Construction site comprehensive management system based on machine vision

Abstract

The invention relates to the technical field of intelligent building management, in particular to a construction site comprehensive management system based on machine vision, which comprises an image capturing and partitioning module, a construction state analysis module, a component progress matching module, a progress deviation analysis and early warning module, a task dynamic allocation module and a construction node optimization module. According to the invention, through regionally extracting and numbering image data and combining a position coordinate subdivision technology, accurate partitioning and ordered data organization of a construction site are realized, the structural property and traceability of image data are guaranteed, real-time image analysis is carried out to compare the image data with a construction plan, the state of a component is identified, and progress deviation is revealed; a dynamic early warning mechanism is established, abnormal areas are marked, risks are prompted in advance, resource allocation is concentrated in key areas based on real-time progress and task priorities, task distribution and resource use priorities are optimized, reasonable and efficient allocation of resources is achieved, and therefore intelligence, reliability and safety of construction management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent building management, and in particular to a machine vision-based integrated construction site management system. Background Art

[0002] The field of intelligent building management technology mainly focuses on the comprehensive monitoring and management of the construction and operation stages of a building. By using advanced sensors, data analysis, machine vision, the Internet of Things (IoT) and other technologies, it achieves all-round dynamic management of building facilities. The core of this technology field is to improve the safety, efficiency and environmental protection of construction sites and building operations, including real-time monitoring of personnel and equipment, material tracking, risk prediction, energy consumption control and other functions. The intelligent building management system can effectively reduce human intervention and make the construction and management process more intelligent and transparent through automated data collection and processing, thereby improving the overall management efficiency of construction projects and ensuring construction safety.

[0003] Among them, the comprehensive construction site management system is a management tool designed specifically for construction sites. It combines machine vision, sensor monitoring, data analysis and other technologies to achieve comprehensive management and safety supervision of the construction site. Its main uses include attendance management of construction site personnel, equipment status monitoring, material usage tracking, construction progress supervision, etc. Managers can understand the dynamic information of the construction site in real time, and promptly warn of potential risks, effectively prevent construction safety hazards, and ensure the orderly progress of the project.

[0004] The existing technology mainly relies on the comprehensive monitoring system for management during the construction and operation stages of the building. Its monitoring and management methods have problems such as low real-time data, delayed deviation warning, and lack of flexibility in resource allocation. Since the existing technology only realizes basic monitoring in the organization and processing of image data, lacks targeted data segmentation and structured management, the zoning management in the construction process is not accurate enough, affecting the accuracy of management. The comparison between the construction progress and the actual construction status in the existing system mostly relies on manual verification, which is difficult to achieve efficient progress deviation identification and recording, delaying the identification of progress deviation and increasing construction risks. The existing technology also lacks a dynamic allocation mechanism in resource allocation, and cannot reasonably allocate resources according to the real-time task priority, which easily causes slow progress of key tasks and inefficient resource utilization, ultimately affecting the reasonable advancement of the overall construction progress. The existing technology has a low degree of automation in construction node optimization and lacks dynamic optimization of resource allocation, which affects the efficient execution of construction tasks. As a result, when facing complex construction tasks, the existing system is difficult to achieve rapid response and reasonable configuration, which is not conducive to improving management efficiency and ensuring construction safety. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a comprehensive construction site management system based on machine vision.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: A machine vision-based construction site integrated management system comprises: The image capture and partitioning module is based on the on-site image data stream, according to the differentiated area numbers, subdividing the area boundaries, and arranging the partition images in sequence to form partition structure image data; The construction status analysis module extracts the real-time image through the area number in the partition structure image data, verifies the construction plan status, identifies and classifies the component status, and generates regional construction status information; The component progress matching module extracts the component status through the regional construction status information, checks the progress difference with the BIM model plan, and forms component progress difference data; The progress deviation analysis and early warning module verifies the component progress difference data with the preset threshold, extracts the deviation component information, records the deviation status of the abnormal component, marks the deviation area and component position, and generates a progress deviation label; The task dynamic allocation module combines the components and partition status of the progress deviation label, extracts the key task area, integrates the partition resources, dynamically allocates resources to the deviation area, rearranges the task priority and execution order, and generates a construction task priority allocation plan; The construction node optimization module screens the current resource configuration data based on the construction task priority allocation scheme, updates the task distribution and resource usage priority, and generates a construction node resource configuration scheme.

[0007] As a further solution of the present invention, the partition structure image data includes a numbering system, boundary definition, and image arrangement; the regional construction status information includes status identification, classification mark, and status record; the component progress difference data includes status matching results, progress verification records, and difference data; the progress deviation label includes deviation component identification, position marking, and deviation record; the construction task priority allocation plan includes task area positioning, dynamic resource configuration, and task sequence adjustment; the construction node resource configuration plan includes resource screening records, task priority updates, and resource configuration optimization.

[0008] As a further solution of the present invention, the image capturing and partitioning module includes: The data stream extraction submodule extracts the content of each frame of the image according to the preset partitioning rules based on the live image data stream, marks the partitioned image frames according to the partition sequence number, stores the image data in sequence and records the partition information of each frame, and classifies it into the corresponding image structure to obtain the image partition sequence number; The difference numbering submodule performs a difference analysis on the image content of each partition in the image frame based on the image partition sequence number, extracts the feature value of each partition content, arranges them in the partition order, marks the feature sequence of each image content, and obtains the difference partition numbering data; The boundary subdivision submodule calls the coordinate value of each numbered partition based on the difference partition number data, refines the boundary of the partition, matches the coordinate data of the area with the numbered partition, splices the image frame and the coordinate sequence, combines the image according to the real-time partition position, and obtains the partition structure image data.

[0009] As a further solution of the present invention, the construction status analysis module includes: The image extraction submodule reads the image data corresponding to each numbered area based on the area number in the partition structure image data, extracts each frame of image in the order of number, decomposes the independent image data of the component area, and arranges them by component number to generate a component image sequence; The status check submodule matches the status identifiers in the construction plan model item by item based on the component image sequence, extracts the status information of the component image, and checks the real-time status of the component with the planned status item by item, classifies the difference content into the corresponding component number, marks the corresponding difference information, and obtains the component status difference data; The status classification submodule applies the random forest algorithm based on the component status difference data to classify and identify the status content of the components in the difference data, integrates the status data into category indexes according to the status categories, extracts the classified status information of the components in the partition, organizes and summarizes the status information of the partition, and obtains the regional construction status information.

[0010] As a further solution of the present invention, the formula of the random forest algorithm is as follows: in, Represents the degree of deviation of the component state. Representative The status value of a component, Representative The average value of the component status, Representative The weight of a component in the classification, and n represents the total number of components.

[0011] As a further solution of the present invention, the component progress matching module includes: a status information verification submodule extracts the current construction stage and attributes of the component based on the regional construction status information and the partition component status, collects physical attribute data and identifies the process information, and records the position information and process number of the component, calibrates the matching relationship between the construction stage and the position data, and obtains real-time component status data; a progress matching comparison submodule extracts the predetermined state and time node of the component in the BIM model planning schedule based on the real-time component status data, judges the advance and lag of each component by comparing the state information with the time node, cross-checks the state and time parameters, and generates a progress difference comparison result; a difference extraction and analysis submodule applies a multivariate linear regression algorithm based on the progress difference comparison result, analyzes the state difference of each component, extracts the time offset, position offset and process deviation parameters of the component, summarizes the parameters, and obtains the component progress difference data.

[0012] As a further solution of the present invention, the formula of the multivariate linear regression algorithm is as follows: in, represents the predicted progress variance, represents the time offset, represents the position offset, represents process deviation, is the intercept term, is the weight coefficient of the variable, represents the error term.

[0013] As a further solution of the present invention, the progress deviation analysis and early warning module includes: The deviation identification submodule extracts the real-time progress and the planned progress values ​​of the components based on the component progress difference data, calculates the difference value, compares the difference value with the preset threshold parameter, marks the components exceeding the threshold range, records the deviation value and component identification information, and obtains the component deviation list; The abnormal record submodule reads the deviation value of each component based on the component deviation list, screens components in deviation state, extracts the position parameters of the components, combines the deviation state and the deviation type with the position parameters, sorts the deviation type and value according to the component number, and obtains the abnormal component deviation state record; The abnormal label mapping submodule reads the deviation area and position parameters of the abnormal component based on the abnormal component deviation status record, locates the component position, assigns the label type according to the deviation status and area information, marks the label at the component position, and generates a progress deviation label.

[0014] As a further solution of the present invention, the task dynamic allocation module includes: The key area identification submodule analyzes the current status of each component and partition based on the progress deviation label, focuses on the key areas as key task areas through deviation value sorting and urgency rating, aggregates and compiles the data of the areas into a list, and obtains a list of key task areas; The resource integration submodule reviews the resource allocation of the area, including manpower, machinery and materials, based on the list of key task areas, and generates a resource allocation plan through matching analysis between demand and current resources; The task priority adjustment submodule evaluates and reorders the tasks based on the resource allocation plan, adjusts the current task list through task urgency analysis and resource matching status check, optimizes the execution order of tasks, and builds a construction task priority allocation plan.

[0015] As a further solution of the present invention, the construction node optimization module includes: The resource data screening submodule reviews the current resource configuration data based on the construction task priority allocation plan, evaluates the performance of each resource point, determines the resource demand intensity of the task through quantitative analysis, screens the resource points and marks them with priority, and obtains the screened resource data; The node priority update submodule analyzes the construction nodes based on the filtered resource data, refers to the task urgency and resource allocation of the nodes, updates the execution priority of the nodes using the resource and task urgency matching index, and generates an updated node priority table; The resource management and scheduling submodule formulates resource allocation and scheduling strategies based on the updated node priority table, adjusts resource configuration of key nodes, optimizes the temporal and spatial distribution of resources, and forms a resource configuration plan for construction nodes.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by regional extraction and numbering of image data, combined with the subdivision technology of position coordinates, the accurate division of construction site partitions and the orderly organization of image data are realized, the structure and traceability of image data are guaranteed, the real-time image analysis process compares the image data with the construction plan model, accurately identifies the actual status of components in the area, and effectively improves the transparency and controllability of the construction progress, in the process of synchronizing the component status with the BIM model, timely reveals the construction deviation through status matching, ensures that any progress difference in the construction process can be quickly identified and recorded, and establishes a dynamic early warning mechanism for the identified deviation information, automatically marks abnormal areas and components, and prompts potential construction progress risks in advance to reduce safety hazards, and the dynamic resource allocation process combines real-time progress information with task priority to reasonably concentrate resources in key construction areas to ensure more efficient and accurate resource allocation, and further optimizes the construction nodes. Priority updates are performed on the distribution of tasks and resource use to ensure the rationality of resource allocation and orderly scheduling during the construction process, and provide full guarantees for key node tasks, and through efficient data extraction, accurate status comparison and automated resource allocation, significantly improve the intelligent level of construction management, reduce human intervention, and improve the safety and reliability of the construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a system flow chart of the present invention; Figure 2 It is a schematic diagram of the system framework of the present invention; Figure 3 It is a flow chart of the image capturing and partitioning module in the present invention; Figure 4 It is a flowchart of the construction status analysis module in the present invention; Figure 5 It is a flow chart of the component progress matching module in the present invention; Figure 6 This is a flow chart of the progress deviation analysis and early warning module in the present invention; Figure 7 It is a flowchart of the task dynamic allocation module in the present invention; Figure 8 It is a flow chart of the construction node optimization module in the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0020] Embodiment 1 See also Figure 1 to Figure 2 , a comprehensive construction site management system based on machine vision includes: The image capture and partitioning module extracts image content according to the set partitions based on the on-site image data stream, numbers the differentiated areas and organizes the image data, subdivides the area boundaries by position coordinates, arranges the partition images in order, and forms partition structure image data; the construction status analysis module extracts the real-time image of the components in the area through the area number in the partition structure image data, verifies the status identification of the construction plan model, and identifies and classifies the component status from the image, marks the status information of the partition, and generates the regional construction status information; The component progress matching module extracts the status of partitioned components based on the regional construction status information, verifies the real-time component status and the BIM model planned progress, identifies the differences in progress item by item through status matching, and extracts the component progress differences to form component progress difference data; The progress deviation analysis and early warning module verifies the difference value with the preset threshold parameters based on the component progress difference data, extracts the deviation component information, records the deviation status of the abnormal component, marks the deviation area and component position, and generates a progress deviation label; The task dynamic allocation module combines the components and partition status of the progress deviation label, extracts the key task area according to the priority, integrates the resource parameters in the partition, dynamically allocates resources to the deviation area, rearranges the task priority and execution order, and generates a construction task priority allocation plan; The construction node optimization module screens the current resource configuration data based on the task priority in the construction task priority allocation plan, updates the task distribution and resource usage priority, identifies key task nodes, optimizes the resource allocation strategy and node execution order, and generates a construction node resource configuration plan.

[0021] The partition structure image data includes numbering system, boundary definition, and image arrangement. The regional construction status information includes status identification, classification mark, and status record. The component progress difference data includes status matching results, progress verification records, and difference data. The progress deviation label includes deviation component identification, position marking, and deviation record. The construction task priority allocation plan includes task area positioning, dynamic resource configuration, and task sequence adjustment. The construction node resource allocation plan includes resource screening records, task priority updates, and resource allocation optimization.

[0022] Specifically, if Figure 2 , 3 As shown, the image capture and partitioning module includes: The data stream extraction submodule is based on the on-site image data stream. It extracts the content of each frame of the image frame according to the preset partitioning rules, marks the partitioned image frames according to the partition serial number, stores the image data in sequence and records the partition information of each frame, and classifies them into the corresponding image structure to obtain the image partition sequence number. The execution process is as follows: Based on the on-site image data stream, the content of each frame of the image is extracted frame by frame according to the preset partitioning rules. First, the image frame is divided according to the defined partitioning rules, and the image data is located and marked according to the partition position. The content partitions of each frame of the image are aligned according to the preset rules. The image content in each partition is pixel-separated by regional division, and the pixel density, color characteristics, edge information and other contents of each partition are analyzed in turn. After the key content features of each area are divided in turn, each partition of the image is feature-marked and numbered, and stored in the array file according to the partition serial number and the image frame order. The content of each partition is continuously marked according to the characteristics of the image sequence, and the partition information of each frame of the image is recorded. In the process of dividing and storing each frame of the image, the image partition sequence number is obtained by continuously recording the structured data content and image position information in the partition.

[0023] The difference number submodule performs differential analysis on the image content of each partition in the image frame based on the image partition sequence number, extracts the feature value of each partition content, arranges them in partition order, marks the feature sequence of each image content, and obtains the execution flow of the difference partition number data as follows: Extract the content feature value of each partition in the image frame according to the formula: Compute the partition difference eigenvalues, where The difference feature value representing the content of each partition in the image frame, Represents the pixel value of the current frame, Represents the corresponding pixel value of the previous frame, Represents the number of pixels in each partition; by monitoring each partition in the image frame sequence, the difference feature value of each partition is calculated frame by frame, and the difference feature of each frame partition is obtained by accumulating the difference of pixel values ​​based on the sum of squares of pixel differences. Suppose in a partition, each frame contains 5 pixels, and the pixel values ​​of the current frame and the previous frame are and , then the calculation formula of the difference eigenvalue is:

[0024]

[0025]

[0026] The result shows that the partition difference feature value is 83, which represents the content difference between the current frame and the previous frame in the partition. By calculating the difference feature values ​​of consecutive frames, the content changes can be tracked frame by frame and the difference partition number data can be extracted.

[0027] The boundary subdivision submodule calls the coordinate value of each numbered partition based on the difference partition number data, refines the boundary of the partition, matches the coordinate data of the area with the numbered partition, splices the image frame and the coordinate sequence, combines the image according to the real-time partition position, and obtains the partition structure image data. The execution process is as follows: Based on the differential partition numbering data, the coordinate values ​​of the acquired numbered partitions are refined and divided. First, the original coordinate value set of the partition is read, and the boundary of the partition is located and matched. According to the serial number and position of the partition and the coordinate characteristics of the partition, the image contour within the partition is pixel-level refined according to each boundary contour. The coordinate data is arranged according to the refinement rules within the partition and the boundary range of the partition is recorded. Through further boundary optimization, the boundary area of ​​each partition is connected in sequence, and it is ensured that there is no overlap and gap between the boundaries of each partition in the image frame to obtain accurate boundary contour information. A refined coordinate sequence is generated for the boundary contour data of each partition, and each image frame is spliced ​​according to the real-time partition position to form a new image structure file. Finally, the boundary data of each partition is combined to generate partition structure image data.

[0028] Specifically, if Figure 2 , 4 As shown, the construction status analysis module includes: The image extraction submodule reads the image data corresponding to each numbered area based on the area number in the partition structure image data, extracts each frame of the image in the order of the number, decomposes the independent image data of the component area, and arranges them by component number. The execution process of generating the component image sequence is as follows: Based on the area number in the partition structure image data, the image data corresponding to each numbered area is read frame by frame, and each frame of the image is partitioned and extracted in the order of the numbers. During the decomposition process, the image data is carefully classified according to the component number, and the independent image data of each component area is extracted and sorted, rearranged based on the component number as the main sequence, and stored as a component image sequence file in the order of the numbers. The entire extraction process depends on the accurate partitioning of the image data, and combined with the identification of the numbered partition sequence, by reading the component data of each frame of the image, it is ensured that the image data of each component in the decomposition process corresponds to a unique number to form a component image sequence, and the generated component image data set is stored by number, thereby realizing the serialized management of the component image data.

[0029] The status check submodule matches the status identifiers in the construction plan model item by item based on the component image sequence, extracts the status information of the component image, and checks the real-time status of the component with the planned status item by item, classifies the difference content into the corresponding component number, marks the corresponding difference information, and obtains the execution process of component status difference data as follows: Based on the component image sequence, the image status information of each component is checked one by one, the real-time status of each component is matched with the planned status, and compared item by item in combination with the status identification in the construction plan model. During the status verification process, the status information of the component image is matched and classified according to the status identification in the plan model. During the comparison process, the status features are extracted from the real-time status of the component image, including the status identification serial number, identification type and time stamp. The system detects the correspondence between the real-time status and the planned status, summarizes the status difference information of each component, and stores the difference content in the database associated with the component number. The difference content will be classified under the corresponding component number to form component status difference data, mark the difference status of the component, and provide basic data for subsequent status analysis and classification.

[0030] The status classification submodule is based on the component status difference data and applies the random forest algorithm to classify and identify the status content of the components in the difference data, integrate the status data into category indexes according to the status categories, extract the classified status information of the components in the partition, organize and summarize the status information of the partition, and obtain the execution process of the regional construction status information as follows: The formula of the random forest algorithm is as follows: in, Represents the degree of deviation of the component state. Representative The status value of a component, Representative The average value of the component status, Representative The weight of a component in the classification, n represents the total number of components; detailed explanation of the formula and the derivation process of the formula calculation: This formula is used to calculate the weighted average deviation of component state differences, where: Indicates the evaluation result of the overall classification accuracy or state deviation; parameter Indicates The actual status value of each component is obtained by collecting data through on-site monitoring equipment such as sensors. The status values ​​of three components are set to be 8, 10 and 12 respectively; parameter Indicates The expected average values ​​of the component states are based on historical data and environmental baseline adjustments, with the average values ​​set at 9, 11, and 13; parameter Representative The weight of each component in the evaluation model reflects the importance of the component or its impact on the overall system performance. The weight is determined based on the functional importance and failure frequency analysis of the component. The weights are set to 0.5, 1.0, and 1.5. Components with higher weights have greater influence in the overall evaluation. Calculate the deviation value of each component : For the first component, the deviation is ; For the second component, the deviation is ; Weighted deviation of the third component ; Then, multiply the bias by the weight: Weighted deviation of the first component ; Weighted deviation of the second component ; Weighted deviation of the third component ; Next, calculate the sum of the squares of the weights: ;calculate : The results show that the average deviation of the overall component status is 1.60 with reference to the weight factor. This value reflects the average status deviation after taking into account the importance of the component, which helps to evaluate the overall quality and accuracy of the regional construction status. By integrating and analyzing specific difference metrics, resource allocation and schedule scheduling during the construction process can be effectively managed and optimized.

[0031] Specifically, if Figure 2 , 5 As shown, the component progress matching module includes: The status information verification submodule extracts the current construction stage and properties of the component based on the regional construction status information and the partition component status, collects physical property data and identifies the process information, records the component's location information and process number, calibrates the matching relationship between the construction stage and the location data, and obtains the real-time component status data. The execution process is as follows: Based on the regional construction status information and the partition component status, the current construction stage and properties of each component are extracted, and the physical property data of the component is collected item by item. The collection process includes the collection of detailed information on the component's size, material composition and construction process. Each data collection records the specific location information of the component, which is processed through the process identification and forms a process number for subsequent calibration. The system matches the current construction stage of the component with its location data, and verifies the accuracy of the match to ensure that the actual position of the component is consistent with the location data in the construction progress. By analyzing the matching relationship between the construction stage and the position one by one, the position parameters are processed in the calibration process to obtain real-time component status data, providing high-precision basic data support for subsequent progress matching.

[0032] The progress matching and comparison submodule extracts the scheduled status and time nodes of the components in the planned schedule of the BIM model based on the real-time component status data. By comparing the status information with the time nodes, the advance and lag of each component is judged, and the status and time parameters are cross-checked to generate the progress difference comparison results. The execution process is as follows: Based on the real-time component status data, the real-time status of each component is compared with the planned schedule data in the BIM model one by one, and the scheduled status information and corresponding time nodes of each component in the BIM model are extracted. The comparison process adopts the cross-check method of component status and time nodes to identify whether the status of each component at the current time point meets the scheduled schedule requirements, calculate the progress offset data of the component item by item, identify the advance or lag status, and summarize the offset values, record the time difference of the component, generate the progress difference comparison result, and match the components with advanced or lagging progress with their planned status item by item to form a detailed difference information record file, which provides data basis for subsequent difference extraction and analysis.

[0033] Based on the progress difference comparison results, the difference extraction and analysis submodule applies the multivariate linear regression algorithm to analyze the state difference of each component, extract the time offset, position offset and process deviation parameters of the component, and summarize the parameters to obtain the component progress difference data. The execution process is as follows: The formula of the multivariate linear regression algorithm is as follows:

[0034] in, represents the predicted progress variance, represents the time offset, represents the position offset, represents process deviation, is the intercept term, is the weight coefficient of the variable, represents the error term; Detailed explanation of the formula and the process of formula calculation and derivation: The multivariate linear regression model used in this example is used to predict progress differences in construction projects. The parameters of the formula are obtained through data analysis. The specific data comes from the historical records and real-time monitoring system of the construction project. The value of (intercept term) is set to 0.5, which is obtained through regression analysis of historical project data and represents the basic schedule difference without any deviation; The value of (Time Offset) is set to 3 days. This data comes from the project's real-time tracking system and indicates the number of days the current checkpoint is delayed relative to the plan. The (time offset weight) is set to 0.2. This weight is determined by analyzing historical data of multiple similar projects and represents the rate at which the schedule variance increases for each additional day of time offset. The value of (position offset) is set to 2 meters. This data is obtained through the GIS positioning system at the project site and represents the distance deviation between the component position and the planned position; (Position offset weight) is set to 0.3. This coefficient is obtained based on historical data analysis and reflects the increase in progress difference for every additional meter of position offset; The value of (process deviation) is set to 1.5, which represents the deviation index between the actual process execution and the plan, obtained through the deviation recording function in the project management software; (Process Deviation Weight) is set to 0.4. This value is obtained by analyzing the correlation between process deviation and schedule impact in similar projects, which shows how the schedule difference increases for each unit increase in process deviation; The error term expresses the residual variation that the model cannot explain and is set to be normally distributed with a mean of 0 and a standard deviation of 0.1. To perform specific calculations, substitute the values ​​in the formula to expand:

[0035]

[0036] The results show that after considering time offset, location offset and process deviation, the predicted schedule difference is 2.3. This value represents the expected degree of schedule delay under given offset conditions, which helps the project management team adjust resource allocation and optimize the schedule to mitigate the negative impact of deviations.

[0037] Specifically, if Figure 2 , 6 As shown in the figure, the progress deviation analysis and early warning module includes: The deviation identification submodule extracts the real-time progress and planned progress values ​​of the components based on the component progress difference data, calculates the difference value, compares the difference value with the preset threshold parameter, marks the components that exceed the threshold range, records the deviation value and component identification information, and obtains the component deviation list. The execution process is as follows: Based on the component progress difference data, the real-time progress and planned progress values ​​of the components are extracted. First, the target time and actual completion time of the component in the construction plan are obtained, and the difference between the two is calculated. The time difference between the actual and planned progress is expressed in days, and then the advance or lag status of each component is analyzed. After calculating the difference value, the difference data is compared with the preset progress threshold one by one, and the components that exceed the preset threshold are marked as abnormal components. At the same time, the corresponding deviation value and component identification information are recorded, and the deviation value of each component is classified and recorded to generate a component deviation list, providing a clear data basis for subsequent abnormal records.

[0038] Based on the component deviation list, the abnormal record submodule reads the deviation value of each component, screens components in deviation status, extracts the position parameters of the components, combines the deviation status and deviation type with the position parameters, sorts the deviation type and value by component number, and obtains the abnormal component deviation status record. The execution process is as follows: Based on the component deviation list, the deviation value of each component is screened and classified, and the components with advanced and lagging progress are classified separately. Then, the position information of the components is read in turn, and each deviation is recorded according to the component number. According to the deviation status, the current coordinate position of the component is extracted, the position and status parameters are matched, the deviation type is sorted and a structured deviation data file is generated. The deviation status and deviation value are recorded item by item according to the component number in the file, and the deviation is refined and classified according to the severity and type of the deviation to form an abnormal component deviation status record. The data file can be directly used for subsequent label mapping processing to ensure the accuracy of the deviation status record.

[0039] The abnormal label mapping submodule reads the deviation area and position parameters of the abnormal component based on the abnormal component deviation status record, locates the component position, assigns the label type according to the deviation status and area information, and marks the label at the component position. The execution process of generating the progress deviation label is as follows: The deviation state of the component is combined with the position parameters according to the formula: Calculate the progress deviation label position, where, Represents the offset position of the label. Indicates the component deviation value, Represents positional parameters, is the deviation state quantity, is the position adjustment coefficient; according to the deviation state record of the component, the deviation value and the corresponding position parameter are obtained, and the display position of the label on the image is determined by calculating the position of the offset label. For example, if the component deviation value is set to 10, the position parameter , the deviation state quantity is 1, the position adjustment coefficient is 1.5, and the calculation formula is as follows:

[0040] The result shows that the label offset position is 21.5, which means that the label position needs to be offset by 21.5 units on the basic coordinates. This calculation can help accurately locate the progress deviation label of each abnormal component and generate label information corresponding to the actual position of the component, thereby achieving intuitive marking and display of the deviation area.

[0041] Specifically, if Figure 2 , 7 As shown, the task dynamic allocation module includes: The key area identification submodule analyzes the current status of each component and partition based on the progress deviation label, and focuses on the key areas as key task areas through deviation value sorting and urgency rating. The data of the area is aggregated and compiled into a list. The execution process of obtaining the key task area list is as follows: Based on the progress deviation label, the current status of components and partitions is analyzed item by item, the deviation value of each component is sorted, and the urgency level is calculated in turn. During the sorting process, the priority is determined according to the relationship between the deviation value and the time node, and the component areas with higher urgency are marked as critical task areas according to the urgency level. The key components and partition information in each partition are integrated, the data is aggregated into a table structure file, and the detailed content of each key area is recorded item by item in the data file, including information such as location, urgency level and deviation value, to form a list of critical task areas, providing data basis for subsequent resource allocation analysis.

[0042] The resource integration submodule reviews the resource allocation of the region based on the list of key mission areas, including manpower, machinery and materials. By matching the demand with the current resources, the execution process of generating a resource allocation plan is as follows: Based on the list of key task areas, the resource demand situation of each key area in the list is reviewed item by item. The system checks the manpower, machinery and equipment, materials and other resources currently configured in each area. By comparing and analyzing each demand with the current resource configuration status, the actual resource availability and task demand rate are calculated in turn, and the resource shortage situation is recorded by category. A resource demand difference table file is generated to further refine the gap situation of each type of resource. According to the results of demand matching analysis, available resources are deployed item by item to meet regional needs, and a resource allocation plan is formed to ensure that the resource allocation situation in key task areas is consistent with actual needs, providing resource data support for the priority configuration of construction tasks.

[0043] The task priority adjustment submodule evaluates and reorders tasks based on the resource allocation plan. It adjusts the current task list and optimizes the execution order of tasks through task urgency analysis and resource matching status check. The execution process of the construction task priority allocation plan is as follows: Task priority adjustment is based on urgency and resource matching status, through the formula:

[0044] Calculate the task priority, where Represents the priority of the task. Indicates the degree of urgency, Indicates the resource matching status. and is the weight coefficient, which reflects the priority of the task; the priority adjustment is based on the urgency of the task and the actual matching of resources. The weight coefficient is used to adjust the weight of each influencing factor, and the task priority is calculated by the weighted sum of the urgency and resource matching status. , resource matching status , weight coefficient , For example, the priority calculation formula is as follows:

[0045]

[0046] The result shows that the priority value of the task is 6.1. The larger the value, the higher the priority. By calculating the priority value, the task list is sorted and adjusted to better optimize the execution order of construction tasks, thereby achieving priority allocation of key tasks and efficient use of resources.

[0047] Specifically, if Figure 2 , 8 As shown in the figure, the construction node optimization module includes: The resource data screening submodule reviews the current resource configuration data based on the construction task priority allocation plan, evaluates the efficiency of each resource point, determines the resource demand intensity of the task through quantitative analysis, screens the resource points and prioritizes them to obtain the filtered resource data. The execution process is as follows: Based on the construction task priority allocation plan, the current resource configuration data is carefully reviewed to quantitatively evaluate the performance of each resource point. By statistically analyzing the application frequency, resource consumption, workload rate and other data of each resource point in historical tasks, a resource efficiency evaluation model is created. The efficiency value of each resource point is obtained through this model, and the degree of match with the task requirements is calculated to form a score. Based on this score, the priority of resource demand intensity is determined. Resource points with scores higher than the set threshold are screened and marked as priority resources to form filtered resource data.

[0048] The node priority update submodule analyzes the construction nodes based on the filtered resource data, and updates the execution priority of the nodes using the resource and task urgency matching index based on the task urgency and resource allocation of the nodes. The execution process of generating the updated node priority table is as follows: Analyze the task urgency and resource allocation of the construction nodes, according to the formula: The execution priority of the compute node, where Indicates the execution priority of the node. Indicates the urgency score of the node task, Indicates the total amount of resources required by the node. Indicates the total amount of resources that can be allocated in the filtered resource data; Urgency score Obtained through the historical construction schedule, assuming that the urgency score of a certain node construction task is 8, the total resource demand The total amount of resources that can be deployed is calculated to be 120 units according to the construction progress requirements. The current resource data table is 100 units, so the priority Calculated as: The result shows that the execution priority of the node is 9.6, which means that in the trade-off between task urgency and resource demand, the node has a higher priority and resource scheduling should be given priority.

[0049] The resource management and scheduling submodule formulates resource allocation and scheduling strategies based on the updated node priority table, adjusts the resource allocation of key nodes, optimizes the spatiotemporal distribution of resources, and forms the execution process of the resource allocation plan for construction nodes as follows: Based on the updated node priority table, a resource allocation strategy is formulated to ensure the optimal allocation of resources required for key nodes. First, all node priorities are counted and arranged from high to low in order of priority. The resource requirements of the top-ranked key nodes are matched one by one from the screened resource data. The resource allocation strategy is adjusted by comparing the gap between resource requirements and available resources. For key nodes with large gaps, excess resources are allocated first to optimize the spatiotemporal distribution structure of resources, thereby forming a resource allocation plan for construction nodes.

[0050] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A comprehensive construction site management system based on machine vision, characterized in that: The system comprises: The image capture and partitioning module is based on the on-site image data stream, according to the differentiated area numbers, subdividing the area boundaries, and arranging the partition images in sequence to form partition structure image data; The construction status analysis module extracts the real-time image through the area number in the partition structure image data, verifies the construction plan status, identifies and classifies the component status, and generates regional construction status information; The component progress matching module extracts the component status through the regional construction status information, checks the progress difference with the BIM model plan, and forms component progress difference data; The progress deviation analysis and early warning module verifies the component progress difference data with the preset threshold, extracts the deviation component information, records the deviation status of the abnormal component, marks the deviation area and component position, and generates a progress deviation label; The task dynamic allocation module combines the components and partition status of the progress deviation label, extracts the key task area, integrates the partition resources, dynamically allocates resources to the deviation area, rearranges the task priority and execution order, and generates a construction task priority allocation plan; The construction node optimization module screens the current resource configuration data based on the construction task priority allocation scheme, updates the task distribution and resource usage priority, and generates a construction node resource configuration scheme.

2. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The partition structure image data includes a numbering system, boundary definition, and image arrangement; the regional construction status information includes status identification, classification mark, and status record; the component progress difference data includes status matching results, progress verification records, and difference data; the progress deviation label includes deviation component identification, position marking, and deviation record; the construction task priority allocation plan includes task area positioning, dynamic resource configuration, and task sequence adjustment; the construction node resource configuration plan includes resource screening records, task priority updates, and resource configuration optimization.

3. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The image capture and partitioning module includes: The data stream extraction submodule extracts the content of each frame of the image according to the preset partitioning rules based on the live image data stream, marks the partitioned image frames according to the partition sequence number, stores the image data in sequence and records the partition information of each frame, and classifies it into the corresponding image structure to obtain the image partition sequence number; The difference numbering submodule performs a difference analysis on the image content of each partition in the image frame based on the image partition sequence number, extracts the feature value of each partition content, arranges them in the partition order, marks the feature sequence of each image content, and obtains the difference partition numbering data; The boundary subdivision submodule calls the coordinate value of each numbered partition based on the difference partition number data, refines the boundary of the partition, matches the coordinate data of the area with the numbered partition, splices the image frame and the coordinate sequence, combines the image according to the real-time partition position, and obtains the partition structure image data.

4. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The construction status analysis module includes: The image extraction submodule reads the image data corresponding to each numbered area based on the area number in the partition structure image data, extracts each frame of image in the order of number, decomposes the independent image data of the component area, and arranges them by component number to generate a component image sequence; The status check submodule matches the status identifiers in the construction plan model item by item based on the component image sequence, extracts the status information of the component image, and checks the real-time status of the component with the planned status item by item, classifies the difference content into the corresponding component number, marks the corresponding difference information, and obtains the component status difference data; The status classification submodule applies the random forest algorithm based on the component status difference data to classify and identify the status content of the components in the difference data, integrates the status data into category indexes according to the status categories, extracts the classified status information of the components in the partition, organizes and summarizes the status information of the partition, and obtains the regional construction status information.

5. The machine vision-based integrated construction site management system according to claim 4, characterized in that: The formula of the random forest algorithm is as follows: in, Represents the degree of deviation of the component state. Representative The status value of a component, Representative The average value of the component status, Representative The weight of a component in the classification, and n represents the total number of components.

6. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The component progress matching module includes: The status information verification submodule extracts the current construction stage and properties of the component based on the regional construction status information and the partition component status, collects physical property data and identifies the process information, records the component location information and process number, calibrates the matching relationship between the construction stage and the location data, and obtains real-time component status data; The progress matching and comparison submodule extracts the predetermined status and time nodes of the components in the BIM model plan schedule based on the real-time component status data, judges the advance and lag of each component by comparing the status information with the time nodes, performs cross-checking of the status and time parameters, and generates progress difference comparison results; The difference extraction and analysis submodule applies a multivariate linear regression algorithm based on the progress difference comparison result to analyze the state difference of each component, extract the time offset, position offset and process deviation parameters of the component, summarize the parameters, and obtain the component progress difference data.

7. The machine vision-based integrated construction site management system according to claim 6, characterized in that: The formula of the multivariate linear regression algorithm is as follows: in, represents the predicted progress variance, represents the time offset, represents the position offset, represents process deviation, is the intercept term, is the weight coefficient of the variable, represents the error term.

8. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The progress deviation analysis and early warning module includes: The deviation identification submodule extracts the real-time progress and the planned progress values ​​of the components based on the component progress difference data, calculates the difference value, compares the difference value with the preset threshold parameter, marks the components exceeding the threshold range, records the deviation value and component identification information, and obtains the component deviation list; The abnormal record submodule reads the deviation value of each component based on the component deviation list, screens components in deviation state, extracts the position parameters of the components, combines the deviation state and the deviation type with the position parameters, sorts the deviation type and value according to the component number, and obtains the abnormal component deviation state record; The abnormal label mapping submodule reads the deviation area and position parameters of the abnormal component based on the abnormal component deviation status record, locates the component position, assigns the label type according to the deviation status and area information, marks the label at the component position, and generates a progress deviation label.

9. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The task dynamic allocation module includes: The key area identification submodule analyzes the current status of each component and partition based on the progress deviation label, focuses on the key areas as key task areas through deviation value sorting and urgency rating, aggregates and compiles the data of the areas into a list, and obtains a list of key task areas; The resource integration submodule reviews the resource allocation of the area, including manpower, machinery and materials, based on the list of key task areas, and generates a resource allocation plan through matching analysis between demand and current resources; The task priority adjustment submodule evaluates and reorders the tasks based on the resource allocation plan, adjusts the current task list through task urgency analysis and resource matching status check, optimizes the execution order of tasks, and builds a construction task priority allocation plan.

10. The machine vision-based integrated construction site management system according to claim 1, characterized in that: The construction node optimization module includes: The resource data screening submodule reviews the current resource configuration data based on the construction task priority allocation plan, evaluates the performance of each resource point, determines the resource demand intensity of the task through quantitative analysis, screens the resource points and marks them with priority, and obtains the screened resource data; The node priority update submodule analyzes the construction nodes based on the filtered resource data, refers to the task urgency and resource allocation of the nodes, updates the execution priority of the nodes using the resource and task urgency matching index, and generates an updated node priority table; The resource management and scheduling submodule formulates resource allocation and scheduling strategies based on the updated node priority table, adjusts resource configuration of key nodes, optimizes the spatial and temporal distribution of resources, and forms a resource configuration plan for construction nodes.

Images