Intelligent construction site dynamic collaborative management and control system based on digital twinning

Abstract

The application relates to the technical field of construction monitoring, in particular to a dynamic collaborative management and control system for a smart construction site based on digital twinning, which comprises an intelligent agent layout module, a region allocation module, a risk verification module, a risk monitoring module and a risk scheduling module; through receiving log information of a current construction project, a plurality of construction regions of the intelligent agent under the construction project allocation are obtained; the construction region is defined with a boundary attribute, each intelligent agent is coupled with a risk factor according to the boundary attribute of the construction interval, and a risk factor coupling model is obtained; the risk propagation path of the risk factor coupling model is verified, and a risk flow direction graph corresponding to each intelligent agent is determined; the coupling factors of each intelligent agent under the risk flow direction graph are monitored, and the risk range of each intelligent agent during construction is evaluated; according to the risk range of each intelligent agent during construction, the states of each intelligent agent under the risk connection path are compared, and the safety threshold of each intelligent agent is updated; the accuracy and reliability of intelligent agent risk early warning are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of construction monitoring, in particular to a dynamic collaborative management and control system for a smart construction site based on digital twinning. BACKGROUND

[0002] In a construction project, safety monitoring and early warning are important links to ensure the smooth progress of construction and the safety of personnel and property. However, the traditional construction project safety monitoring method often has problems such as limited monitoring range, low precision, and unclear risk propagation path, which is difficult to meet the high requirements of modern construction projects for safety monitoring and early warning.

[0003] For example, Chinese Patent Publication No. CN118411067A discloses a construction quality supervision method and system for building construction, which belongs to the technical field of quality management, and specifically includes: determining the construction process complexity of different construction projects in the construction area based on the construction process of the construction project, obtaining the number of construction personnel of different construction projects in the construction area, and determining the process construction risk of different construction projects in combination with the historical construction data of the construction personnel of the construction project. Determine the risk construction project of the construction area based on the process construction risk of the construction project, determine the area type of the construction area according to the risk construction project of the construction area, the construction process complexity and the process construction risk of different construction projects, and determine the quality supervision strategy of different construction areas through the area type of different construction areas of the target construction project.

[0004] For example, Chinese Patent Publication No. CN118396576A discloses a construction progress monitoring and early warning method and system, which includes: taking a target construction monitoring image as a starting point, and finding a first historical target construction monitoring image and a second historical target construction monitoring image in a historical target construction monitoring image sequence with a preset time difference as a span, and then determining the first feature similarity between the target construction monitoring image and the first historical target construction monitoring image, and the second feature similarity between the first historical target construction monitoring image and the second historical target construction monitoring image.

[0005] The prior art describes a management method for describing construction quality through the complexity of the construction project process, and uses the similarity of the construction object to describe the progress of the construction project, but in describing the progress and quality of the construction project, safety monitoring and early warning are still needed for the current construction project to identify the safety risks existing in the progress of the construction project, as well as the propagation path and flow direction of the risks in the construction project, to provide a basis for risk early warning and response. SUMMARY

[0006] In order to solve the above technical problems, the technical scheme adopted by the present application is: a dynamic collaborative management and control system for a smart construction site based on digital twinning, comprising: an intelligent agent deployment module for receiving log information of a current construction project, generating at least one intelligent agent, and dividing the construction project by each intelligent agent to obtain a plurality of construction areas under the allocation of the intelligent agent.

[0007] A region allocation module for defining boundary attributes of the construction area, the boundary attributes including geographic space attributes, construction stage attributes and risk factor attributes; coupling each intelligent agent with a risk factor based on the boundary attributes of the construction interval to obtain a risk factor coupling model corresponding to each intelligent agent.

[0008] A risk verification module for verifying the risk propagation path of the risk factor coupling model according to the risk factor coupling model corresponding to each intelligent agent, and determining the risk flow direction graph of each intelligent agent under the construction area based on the boundary attributes of the risk propagation path.

[0009] A risk monitoring module for monitoring the coupling factor of each intelligent agent under the risk flow direction graph according to the risk flow direction graph corresponding to each intelligent agent, and evaluating the risk range of each intelligent agent during construction.

[0010] A risk scheduling module for connecting the intelligent agents according to the position of the generated risk to obtain a risk connection path under different risk scheduling, comparing whether the states of each intelligent agent under the risk connection path are consistent, and if so, outputting the risk connection path, and if not, updating the safety threshold of each intelligent agent according to the state of each intelligent agent in the risk connection path.

[0011] The beneficial effects of the present application are: first, the present application generates intelligent agents by receiving log information of a construction project, divides the construction project, and obtains a plurality of construction areas, which can sequentially divide the current construction project into a plurality of intelligent agents to focus on the risks existing in the current construction project under different task execution, and can detect the construction project from multiple levels through multi-agent collaboration to identify problems in the multi-agent collaboration scenario and discover and identify potential safety risks in time.

[0012] Second, the present application couples the risk attributes of each intelligent agent after the node boundary attributes of the construction area to describe the mutual relationship between each risk factor, identify the risk factor coupling model composed after the risk factor coupling, and quantify the coupling relationship between each risk factor, which provides a scientific basis for risk analysis and early warning; at the same time, the risk propagation path existing in the risk factor coupling model can show the propagation between each risk factor, which is convenient for staff to analyze and decide the risk to improve the accuracy of safety warning of the construction project.

[0013] Thirdly, the application can verify the correctness and effectiveness of the risk propagation path, so as to ensure that the propagation of the risk factors from the construction area is identified when the multiple intelligent agents interact in the covered area, and the other construction areas are warned according to the propagation of the risk factors, so as to reduce the loss caused by the risk factors.

[0014] Fourthly, the application can further compress the range of the warning corresponding to the risk factors, reduce the risk false alarm of the intelligent agent when the risk factors appear, and improve the adaptability of the construction project to the abnormal threshold according to the adjustment of the safety threshold, so as to finally improve the safety and reliability of the construction project under monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0015] The application will be further described below in combination with the drawings and embodiments.

[0016] Figure 1 is a system schematic diagram of a dynamic collaborative management and control system of a smart construction site based on digital twinning.

[0017] Figure 2 is a flowchart of a regional allocation module of a dynamic collaborative management and control system of a smart construction site based on digital twinning.

[0018] Figure 3 is a flowchart of a risk verification module of a dynamic collaborative management and control system of a smart construction site based on digital twinning. DETAILED DESCRIPTION

[0019] The embodiments of the application will be described in detail below. The embodiments described below are exemplary and are only used to explain the application, and cannot be understood as a limitation of the application. If the specific technology or condition is not specified in the embodiments, the technology or condition described in the literature in the art or according to the product manual is used.

[0020] Generally, the system of multiple intelligent agent cooperation includes mixing intelligent agent, transportation intelligent agent, paving intelligent agent and compaction intelligent agent and the like. These intelligent agents represent the management mode of the material used and laid under the construction environment, and multiple intelligent agents are divided under each intelligent agent according to different needs, so as to reflect multiple information of the management mode, work equipment and work scene existing in the construction project. The information is classified and managed, and the risk existing in the mixed multiple information is identified, the intelligent agent with the existing problem can be quickly located, the risk perception or risk warning and other multiple working conditions for different environmental scenes and modes are completed.

[0021] Referring to Figure 1 The intelligent construction site dynamic collaborative management and control system based on digital twinning includes an intelligent agent layout module, a region allocation module, a risk verification module, a risk monitoring module, and a risk scheduling module.

[0022] The intelligent agent layout module is configured to receive log information of a current construction project, generate at least one intelligent agent, divide the construction project by each intelligent agent, and obtain a plurality of construction regions under the allocation of the intelligent agent.

[0023] The region allocation module is configured to define boundary attributes of the construction regions, including geographical space attributes, construction stage attributes, and risk factor attributes.

[0024] The risk verification module is configured to verify a risk propagation path of the risk factor coupling model according to the risk factor coupling model corresponding to each intelligent agent, and determine a risk flow direction graph of each intelligent agent in the construction region according to the boundary attributes of the risk propagation path.

[0025] The risk monitoring module is configured to monitor coupling factors of each intelligent agent in the risk flow direction graph according to the risk flow direction graph corresponding to each intelligent agent, and assess a risk range of each intelligent agent during construction.

[0026] The risk scheduling module is configured to connect the intelligent agents according to the positions where the risks are generated, obtain a risk connection path under different risk scheduling, compare the states of each intelligent agent in the risk connection path, output the risk connection path if the states are consistent, and update the safety threshold of each intelligent agent according to the states of each intelligent agent in the risk connection path if the states are inconsistent.

[0027] The event back-to-graph represents the connection between a plurality of events and a single risk after the combination of risk factors and construction stages covered by the plurality of intelligent agents under different attribute combinations, explains the risks that can be generated by the current construction and the content that causes the risks, and then combines the plurality of intelligent agents according to the event back-to-graph of each construction region to obtain the parts controlled by those intelligent agents in the current intelligent agents that are prone to risk events.

[0028] In the subsequent description, the agent will consider the work tasks such as material management, equipment use, communication interaction under each process as an agent according to the processes existing under the construction project, and according to the risk of multiple agents, it is explained whether the risk of each agent exists between other agents, which leads to large-scale failure problems caused by partial equipment failure, or the overall construction project is affected because the completion degree of partial tasks is low.

[0029] The agent used here can be regarded as an independent computing unit responsible for managing a certain sub-task or construction activity in a certain area of the construction project; the collected log information of the construction project includes construction progress, resource usage, environmental data, abnormal event records and other contents, and then these contents are divided into data and areas contained in a single agent according to the sub-tasks contained in the project during construction, to obtain multiple construction areas corresponding to the current agent.

[0030] When generating at least one agent, the implementation of the agent arrangement module further includes: if the positions of the agents do not correspond to the log information of the current construction project, extracting the local features corresponding to each agent according to the positions corresponding to the construction project, and determining the next construction area of each agent during the execution of the construction project according to the connection relationship between the local features of each agent, and taking the next construction area as the construction area of each agent under the project allocation.

[0031] If the positions of the agents correspond to the log information of the current construction project, the corresponding construction area is extracted according to the positions of the agents.

[0032] When the positions of the agents do not correspond to the log information of the current construction project, it generally indicates that the construction log has data missing, or after multiple agents are combined, there are some data conflicts; after the equipment enters the work, due to poor communication environment or equipment abnormalities, the recorded log has a large deviation, at this time, the next task corresponding to the construction area is needed to verify whether the equipment and area represented by the current agent exist risks, in order to realize the safety warning of the construction project.

[0033] In one embodiment of the present application, the geospatial attribute refers to the boundary conditions related to geographical location, which defines the restrictions on physical space that affect how risks propagate between different locations, and can include physical barriers, proximity relationships, and access controls. Physical barriers, such as the presence of natural or man-made obstacles such as buildings, fences, roads, etc., can prevent or limit the spread of risks, such as fires, pollution, etc., from one area to another; proximity relationships take into account the relative position relationship between different construction areas, and how this relationship affects the speed and direction of risk propagation; access controls are that some areas may have strict access permissions, only certain personnel or equipment can enter, which will affect the possibility of risk spreading through personnel or equipment flow.

[0034] The construction phase attribute relates to different stages of the project life cycle and their specific risk management needs. Each stage has its specific tasks and potential risks, so it is necessary to develop targeted risk management strategies. It usually includes the planning phase, preparation phase, implementation phase and completion phase, each of which has its unique set of tasks and risk points. Alternatively, it can be divided into time windows, so that each stage has a clear time frame during which specific work must be completed, failure to complete on time may result in schedule delays or other increased risks. At the same time, it also indicates the resource requirements under each construction phase, for example, in the infrastructure stage, material supply and heavy machinery use are critical; while in the decoration stage, more attention is paid to detail handling and quality control. Then according to the different construction phases, the risks that may exist under the demand of each construction phase can be known, which is convenient for subsequent safety warning of construction projects.

[0035] The risk factor attribute focuses on various internal and external factors that cause risks to occur, which determine the essential characteristics of risks and the possible consequences they may cause. At this time, all possible risk types that the agent may contain are identified, such as technical risk, economic risk, environmental risk, social risk, etc.; these risks are treated as their risk factor attributes, and whether there is a corresponding relationship between the risk factors is evaluated to complete the coupling of their risk factors and the interaction between different risk factors.

[0036] By comprehensively considering the boundary attributes of these three aspects, the risk propagation mechanism in construction projects can be more comprehensively understood, and more effective preventive measures and emergency plans can be developed accordingly. This helps to improve the overall safety of the project and reduce unnecessary losses.

[0037] Defining the boundary attribute of the construction area also includes: based on the size of the range contained by each agent, screening each construction area to determine whether it is an adjacent area; if it is an adjacent area, set the boundary attribute of each agent based on the overlapping area of each agent in the adjacent area.

[0038] If the overlapping area is too large, it indicates that the agents need to work together more, and a higher priority is needed for multi-aspect collaboration processing, that is, the spatial position is added to the common boundary length and overlapping area between adjacent construction areas to indicate whether there is an association between each construction area.

[0039] If not adjacent, the distance between the closest agents in each construction area is used as the boundary attribute.

[0040] If there is no adjacent situation, the relative position of the boundary is described by the distance between the areas where the agents are located. This position will represent whether there will be a corresponding impact in the areas covered by these agents, and whether there will be a coupling phenomenon when risks occur, so as to identify the main risks in the current construction project, facilitating subsequent early warning for these risks.

[0041] As shown in Figure 2 The implementation of the risk factor coupling model of the area allocation module includes: based on the boundary attributes corresponding to each agent, at least one risk label is obtained, and the risk coupling dimension corresponding to each agent is identified according to the risk label. At this time, the risk label set represents the risks existing in the data association or task collaboration of multiple agents, such as equipment failure, operation error, and man-machine interface mismatch caused by equipment personnel interaction risk; when implementing the construction project in the corresponding scene, the mismatch or data comparison difference existing in the material calling, product quality control, and communication coordination of multiple projects causes the actual scheme to be inconsistent with the implementation scene; these problems will represent the risk factors in the current multi-agent joint construction control, which will affect the quality and efficiency of the construction. These combined contents will be considered as risk coupling dimensions, such as personnel equipment, management execution, and other dimensions of risk factors.

[0042] The instructions sent and received by each agent are summarized using the risk coupling dimension, the risk association probability between each agent is determined by performing risk association calculation on the time window corresponding to the log information according to the log information reported by each agent, and the risk association probability is corrected based on the dependency condition of the risk factors of each agent in the historical data. The risk factor coupling matrix is established based on the corrected risk association probability.

[0043] The dependency condition of the risk factor represents the conditional probability of the risk of each agent, and the data corresponding to the conditional probability is used as the dependency condition of the risk factor.

[0044] For example, the dust concentration in the construction area associated with a certain agent is currently detected to exceed the standard, and at the same time, the dust emission amount of a certain mixing device increases, at this time, the two obviously associated parts can be analyzed to calculate their updated association probability.

[0045] That is, a risk sequence is generated in a time window corresponding to the log information Where n represents the number of risk factors present in the log information; after normalizing each element in the risk sequence, the difference between different risk factors is calculated to represent the risk association probability.

[0046] Wherein, represents the risk association probability of risk factor i and risk factor j, represents the absolute difference between risk factor i and risk factor j at time k, m represents the time length value of the time window corresponding to the log information, and k takes a value in the range of 1 to m; represents the minimum value of the absolute difference between risk factor i and risk factor j at time k; represents the maximum value of the absolute difference between risk factor i and risk factor j at time k, which represents the maximum value and minimum value in the time window where the log information is located, for standardization processing; represents the resolution coefficient, which is usually 0.5, used to adjust the contrast to avoid unstable calculation due to too small denominator. At this time, it is explained that the risk association probability between multiple risk factors existing in the time window where the log information is located is considered, and then a relative risk factor coupling matrix is composed according to the risk association probability corresponding to the risk factor, to explain whether there is a corresponding relationship between different risk factors.

[0047] After obtaining the risk association probability value, conditional probability is also used to identify whether the association between the risk factors is normal, such as Wherein, represents the conditional probability of risk factor i and risk factor j, represents the probability of the occurrence of risk factor i and risk factor j together, represents the probability of risk factor j.

[0048] Then the conditional probability is used to correct the risk association probability to obtain the risk factor coupling matrix Wherein, represents the corrected risk association probability of risk factor i and risk factor j, which will be filled into the risk factor coupling matrix according to the number of risk factors, and the size of the matrix is n x n; Wherein, , respectively, and the values of the weight coefficients are 0.6 and 0.4 respectively; the weight of the risk factor association in the actually collected log information and the weight of the risk factor in the historical data in the common occurrence are used to explain the mutual association between the current risk factors.

[0049] The key risk nodes are extracted from the risk coupling matrix, and the risk coupling network graph is composed of the risk propagation paths corresponding to the key risk nodes, and is output as the risk factor coupling model.

[0050] The extraction method of the key risk nodes can be that, for the risk factor corresponding to the key risk node, the part with a risk association probability sum greater than a preset threshold is selected as a plurality of key risk nodes, and the threshold is set according to the average of the risk association probability sum of a single risk factor and other risk factors in the historical data; or a point with the maximum risk association probability sum is selected as a key risk node.

[0051] When the risk coupling network graph is composed, the implementation further includes: connecting each element in the risk coupling matrix according to the derived events of the risk factors to obtain a plurality of risk paths, and calculating the path risk probability of each risk path; the path risk probability represents the product of the risk association probabilities on the risk path.

[0052] Starting from the key risk nodes, the associated risk nodes and the risk paths are connected to obtain the shortest paths between the plurality of key risk nodes, and the path coverage range of the shortest path and the path risk probability of the corresponding risk path are used to select the risk propagation path. At this time, the path coverage range represents the number of risk factors involved between the plurality of key risk nodes, and the ratio of the number of risk factors involved to the total number of risk factors is used to represent the construction range that can be covered when the risk occurs, and whether the construction project will be affected.

[0053] When the paths are connected, the reciprocal of the risk association probability is regarded as the weight of the edge, and the sum of the weights of the connected nodes of each key risk node is required to be the minimum, so as to obtain the shortest path between the plurality of associated risk nodes. The path represents whether there is a relative relationship between different key risk nodes, and whether there is corresponding risk transmission between the plurality of risk paths.

[0054] Preferably, when the risk propagation path is selected, the corresponding confidence and support degree in the historical data are obtained according to the path risk probability of the risk path and the path coverage range of the shortest path, and when the support degree and the confidence are both greater than 0.6, the shortest path between the corresponding risk path and the plurality of key risk nodes is output as the risk propagation path.

[0055] Preferably, selecting the risk propagation path further comprises: identifying risk factors intersecting the shortest path between the risk path and the plurality of key risk nodes, dividing the shortest path between the risk path and the plurality of key risk nodes into a plurality of target paths according to the positions of the intersecting risk factors, and obtaining the plurality of segmented target paths by splitting the path through the intersection between the paths.

[0056] The boundary attribute of each target path is extracted, the relative distance of the agent corresponding to each target path is obtained, the spatial clustering of each target path is performed, and the spatial clustering relationship between each agent and the risk factor is obtained. At this time, the spatial clustering is performed in the following manner: the position of each agent in the construction area is clustered with the spatial position mapped by the target path with the risk, the distance between each agent and the corresponding risk factor in the target path is calculated using the Euclidean distance, the clustering is performed according to the distance value, and the target path existing between the regions with risks in the construction area is identified according to the clustering. The target path is described as a hidden path in the form of a risk factor propagation in the construction area, and the target path is described as a hidden path in the form of a risk factor propagation in the construction area.

[0057] The spatial clustering relationship between each agent and the risk factor is used to map the target path to the construction area, and the connected path is output as the risk propagation path. At this time, the propagation of each risk factor in the construction area is verified, and whether the path in the propagation is consistent with the rule of the risk factor is verified, so as to complete the verification of the path.

[0058] In an embodiment of the present application, when verifying the risk propagation path of the risk factor coupling model, the verification manner is to determine whether the flow direction of the risk event is consistent with the path represented by the risk propagation path, so as to eliminate some risk propagation paths with small correlation.

[0059] As shown in Figure 3 , the implementation manner of the risk verification module comprises: performing risk matching on the risk event corresponding to the risk factor attribute and the risk propagation path, and checking whether the occurrence of each risk event is consistent with the risk propagation path.

[0060] In the risk matching, the connection mode of each agent corresponding to the risk propagation path is verified, and the verification state includes the time of the risk event occurring in each process link during construction, the abnormal data, and whether each construction area is abnormal or risk events occur at the same time. These contained data are used as the main data of risk matching. If there is relevant data and it is in the same position as the direction of the risk propagation path, it is considered that the risk matching is consistent. If the checked data is different from the area where the risk propagation path is located, it is considered to be inconsistent. In this way, it is found out which risk events will generally occur when the current construction area appears risks, causing multiple construction areas or multiple agent controlled areas to appear abnormal. Finally, the construction project is warned according to these conditions.

[0061] Preferably, the boundary attribute of the risk propagation path can also be added to process the risk propagation path, that is, the implementation mode of the risk matching also includes: based on the connection mode of the agent, the construction area corresponding to each agent is compared with the risk propagation path. At this time, the construction area corresponding to each agent and the risk propagation path are mainly compared to see whether the spatial position of the construction area corresponding to the risk propagation path is consistent with the spatial position of the construction area corresponding to each agent, and whether these spatial positions are reachable under the connection mode of the agent, whether they are in adjacent or same construction stages, and whether the risk factors exist correlation; if reachable, record as a space constraint condition, if in adjacent or same construction stages, record as a construction stage constraint, and if the risk factors exist correlation, it is a risk constraint condition. The space constraint condition, construction stage constraint and risk constraint condition at this time are used as the constraint condition of each risk propagation path; determine the constraint condition of each risk propagation path.

[0062] According to the time, geographical position and risk factor of the risk event, the risk event is mapped to the risk propagation path, and whether the risk event and the constraint condition of the risk propagation path meet the preset constraint condition is compared.

[0063] If it is consistent, the boundary attribute of the construction area connected by the risk propagation path is used to generate the risk flow direction graph corresponding to each agent. If it is not consistent, the construction area where each risk event occurs and the construction area corresponding to the risk propagation path are combined, and the combined construction area is used as the risk flow direction graph corresponding to each agent.

[0064] The preset constraint conditions include: checking whether the construction area where the risk event occurs is within the spatial constraint condition of the risk propagation path; checking whether the time when the risk event occurs is within the construction stage constraint of the risk propagation path; checking whether the risk factor of the risk event is within the risk constraint condition of the risk propagation path, if all the constraint conditions of the risk propagation path are met, the risk propagation path is marked as conforming, and if the risk event does not match any constraint condition of the risk propagation path, the risk propagation path is marked as not conforming.

[0065] The generated risk flow diagram shows how the risk is transmitted from one agent representing a subtask or a construction area to another agent, which indicates which agent will be the source of a specific risk, facilitating subsequent positioning of the initial location and cause of the risk. After describing the path of the flow, the whole process of the risk from occurrence to diffusion can be displayed, facilitating individual warning of the agent and taking targeted preventive measures. Because the boundary attributes of each agent are different, the key risk points and other contents can be identified after the path and flow are displayed.

[0066] When generating the risk flow diagram, the description of the geographical space attribute, the construction stage attribute and the risk factor attribute on the current agent corresponding construction area is pointed, and a plurality of risk factor attribute pointing single risk factor or multiple risk factor pointing diagram is generated to represent the risk tendency in the construction area corresponding to the current agent.

[0067] Preferably, the implementation manner of generating the risk flow diagram corresponding to each agent comprises: connecting the construction areas meeting the preset constraint conditions according to the constraint conditions met by each construction area, setting the direction vector of each construction area after connection with the construction area where the risk event occurs as the starting point, and taking the maximum direction vector existing after fitting of each direction as the risk flow diagram corresponding to each agent. The direction vector of the connected construction area is set according to the position of each construction area when connected, and then the direction vector is fitted to know the area from the starting point of the risk to the end point of the influence. The construction areas contained in these directions need to be warned to prevent the harm caused by the risk.

[0068] If there is a construction area without preset constraints, the construction area corresponding to the risk event is taken as a center point, and each construction area is connected according to any one of the preset constraints to obtain a risk flow direction graph corresponding to each agent. If it does not meet the preset constraints, the construction area has one or more constraints that do not meet the preset constraints under the space, risk association and construction stage. If there is a region that does not meet the preset constraints at all, the construction area is completely irrelevant to the current risk event, and only the part of the data that meets at least one preset constraint needs to be identified to find out how each construction area points to the construction area with the current risk event when the preset constraints are not fully met, thereby assisting subsequent staff in judging whether the current risk will affect the project, and then warning the related areas associated to reduce the harm caused by the risk.

[0069] In an embodiment of the present application, when monitoring the risk flow direction graph, the risk range existing in the construction area under different risk factors is identified according to the direction and position corresponding to the risk flow direction graph, so as to describe the risk problem of the construction project under safety monitoring.

[0070] The implementation manner of the risk monitoring module includes: taking the construction area contained in the risk flow direction graph in each time period as the influence range of the risk flow direction graph, and taking the duration of the risk flow direction graph pointing to the risk factor as the duration of the risk flow direction graph; extracting the risk association probability corresponding to each risk factor in the risk flow direction graph from the risk factor coupling model as the risk association probability of the risk flow direction graph; and taking the influence range, duration and risk association probability of the risk flow direction graph as the coupling factor corresponding to the risk flow direction graph.

[0071] The construction area contained in the risk flow direction graph in each time period indicates the size of the risk propagation area when the risk propagation occurs; the duration of the risk flow direction graph pointing to the risk factor indicates the time required from one risk factor to another risk factor in the risk flow direction graph, at this time, it indicates the time when the two risk factors have a mutual influence relationship when the pointing risk factor is the risk factor corresponding to the starting point to the ending point in the risk flow direction graph, representing the association degree between different risks; and the extracted risk association probability is the value corresponding to the multiple risk factors represented by the risk factor coupling model, thereby completing the coupling analysis of the risk factors between the current multiple agents.

[0072] The coupling factor presents the diffusion degree of the risk in the space, helps to identify the high-risk area for resource allocation, quantifies the duration of the risk, can reflect the potential cumulative impact when the risk occurs, identifies the risk problem of the construction project under long-time accumulation, and finally collects the risk association probability to clearly understand the interaction between different risk factors, thereby reducing the problem of reduced risk warning accuracy caused by single index evaluation.

[0073] The product of the risk correlation probability, the influence range and the duration corresponding to the coupling factor after normalization processing is preferably used to represent the risk situation of each agent after coupling, that is, the risk factors are correlated through multiple nodes and multiple paths, and the main influence of a specific node under the overall risk propagation in the correlation is used to quantify the risk range that can be caused by the anomaly during construction after each agent identifies the anomaly. According to the time length corresponding to the risk propagation path, it is determined whether each risk factor will cause the corresponding risk problem within a certain time and timely warning is given.

[0074] The risk flow diagram is sequentially arranged according to the coupling factor. When arranging, the influence range, the duration and the risk correlation probability of the construction area covered by the risk flow diagram are used to describe each construction area, and the descending order is arranged according to the values of the influence range, the duration and the risk correlation probability from large to small, as shown in Table 1.

[0075] Table 1. Sequential arrangement of coupling factors

[0076]

[0077] Each coupling factor in Table 1 represents a construction area. After the comprehensive value of the influence range, the duration and the risk correlation probability of the construction area at this time is calculated and sorted, the similarity of adjacent coupling factors is calculated, the target coupling factor is selected, and the score finally represents the product of the risk correlation probability, the influence range and the duration after normalization processing, which indicates the influence of each coupling factor on the corresponding construction area. This is convenient for finding the target coupling factor according to the score in the subsequent process, so as to improve the accuracy of the identification of multiple risk factors.

[0078] The similarity of adjacent coupling factors after sequential arrangement and the average similarity of all coupling factors are used to determine the target coupling factor, and the target coupling factor is output as the risk range of each agent during construction.

[0079] The similarity of the coupling factors is calculated by the risk correlation probability, the influence range and the duration, and the similarity between the coupling factors is obtained in the form of cosine similarity. When determining the target coupling factor, the coupling factor that is obviously higher than the global average can be obtained by taking the similarity of adjacent coupling factors greater than the average similarity plus the standard deviation. If the similarity of adjacent regions is significantly higher than the global average, the coupling factors of the two regions form a strong correlation pair. The coupling factor with a higher product score in the strong correlation pair is selected as the target candidate. The finally selected target coupling factor is the coupling factor with the maximum product score among the target candidates in the data of the risk flow diagram being processed, which indicates the main risk at this time, and the risk is scheduled according to the risk.

[0080] In an embodiment of the present application, the risk scheduling module mainly connects according to the risk corresponding to each agent to determine whether the states of each agent under the risk distribution are consistent, to describe whether each agent can intelligently feedback to other faulty agents.

[0081] Preferably, the state of the agent includes a safety threshold, a fault flag, and an abnormality level, and at this time, the obtained data includes a normal range of device operating parameters, whether it is in a fault state, and an abnormality level marked when an abnormality occurs.

[0082] The finally output risk connection path outputs the corresponding agents in the same state, and if they are inconsistent, the safety threshold and other parts of each agent are checked to find the agents with state deviation at this time to analyze the abnormality type such as value overrun, logic conflict, and protocol mismatch; to realize the intelligent feedback of the states of multiple agents, and finally complete the early warning of each agent in the corresponding area.

[0083] At the same time, the safety threshold of the agent with state deviation is adjusted based on the abnormality degree, historical data, and path characteristics of the risk connection path to monitor the updated state, to reduce the problem of causing multiple agent failures around a single agent failure in a self-adaptive threshold manner, and to timely identify the distribution of each abnormal agent under the current construction project based on the adjusted safety threshold and the output consistent risk connection path, to improve the efficiency of the intelligent agent early warning.

[0084] Therefore, the implementation manner of the risk scheduling module further includes verifying the state of each agent on the risk connection path, determining the state difference corresponding to each agent; detecting the feedback strategy of each agent according to the state difference of each agent, and updating the safety threshold of each agent according to the update time of the feedback strategy.

[0085] At this time, it is detected whether the coupling factor and other data recorded on each agent are different from the average value represented by the historical data, and the difference value is regarded as the state difference at this time, and then the feedback strategy of each agent under the state difference is compared, the feedback strategy represents the ratio of the number of false positives and false negatives to the total number of early warnings when the agent outputs, uploads, and feeds back in the corresponding construction area after the agent is allocated with tasks and resources, and finally the safety threshold is adjusted according to the value obtained by the feedback strategy.

[0086] For example, based on the abnormality degree, if the deviation of an agent is small, the safety threshold of the agent can be adjusted slightly; based on the path characteristics, if the overall risk of the risk connection path is high, the safety threshold of all nodes can be uniformly tightened; based on the historical data, the average value of the historical data on the corresponding agent is referred to to dynamically adjust the threshold to adapt to system changes.

[0087] For example, the difference value corresponding to the state difference is taken as the main body, and the feedback strategy corresponding score is taken as the auxiliary, and the feedback strategy score is represented as the sum of the ratio of the number of false alarms and the number of missed alarms to the total number of early warnings; if the difference value of the state difference is greater than the average value of the state difference in the historical data, it is considered that the deviation is large, otherwise it is considered that the deviation is small, and the adjustment value based on the abnormal degree is set by the deviation coefficient, that is, the difference value of the state difference is multiplied by the deviation coefficient, and the deviation coefficient is selected as 0.25 and 0.05 according to the deviation large and deviation small conditions; then the adjustment value based on the path characteristic is set, at this time the average value of the coupling factor on the current identified risk connection path is obtained, and then the average value of the coupling factor is based on the average value of the coupling factor in the historical data. The difference value corresponding to the coupling factor is obtained, the average value of the coupling factor is calculated according to the coupling factor corresponding score, and then the adjustment value based on the path characteristic is obtained by subtracting the weighted value of the difference value corresponding to the coupling factor from the adjustment value of the last path characteristic. The average value in the historical data is used as the adjustment value based on the historical data at this time, the adjustment value based on the historical data, the adjustment value based on the path characteristic and the adjustment value based on the abnormal degree are weighted and summed to obtain the final security threshold value used, which can indicate the value condition of the coupling factor on the construction area contained by the intelligent agent under what circumstances. If the value is too large after the risk coupling, it means that there is a serious risk, and the surrounding area needs to be authenticated.

[0088] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application, which are still covered by the protection scope of the present application.

Claims

1. The dynamic collaborative management and control system of smart construction sites based on digital twins is characterized by: include: An agent placement module is configured to receive log information of a current construction project, generate at least one agent, divide the construction project into different agents, and obtain multiple construction areas assigned to the agents under the construction project. The area allocation module is used to define the boundary attributes of the construction area. The boundary attributes include geographic space attributes, construction stage attributes, and risk factor attributes. The risk factor coupling model corresponding to each agent is obtained by coupling the risk factors of each agent based on the boundary attributes of the construction area. The risk verification module is used to verify the risk propagation path of the risk factor coupling model based on the risk factor coupling model corresponding to each intelligent agent, and determine the risk flow diagram corresponding to each intelligent agent in the construction area based on the boundary attributes of the risk propagation path; The implementation method of the risk verification module includes: matching the risk events corresponding to the risk factor attributes with the risk propagation path, and comparing the construction area corresponding to each intelligent agent with the risk propagation path based on the connection method of the intelligent agents; mapping the risk events to the risk propagation path according to the time, geographical location and risk factors of the risk events, and comparing whether the constraints of the risk events and the risk propagation path meet the preset constraints; if they meet, generating the risk flow diagram corresponding to each intelligent agent based on the boundary attributes of the construction area connected by the risk propagation path; if they do not meet, combining the construction area where each risk event occurs with the construction area corresponding to the risk propagation path, and using the combined construction area as the risk flow diagram corresponding to each intelligent agent; The preset constraints include: checking whether the construction area where the risk event occurs is within the spatial constraints of the risk transmission path; checking whether the time of the risk event is within the construction phase constraints of the risk transmission path; checking whether the risk factors of the risk event are within the risk constraints of the risk transmission path. If all the constraints of the risk transmission path are met, the risk transmission path is marked as compliant; if the risk event does not match any of the constraints of the risk transmission path, the risk transmission path is marked as non-compliant. The risk monitoring module is used to monitor the coupling factors of each agent under the risk flow diagram corresponding to each agent and assess the risk range of each agent during construction; The implementation method of the risk monitoring module includes: taking the construction area included in the risk flow diagram in each time period as the influence range of the risk flow diagram, and taking the duration of the risk flow diagram pointing to the risk factor as the duration of the risk flow diagram; extracting the risk association probability corresponding to each risk factor in the risk flow diagram from the risk factor coupling model as the risk association probability of the risk flow diagram; taking the influence range, duration and risk association probability of the risk flow diagram as the coupling factors corresponding to the risk flow diagram, and each coupling factor represents a construction area; arranging the risk flow diagram in sequence according to the coupling factors, and using the similarity of adjacent coupling factors after the sequence arrangement and the average similarity of all coupling factors to determine the target coupling factor, and outputting the target coupling factor as the risk range of each intelligent agent during construction; The risk scheduling module is used to connect the intelligent agents according to the risk range of each intelligent agent during construction, according to the location where the risk occurs, to obtain the risk connection path under different risk scheduling, and to compare whether the status of each intelligent agent under the risk connection path is consistent. If they are consistent, the risk connection path is output; if they are inconsistent, the safety threshold of each intelligent agent is updated according to the status of each intelligent agent in the risk connection path.

2. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 1 is characterized in that: The implementation of the agent deployment module also includes: If the location of each agent does not correspond to the log information of the current construction project, the local features corresponding to each agent are extracted according to the location corresponding to the construction project, and the next construction area of ​​each agent when executing the construction project is determined according to the connection relationship between the local features of each agent. The next construction area is used as the construction area of ​​each agent under the project allocation; If the location of each intelligent agent corresponds to the log information of the current construction project, the corresponding construction area is extracted according to the location of each intelligent agent.

3. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 1 is characterized in that: The boundary attributes for the construction area also include: Based on the size of each agent's coverage, each construction area is screened to determine whether each construction area is adjacent to another area. If so, the boundary attributes of each agent are set based on the overlapping area of ​​each agent in the adjacent area. If they are not adjacent, the distance between the closest agents in each construction area is used as the boundary attribute.

4. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 1 is characterized in that: The implementation of the risk factor coupling model of the regional allocation module includes: Based on the boundary attributes corresponding to each intelligent agent, at least one risk label is obtained, and the risk coupling dimension corresponding to each intelligent agent is identified according to the risk label; The risk coupling dimension is used to summarize the instructions sent and received by each agent. According to the log information reported by each agent, the risk association calculation is performed in the time window corresponding to the log information to determine the risk association probability of each risk factor. The risk association probability is corrected based on the dependency conditions of the risk factors of each agent in the historical data, and the risk factor coupling matrix is ​​established based on the corrected risk association probability. The key risk nodes are extracted from the risk coupling matrix, and the risk coupling network diagram is composed of the risk propagation paths corresponding to the key risk nodes.

5. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 4 is characterized in that: The implementation methods of forming the risk coupling network diagram also include: Connect the elements in the risk coupling matrix according to the derivative events of the risk factors to obtain multiple risk paths, and calculate the path risk probability of each risk path; Starting from the key risk nodes, the associated risk nodes are connected with the risk paths to obtain the shortest path between multiple key risk nodes. The risk propagation path is selected based on the path coverage of the shortest path and the path risk probability of the corresponding risk path.

6. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 1 is characterized in that: The implementation methods for generating the risk flow diagram corresponding to each agent include: The construction areas that meet the preset constraints are connected according to the constraints that each construction area meets. The construction area where the risk event occurs is used as the starting point. The direction vectors of each construction area after connection are set, and the maximum direction vector after fitting is used as the risk flow diagram corresponding to each intelligent agent. If there is a construction area that does not meet the preset constraints, the construction area corresponding to the risk event is taken as the center point, and each construction area is connected according to any preset constraint that is met to obtain the risk flow diagram corresponding to each intelligent agent.

7. The digital twin-based smart construction site dynamic collaborative management and control system according to claim 1 is characterized in that: The implementation of the risk scheduling module also includes: Verify the status of each agent on the risk connection path and determine the corresponding state difference of each agent; use the state difference of each agent to detect the feedback strategy of each agent, and update the safety threshold of each agent according to the update time of the feedback strategy.

Images