site monitoring.

[0057] In practical applications, the effect of this process is particularly significant. For example, in a high-risk construction site environment, the monitoring demand around a key device is usually high. Through the resource allocation game model, the system can prioritize the allocation of more resources to the nodes that monitor the device, and at the same time, through the strategy optimization game model, ensure that these nodes can reasonably adjust the monitoring strategy, such as increasing the monitoring frequency, adjusting the camera angle, etc., to fully cover the surrounding area of ​​the device. The global collaborative strategy data finally generated can ensure that even when resources are tight, the system can still monitor key areas efficiently and accurately to avoid potential accidents.

[0058] Preferably, the interactive dynamic game model determines the optimal resource allocation in the resource allocation game by the following formula:

[0059] in, Indicates The resource allocation vector of a node, that is, the amount of resources actually allocated to the node in the limited total amount of resources; Indicates The strategy selection vector of a node represents the specific strategy adopted by the node in the monitoring task, such as the camera angle, focal length setting or sensor sensitivity adjustment; Indicates The utility function of each node, which measures the monitoring effect or benefit that the node can achieve under different resource allocation and strategy selection conditions. Usually, the utility function is proportional to the effect of the node in completing the monitoring task; Indicates the total amount of available resources. It is a constant that indicates the upper limit of resources that can be allocated in the entire monitoring system. This formula is used to optimize resource allocation among monitoring nodes and generate global coordination strategy data.

[0060] The goal of this formula is to maximize the total utility of all monitoring nodes while meeting the total system resource limit. In other words, the system must ensure that the resource utilization of the entire system is optimal while each node obtains the resources it needs. This is crucial for construction site monitoring systems, especially when resources are limited, and it is necessary to ensure that monitoring of key areas is prioritized.

[0061] In practical applications, the complexity of construction site monitoring lies in the different needs and importance of each monitoring node. For example, in a large-scale construction site, some areas may require more monitoring resources such as high-definition cameras and fast-response sensors due to intensive construction activities or high potential risks; while other areas may be relatively stable and require fewer resources. Therefore, the system needs to finely manage the resource allocation of each node to ensure the overall monitoring efficiency.

[0062] By applying the above formula, the system can dynamically adjust the resource allocation of each node. For example, if the system detects that the work activities in the northern area of ​​the construction site suddenly increase during a certain period of time, while the southern area is relatively quiet, the system may adjust the resource allocation vector , more power, bandwidth and other resources are allocated to the monitoring nodes responsible for the northern region to improve the monitoring frequency and clarity of the northern region. Monitoring nodes in the southern region may temporarily reduce resource usage to ensure that total resource usage does not exceed the system's upper limit. .

[0063] The global coordination strategy data generated by this process not only includes the resource allocation plan for each monitoring node, but also the optimal monitoring strategy for these nodes under the current resource conditions. These strategy data play a guiding role in the execution of monitoring tasks, ensuring that the nodes in the system can work in coordination, maximizing the monitoring coverage and response speed, and avoiding resource waste or the appearance of monitoring blind spots.

[0064] Through this optimized resource allocation and strategy adjustment, the construction site monitoring system can continue to operate efficiently in a complex and changing environment. Especially when resources are limited, the system can flexibly dispatch resources to prioritize the monitoring needs of key areas and ensure the safety and efficiency of construction site operations. For example, when a monitoring node near a key device detects a possible fault signal, the system can immediately focus more resources on monitoring that area, strengthen monitoring efforts, and issue early warnings in a timely manner, thereby avoiding potential accidents.

[0065] During the execution of the monitoring task, real-time monitoring feedback data is collected and processed, and multi-level iterative optimization is performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating the global collaborative strategy data.

[0066] like Figure 4 As shown, preferably, during the execution of the monitoring task, the following iterative optimization steps are performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network: In the first-level feedback mechanism, preliminary feedback optimization data is generated based on the current monitoring results; In the second-level feedback mechanism, the preliminary feedback optimization data is compared and analyzed with the historical monitoring data to generate the second feedback optimization data; In the third-layer feedback mechanism, the abnormal situation handling records and long-term trend analysis are combined to generate multi-level feedback optimization data for adjusting the parameters and strategies of the construction site monitoring nodes.

[0067] During the execution of the monitoring task, the system will collect feedback data from each monitoring node in real time. These feedback data include the latest environmental information, equipment operation status, personnel activities, etc. captured by the monitoring node. The collection of real-time monitoring feedback data is the basis for dynamic adjustment of the system, ensuring that the system can timely reflect the latest changes in the construction site environment.

[0068] These real-time monitoring feedback data will be processed and optimized through a multi-level intelligent feedback network. The network consists of multiple feedback mechanisms, each of which has its own specific function in data processing and optimization, gradually enhancing the accuracy and adaptability of feedback data.

[0069] In the first level of feedback mechanism, the system generates preliminary feedback optimization data based on the current monitoring results. This stage mainly analyzes and processes real-time monitoring data quickly. The generated preliminary feedback optimization data can reflect the initial results of the current monitoring task and the parts that need to be adjusted immediately. For example, in a certain area of ​​the construction site, if the real-time data shows that the activity of personnel has increased significantly, the system may adjust the angle of the camera or increase the monitoring frequency at this stage to better cover the area.

[0070] The second-level feedback mechanism further compares and analyzes the initial feedback optimization data with the historical monitoring data to generate more in-depth second feedback optimization data. Through comparison and analysis, the system can identify the difference between the current state and the historical pattern, so as to better predict possible future development trends. For example, if a device gradually shows an abnormal trend in the past few days of monitoring, the comparison of the current monitoring results with the historical data can enable the system to detect the signs of potential failure earlier, so as to take countermeasures in advance.

[0071] In the third-level feedback mechanism, the system combines abnormal situation handling records and long-term trend analysis to generate the final multi-level feedback optimization data. This level of analysis not only considers current and historical data, but also incorporates the system's past experience in handling similar abnormal situations, as well as trend changes reflected in long-term data. This process ensures the comprehensiveness and accuracy of feedback optimization data. For example, based on long-term trend analysis, the system can identify the potential impact of certain environmental changes (such as seasonal climate change) on equipment operation, thereby adjusting the monitoring strategy to prevent these changes from having an adverse impact on site operations.

[0072] The generated multi-level feedback optimization data is used to update the global collaborative strategy data in real time. This means that the system can dynamically adjust the parameters and strategies of the monitoring nodes based on the latest feedback information to ensure that the system is always in the best operating state during the execution of the monitoring task. For example, when monitoring a construction site at night, if the system detects that the lighting conditions are gradually getting worse, the real-time feedback data and historical data will prompt the system to automatically adjust the exposure settings of the camera, and may also adjust the work priorities of the monitoring nodes to ensure that key areas are still fully monitored.

[0073] Through iterative optimization of the multi-level intelligent feedback network, the system can effectively adapt to the rapid changes in the construction site environment and continuously improve the accuracy and response speed of monitoring during the execution of monitoring tasks. This not only improves the system's ability to respond to emergencies, but also ensures the stability and effectiveness of long-term monitoring tasks.

[0074] For example, during nighttime construction at a large construction site, the system discovered changes in light conditions through the first-level feedback mechanism, generated preliminary feedback optimization data, and prompted the need to adjust the camera's sensitivity. Subsequently, the second-level mechanism compared this real-time data with the monitoring data of the past few nighttime constructions, and found that the current light changes were similar to the light change patterns during the previous constructions, thereby generating more detailed feedback optimization data and further adjusting the camera's angle and shooting frequency. Finally, the third-level mechanism combined the experience of nighttime construction in long-term monitoring and predicted that the light changes might be accompanied by abnormal operation of some equipment. Therefore, the system decided to adjust the priority configuration of the monitoring nodes and increase the monitoring of key equipment. Finally, through these hierarchical feedback and optimization, the system successfully ensured the safety and smooth progress of the entire nighttime construction process.

[0075] Preferably, the multi-level intelligent feedback network includes multiple feedback levels, each level analyzes and optimizes data of different dimensions, wherein the first level is for current monitoring results, the second level is for historical monitoring data, and the third level is for exception handling records, thereby generating comprehensive feedback optimization data for further optimizing global collaborative strategy data.

[0076] The multi-level intelligent feedback network contains multiple feedback levels, each of which performs specific analysis and optimization on data of different dimensions to provide multi-faceted information support. The first-level feedback mechanism focuses on data analysis of current monitoring results. This level mainly processes real-time collected data to quickly identify and respond to changes in the current construction site environment. Through this level, the system can instantly adjust the working parameters of the monitoring nodes, such as the focal length of the camera, the sensitivity of the sensor, etc., to ensure that the effectiveness of monitoring is improved in a short period of time. For example, when the system detects that a certain construction site area is crowded with people, the first-level feedback mechanism can immediately adjust the camera's viewing angle to cover a larger area to ensure that all personnel activities are within the monitoring range.

[0077] The second level of feedback mechanism focuses on the comparison and analysis of historical monitoring data. By comparing the current monitoring data with historical data, the system can identify the difference between the current state and the previous pattern, thereby predicting potential risks or trends. For example, if the system finds that the current equipment temperature is slightly higher than the historical level during the same period, and the temperature is gradually rising, the system can infer that the equipment may be facing the risk of overheating and issue an early warning. This level of feedback mechanism enables the system to not only respond to current changes, but also make more accurate medium-term forecasts based on historical data, providing forward-looking decision support for site management.

[0078] The third level of feedback mechanism focuses on the comprehensive analysis of exception handling records and long-term trends. This level mainly processes long-term accumulated data and records of abnormal situations, identifies periodic or sudden events that may occur during site operations, and incorporates this information into the adjustment of monitoring strategies. For example, the system may find through analyzing the monitoring data of the past few months that the performance of a particular device often fluctuates during the rainy season. Therefore, when the rainy season is about to begin, the system will automatically increase the monitoring of the device and adjust the monitoring strategy to adapt to possible abnormal situations. This level of feedback mechanism ensures that the system can effectively manage the site environment in the long term and maintain stable operation under complex conditions.

[0079] Through the collaborative work of these three levels, the multi-level intelligent feedback network can generate comprehensive feedback optimization data. These data not only include responses to the current state, but also integrate the comparison results of historical data and analysis of abnormal records, thus providing a multi-dimensional optimization solution for the system. These feedback optimization data are used to further adjust and optimize the global collaborative strategy data to ensure that the system achieves optimal resource allocation and strategy coordination among all monitoring nodes.

[0080] In actual applications, the effect of this multi-level feedback mechanism is very significant. For example, at a large construction site, the system discovered through the first-level feedback mechanism that a certain device was operating abnormally in hot weather, and immediately adjusted the monitoring parameters for more intensive monitoring. Subsequently, through the second-level feedback mechanism, this anomaly was compared with historical data to confirm that the temperature rise of the equipment was consistent with previous similar situations. Then, through the third-level feedback mechanism, the past abnormal handling records were analyzed and it was found that this situation usually leads to equipment failure. Therefore, the comprehensive feedback optimization data finally generated by the system will not only recommend immediate inspection of the equipment, but also adjust the overall monitoring strategy, increase the monitoring frequency of the equipment, and prioritize resource allocation in related areas to prevent the expansion of the failure and affect the site operation.

[0081] like Figure 5 As shown, a system for implementing the integrated multi-dimensional monitoring method for a construction site includes: The multi-spectral imaging module, the acoustic wave detection and analysis module, and the LIDAR module are used to collect environmental data, acoustic wave feature data, and three-dimensional spatial data at key locations on the construction site, and pre-process and format the environmental data, acoustic wave feature data, and three-dimensional spatial data to generate a processed comprehensive data set; multi-dimensional data collection is performed at key locations on the construction site through the multi-spectral imaging module, the acoustic wave detection and analysis module, and the LIDAR module. These modules are responsible for collecting environmental data, acoustic wave feature data, and three-dimensional spatial data, respectively, and then pre-processing and formatting these data to generate a processed comprehensive data set. This comprehensive data set provides rich and diverse input data for subsequent multi-level spatiotemporal correlation modeling and feature extraction, thus laying the foundation for comprehensive monitoring of the system.

[0082] The multi-level spatiotemporal graph convolutional network module is used to receive the processed comprehensive data set, perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data; the multi-level spatiotemporal graph convolutional network module receives and processes these comprehensive data. Through multi-level spatiotemporal correlation modeling, the system can capture the dynamic change characteristics in the construction site environment. This module extracts valuable spatiotemporal correlation data by analyzing the spatiotemporal correlation in the data, and further provides a basis for the system's state analysis and prediction.

[0083] The adaptive Bayesian-Markov model module is used to perform uncertainty analysis and state transition prediction on the multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data; in the state analysis and prediction stage, the adaptive Bayesian-Markov model module performs uncertainty analysis and state transition prediction on these spatiotemporal correlation data. This module generates comprehensive environmental prediction data by quantifying the uncertainty factors in the data and predicting the future state. These prediction data help the system predict possible changes or risks in the construction site in the future, thereby providing an important basis for adjusting the monitoring strategy.

[0084] The interactive dynamic game model module is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes based on the comprehensive environmental prediction data, generate global coordination strategy data, and perform monitoring tasks according to the global coordination strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; the interactive dynamic game model module optimizes the resource allocation and monitoring strategy of the construction site monitoring nodes based on the comprehensive environmental prediction data. Through dynamic game, the module generates the optimal global coordination strategy data, and adjusts the parameters and priority configuration of the monitoring equipment accordingly to ensure the effective use of resources and efficient execution of monitoring.

[0085] The multi-level intelligent feedback network module is used to collect and process real-time monitoring feedback data during the execution of monitoring tasks, and to perform multi-level iterative optimization on real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating global collaborative strategy data. The multi-level intelligent feedback network module collects and processes feedback data in real time during the execution of monitoring tasks. Through multi-level iterative optimization, the module compares and analyzes real-time feedback data with historical data, generates optimized data and updates the global collaborative strategy. In this way, the system can maintain efficient and accurate monitoring in a dynamically changing construction site environment, and can quickly adapt to changes in the environment, thereby ensuring the safety and operational efficiency of the construction site.

[0086] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0087] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An integrated multi-dimensional monitoring method for a construction site, characterized in that: The following steps are involved: Through multi-spectral imaging technology, acoustic wave detection and analysis technology and LIDAR technology, environmental data, acoustic wave characteristic data and three-dimensional spatial data are collected at key locations on the construction site, and the collected data are pre-processed and formatted to generate a processed comprehensive data set; The processed comprehensive data set is input into a multi-level spatiotemporal graph convolutional network to perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data. The multi-level spatiotemporal correlation data is then subjected to uncertainty analysis and state transition prediction through an adaptive Bayesian-Markov model to generate comprehensive environmental prediction data. Based on comprehensive environmental prediction data, the interactive dynamic game model is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes, generate global collaborative strategy data, and perform monitoring tasks according to the global collaborative strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; During the execution of the monitoring task, real-time monitoring feedback data is collected and processed, and multi-level iterative optimization is performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating the global collaborative strategy data.

2. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: The preprocessing and formatting include the following steps: denoising the environmental data through a filtering algorithm to eliminate environmental noise interference; normalizing the sound wave feature data to standardize the signal strength of different sound sources; converting the data format of the three-dimensional space data and converting the point cloud data into a raster data format, thereby generating a comprehensive data set suitable for a multi-level spatiotemporal graph convolutional network.

3. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: The multi-level spatiotemporal graph convolutional network includes the following steps: by constructing a multi-layer convolutional neural network, multi-level spatiotemporal correlation modeling is performed on the processed comprehensive data set, wherein each layer of the convolutional network processes data of different time scales and spatial resolutions respectively to extract dynamic change characteristics in the construction site environment, and finally generates multi-level spatiotemporal correlation data containing multi-dimensional spatiotemporal relationships.

4. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: The adaptive Bayesian-Markov model performs state transition analysis on multi-level spatiotemporal correlation data by the following steps: Based on historical observation data and current observation data, the Bayesian inference method is used to recursively update the state transition probability and uncertainty parameters; Based on the state transition model of Markov chain, the construction site environmental status at future moments is predicted, thereby generating comprehensive environmental prediction data including short-term and medium-term prediction results.

5. The integrated multi-dimensional monitoring method for construction sites according to claim 4, characterized in that: The state transition probability of the adaptive Bayesian-Markov model is calculated by the following formula: in, Indicates at time state Transfer to time state The probability of Indicates that in observation Down state The conditional probability of Indicates observation The probability of Indicates in status Observation The conditional probability of This formula is used to quantify the state transition relationship in multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data.

6. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: The interactive dynamic game model models the resource allocation and strategy optimization of the construction site monitoring nodes through the following steps: Construct a resource allocation game model, in which the resource allocation decision of each monitoring node is based on the principle of maximizing the global utility function to generate the optimal resource allocation vector; Based on the output of the resource allocation game model, a strategy optimization game model is further constructed. The strategy selection vector of each monitoring node is generated through a dynamic game process, and finally the global collaborative strategy data is generated.

7. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: The interactive dynamic game model determines the optimal resource allocation in the resource allocation game through the following formula: in, Indicates The resource allocation vector of nodes; Indicates The strategy selection vector of nodes; Indicates The utility function of each node; Indicates the total amount of available resources; This formula is used to optimize resource allocation among monitoring nodes and generate global coordination strategy data.

8. The integrated multi-dimensional monitoring method for construction sites according to claim 1, characterized in that: During the execution of the monitoring task, the following iterative optimization steps are performed on the real-time monitoring feedback data and historical data through a multi-level intelligent feedback network: In the first-level feedback mechanism, preliminary feedback optimization data is generated based on the current monitoring results; In the second-level feedback mechanism, the preliminary feedback optimization data is compared and analyzed with the historical monitoring data to generate the second feedback optimization data; In the third-layer feedback mechanism, the abnormal situation handling records and long-term trend analysis are combined to generate multi-level feedback optimization data for adjusting the parameters and strategies of the construction site monitoring nodes.

9. The integrated multi-dimensional monitoring method for construction sites according to claim 8, characterized in that: The multi-level intelligent feedback network includes multiple feedback levels, each level analyzes and optimizes data of different dimensions, where the first level is for current monitoring results, the second level is for historical monitoring data, and the third level is for exception handling records, thereby generating comprehensive feedback optimization data for further optimizing global collaborative strategy data.

10. A system for implementing the integrated multi-dimensional monitoring method for a construction site according to any one of claims 1 to 9, characterized in that: include: The multispectral imaging module, the acoustic wave detection and analysis module and the LIDAR module are used to collect environmental data, acoustic wave characteristic data and three-dimensional spatial data at key locations on the construction site, and pre-process and format the environmental data, acoustic wave characteristic data and three-dimensional spatial data to generate a processed comprehensive data set; A multi-level spatiotemporal graph convolutional network module, which is used to receive the processed comprehensive data set and perform multi-level spatiotemporal correlation modeling and feature extraction to generate multi-level spatiotemporal correlation data; An adaptive Bayesian-Markov model module, used for performing uncertainty analysis and state transition prediction on the multi-level spatiotemporal correlation data to generate comprehensive environmental prediction data; An interactive dynamic game model module is used to optimize the resource allocation and monitoring strategy of the construction site monitoring nodes based on the comprehensive environmental prediction data, generate global collaborative strategy data, and perform monitoring tasks according to the global collaborative strategy data, including real-time adjustment of monitoring equipment parameters and priority configuration; The multi-level intelligent feedback network module is used to collect and process real-time monitoring feedback data during the execution of monitoring tasks, and to perform multi-level iterative optimization on real-time monitoring feedback data and historical data through a multi-level intelligent feedback network to generate optimized data for updating global collaborative strategy data.

Images