Traffic engineering construction management method and system based on intelligent perception

Abstract

The invention discloses a traffic engineering construction management method and system based on intelligent perception, and belongs to the technical field of construction management. The system comprises an environment sensing module, an intelligent analysis module, a risk assessment module and an early warning prompt module. The environment sensing module is used for collecting GIS map and environment data, azimuth parameters of operation objects in a construction area and operation parameters of moving objects in a road area, and constructing a three-dimensional scene; the intelligent analysis module analyzes the state of the operation object according to the azimuth parameters and sets equivalent points in the three-dimensional scene; analyzing and calculating a weight coefficient of the moving object in combination with the environment data and the operation parameters; the risk assessment module sets detection points for the equivalent points, analyzes the parameters of the driven-in moving object, calculates the risk coefficient of the equivalent points and the risk index of the operation object in combination with the weight coefficient, marks the operation object and sets a safety point; the early warning prompt module gives out acousto-optic early warning through the helmet, and voice prompt marks that the operation object immediately moves to a safety point for danger avoiding.

Description

Technical Field

[0001] This invention relates to the field of construction management technology, specifically to a method and system for construction management of traffic engineering based on intelligent sensing. Background Technology

[0002] In the field of traffic engineering construction safety management, ensuring the safety of construction personnel is the primary objective, especially in construction scenarios near roads where the construction area is easily disturbed by passing vehicles, posing a significant collision risk. Currently, traditional monitoring technologies mainly rely on video surveillance, GPS positioning, and basic data analysis. However, with the increasing complexity of projects, these systems have exposed a series of key flaws that seriously endanger personnel safety.

[0003] First, environmental perception capabilities are insufficient. Existing technologies generally rely on single sensors to collect information, making it difficult to effectively integrate and collaboratively process diverse and heterogeneous data. This results in low scene modeling accuracy, blurred construction boundary identification, and difficulties in determining operational status. Furthermore, location tracking relies solely on coarse video analysis or static coordinate input, lacking real-time dynamic spatial mapping capabilities. Second, intelligent analysis mechanisms are lagging. Operational risk assessments often use fixed height thresholds as a binary judgment criterion, ignoring the impact of dynamic buffering effects on support surfaces, such as the energy absorption characteristics of facilities. In addition, the lack of a systematic vehicle stability analysis mechanism leads to a significant increase in the misjudgment rate of high-risk vehicles. Finally, early warning mechanisms are ineffective. Existing technologies mostly use a unified alarm mode rather than being driven by individualized risk indices. Audible and visual alarm signals fail in strong light or long-distance environments, and the lack of guidance on safe evacuation points prolongs emergency response delays. These problems collectively exhibit fragmented, static, and data-siloed characteristics, severely restricting the overall effectiveness of the safety protection system. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for traffic engineering construction management based on intelligent perception, so as to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides a traffic engineering construction management method based on intelligent sensing, comprising:

[0006] S100 collects GIS maps and environmental data, as well as the orientation parameters of objects working within the construction area and the running parameters of moving objects within the road area, thereby constructing a three-dimensional scene.

[0007] Environmental data includes image data, BIM data, and scan data; image data refers to surveillance videos of construction areas and road areas.

[0008] BIM data includes 3D models of various facilities and equipment used for engineering operations within the construction area. Scan data refers to 3D physical space data of the construction area collected through multi-source sensing devices.

[0009] GIS maps provide a global geographic coordinate system, BIM data injects semantic information about engineering facilities, and laser scanning point clouds fill in spatial details. The three are integrated with centimeter-level accuracy through spatial registration algorithms.

[0010] A construction zone refers to a specific area pre-demarcated during the implementation of traffic engineering projects to ensure the safety of engineering operations. A road area refers to the portion of the road coverage area whose average distance from the boundary of the construction zone is less than a threshold.

[0011] The work object refers to the workers performing operations within the construction area. Orientation parameters refer to the spatial position of the human body, which is collected in real time by sensors on the safety helmets worn by the workers.

[0012] The moving object refers to a vehicle traveling within the road area. Operating parameters include the vehicle's mass, position, speed, and direction. Mass is acquired by a piezoelectric thin-film sensor, while position, speed, and direction are obtained through real-time image data analysis.

[0013] The helmet sensor updates the personnel coordinates at a frequency of 10Hz, and the vehicle operation parameters are extracted from the monitoring video through the target tracking algorithm to form a dynamic object trajectory data stream.

[0014] Digital twin technology is used to construct a three-dimensional scene based on GIS maps, scanned data, and BIM data, and construction areas and road areas are divided at corresponding locations in the three-dimensional scene.

[0015] The terrain mesh is reconstructed based on the scanned data using a 3D engine, parametric components from the BIM model library are overlaid, and the construction area and road area are distinguished through shader programming to achieve visual labeling of risk areas.

[0016] The relative position of the work object within the construction area is analyzed using orientation parameters and image data, and then mapped in real time. The relative position of the moving object within the road area is analyzed based on image data, and then mapped in real time.

[0017] By fusing multi-source data to construct a real-time mapped digital twin scenario, a high-precision spatial foundation is provided for dynamic risk analysis.

[0018] S200: Analyze the state of the work object based on the orientation parameters and set equivalent points in the 3D scene. Combine environmental data and operational parameters to analyze and calculate the weight coefficients of the moving object. Specifically, this includes:

[0019] S201. Analyze the three-dimensional coordinates corresponding to the spatial positions of all work objects in the three-dimensional scene. ,Will Work objects with values ​​greater than the threshold are set to the off-ground state, while other work objects are set to the stationed state.

[0020] The threshold for the off-ground state is set based on the statistics of facility and equipment height. The coordinates of the support surface are extracted by point cloud segmentation. If the distance between the point cloud of a person's feet and the support surface is greater than the threshold, it is judged as an off-ground state.

[0021] S202, Transfer the two-dimensional coordinates of the work objects in the site status. The equivalent point is the two-dimensional coordinate of each position where the facilities and equipment to which the work object is attached contact the ground. These are respectively used as the positions of their equivalent points.

[0022] S203. Analyze the running parameters of all moving objects in the 3D scene, and combine historical image data to analyze following distance and following speed. Preset standard speed. and standard spacing The weight coefficients of each moving object are calculated. Specifically, this includes:

[0023] S2031. Obtain historical image data and extract images containing moving objects. Video clips, calculating moving objects average speed and the average speed of all moving objects. .

[0024] S2032, Moving objects The video clip is divided into different following time periods based on the condition that the vehicle is driving directly behind other moving objects and the distance between them is less than a threshold. Within each following time period, different following time periods are evenly distributed. A specific point in time.

[0025] S2033. Analyze moving objects at each time point. The instantaneous velocity and the instantaneous time distance between the vehicle and the moving object in front are used to calculate the standard deviation of the instantaneous velocity at all time points within each following period. and instantaneous time distance standard deviation .

[0026] Spatiotemporal slicing technology is used to segment the video stream, and the rate of change of vehicle spacing and standard deviation are calculated based on optical flow. Reflecting speed fluctuations, It indicates the degree of following closely behind the vehicle.

[0027] S2034. Calculate the moving objects within each following time period. average speed and the average distance between the moving object in front. Substitute into the formula to calculate the moving object Weighting coefficients :

[0028] ;

[0029] In the formula, A constant greater than 1 This represents the number of times the vehicle will accompany the passenger during all scheduled times.

[0030] The risk level of vehicle behavior is quantified by identifying abnormal speeding vehicles by comparing the deviation of individual speeds from the overall traffic flow speed. At the same time, the instantaneous speed fluctuations and spacing during the following period are combined to reflect the vehicle's driving stability.

[0031] The dynamic weighting of historical data and preset safety benchmarks can accurately identify high-risk vehicles, providing key input for subsequent collision risk assessments.

[0032] S2035, and so on, extract video segments for each moving object and calculate weight coefficients.

[0033] Accurately identify personnel's working posture, establish vehicle behavior risk indicators, and solve the problem of heterogeneous data correlation between "people and vehicles" in traditional methods.

[0034] S300: Set different detection points for equivalent points and analyze the parameters of moving objects entering the detection points. Combine the weighting coefficients to calculate the risk coefficient and risk index of the corresponding equivalent point and the work object, mark the work object, and set safety points. Specifically, this includes:

[0035] S301. Analyze the positions of all equivalent points in the 3D scene, and set a location for each equivalent point within the road area. Each detection point has a reference velocity and reference direction determined by analyzing historical image data. Specifically, this includes:

[0036] S3011, evenly distributed within the road area Using the sampling point as the starting point, a vector is first established according to the lane direction. Then according to the equivalent point Establish a vector for position and direction Calculate vector and The included angle .

[0037] S3012. Analyze the relationship between each sampling point and the equivalent point. Location distance Therefore, priority coefficients are calculated, and sampling points are arranged in ascending order of priority coefficients. The first sampling point is selected. Each sampling point is used as an equivalent point. The detection point.

[0038] Priority coefficient The formula is:

[0039] ;

[0040] In the formula, It is a constant greater than 1. Distance weights The included angle weight.

[0041] Prioritize the deployment of detection points and comprehensively evaluate the spatial relationship between road sampling points and construction equivalent points. Priority coefficients assign higher weight to distance factors, while introducing directional angle corrections to ensure that the detection point layout covers potential conflict areas and conforms to the natural trends of vehicle movement trajectories, thereby optimizing risk monitoring efficiency.

[0042] S3013. Extract the image frame where the detection point is triggered from the historical image data, and analyze the speed and direction of the moving object when it reaches the detection point in each image frame. "Trigger" refers to the image frame captured at the instant the moving object enters the detection point.

[0043] S3014. Calculate the average velocity across all image frames at the same detection point, and use it as the reference velocity for that detection point. Fit the direction vector across all image frames at the same detection point, and use it as the reference direction for that detection point.

[0044] Since all direction vectors originate from the same detection point, the fitting methods include single-step fitting and successive fitting.

[0045] A single fitting requires a reference direction vector. The average angle between the fitting and all other direction vectors is calculated to obtain the fitted reference direction.

[0046] Successive fitting involves constructing an angle between two direction vectors in consecutive time and taking the median vector, then continuing to construct an angle between the vectors and subsequent direction vectors and taking the median vector, and so on, until all iterations are completed to obtain the fitted reference direction.

[0047] S3015, and so on, sets for each equivalent point respectively. The system identifies several detection points and analyzes the reference velocity and reference direction for each point.

[0048] S302. Analyze the parameters of each moving object entering the detection point through image data analysis, set the kinetic energy transfer coefficient according to the status of the work object corresponding to the equivalent point, and calculate the risk coefficient of each equivalent point in combination with the weighting coefficient. Specifically, this includes:

[0049] S3021. Analyze the current image data and capture the entry equivalent point. The moving objects at each detection point are analyzed, including their weight coefficients, instantaneous speed, instantaneous direction, and time of entry.

[0050] The detection point uses electronic fence technology. When the vehicle's center of gravity enters a virtual circle with a radius of 1.5m, it triggers a snapshot with a millisecond-level response.

[0051] S3022 is the equivalent point. Set the kinetic energy transfer coefficient for the corresponding work object. The state is analyzed. In the stationary state, the kinetic energy transfer coefficient is 1, and in the off-ground state, the kinetic energy transfer coefficient is less than 1 and greater than 0.

[0052] The kinetic energy transfer coefficient is a core parameter characterizing the proportion of kinetic energy effectively transferred from a moving object to the working object during a collision.

[0053] When the object being worked on is on a rigid ground, the feet / support points form a displacement constraint boundary, and the impact energy of the vehicle is directly transmitted to the incompressible foundation through the human body, causing the energy reflectivity to approach zero.

[0054] At this point, the human body is almost a rigid coupling medium, and the kinetic energy conversion efficiency reaches its theoretical maximum value.

[0055] When the object being worked on is located on the facility or equipment, there is a non-holonomic constraint between the worker and the equipment. Collisions induce a throwing motion of the human body, and some of the kinetic energy is converted into potential energy and rotational kinetic energy. The effective impact energy is dissipated in multiple stages.

[0056] Scaffolding components, lifting platforms, and other equipment can be considered as composite damping systems, where some kinetic energy is dissipated as structural vibration energy.

[0057] The kinetic energy transfer coefficient is a bridge variable that converts the risk of vehicle impact into the degree of damage to the workpiece. Its core logic lies in:

[0058] Collision system stiffness matching: When working on the ground, the vehicle, human body and foundation form an infinitely rigid system, while when working off the ground, it is reduced to a finitely rigid system.

[0059] Differences in energy form conversion: rigid connections transmit pure kinetic energy, while flexible connections convert some of it into sound energy, heat energy, and potential energy.

[0060] S3023. Count the number of moving objects entering at each detection point. Get the current time Substitute into the formula to calculate the equivalent point risk factor :

[0061] ;

[0062] In the formula, The time decay coefficient, and The first The detection point was the first The time and instantaneous speed at which a moving object enters. The preset standard duration, It is a constant. For the first Reference velocity of each detection point.

[0063] and The first The first detection point The weighting coefficients and quality of each moving object. This represents the average weighting coefficient of all moving objects within the road area.

[0064] For the first The detection point was the first The instantaneous direction of the moving object entering is the same as that of the first moving object. The angle between the reference directions of each detection point, For the first The detection point was the first The instantaneous direction and equivalent point of a moving object entering. The angle between the position and direction.

[0065] The risk coefficient is used to dynamically calculate the real-time threat level of vehicle collisions to construction sites. The formula incorporates multi-dimensional parameters:

[0066] Time sensitivity: The impact of recently arrived vehicles is greater than that of earlier events.

[0067] Kinetic energy transfer efficiency: Adjust the collision energy conversion ratio according to the working height.

[0068] Behavioral hazards: speeding, deviating from the intended direction, or vehicles with high weighting coefficients significantly amplify the risk.

[0069] Spatial correlation: The smaller the angle between the vehicle's driving direction and the equivalent point's orientation, the higher the probability of a collision.

[0070] By normalizing the data to offset the interference of environmental variables, continuous risk values ​​are output to drive early warning decisions.

[0071] Similarly, calculate the risk coefficient for each equivalent point.

[0072] S303. Calculate the risk index of the work object based on the risk coefficient, mark work objects with a risk index higher than the threshold, and set safety points around the marked work objects. Specifically, this includes:

[0073] S3031. The average risk coefficient of all equivalent points under the work object is used as the risk index, and a risk index greater than the threshold is marked. The task object.

[0074] S3032, Set Distance Using the position of each equivalent point under the marked work object as the center, the distance is... Divide the 3D scene into circular regions with a radius.

[0075] S3033. Set a two-dimensional coordinate axis at the center of each circular area, and take the four points where the boundary of the circular area intersects with the coordinate axis as planning points, and calculate the risk coefficient of each planning point.

[0076] S3034. Determine whether the risk coefficient of all planning points is greater than the threshold. If the result is negative, then the planning point with the lowest risk coefficient is taken as the safe point.

[0077] S3035 If the result is yes, then divide the area into circular regions with the position of each planning point as the center, and continue to set planning points until the conditions in step S3034 are met. Then stop setting planning points and take the planning point with the lowest risk coefficient as the safe point.

[0078] The algorithm intelligently plans safe haven locations for high-risk workers. It delineates a circular safety zone with an equivalent point as the center and a fixed distance as the radius, prioritizing the location with the lowest risk coefficient among the boundary intersection points.

[0079] If all locations exceed the threshold, the search range is expanded incrementally, iterating until a location meeting the safety criteria is found. This rule ensures the shortest possible escape path and consistently avoids real-time risk hotspots.

[0080] Predicting collision probabilities in the spatiotemporal dimensions and automatically planning the optimal avoidance path enables "risk-driven" proactive protection.

[0081] The S400 system automatically issues an audible and visual warning when marking the work object's head and prompts the user to move to a safe point for safety.

[0082] The helmet has a built-in LoRa module to receive warning commands and uses PWM to control the flashing of the three-color LEDs and the frequency of the buzzer. It is visible for more than 50m in strong light.

[0083] Risk assessment results drive precise individual early warnings, shortening emergency response delays.

[0084] The present invention also provides a traffic engineering construction management system based on intelligent perception, including an environmental perception module, an intelligent analysis module, a risk assessment module, and an early warning module.

[0085] The environmental perception module is used to collect GIS maps and environmental data, the orientation parameters of objects working in the construction area, the running parameters of moving objects in the road area, and to construct a three-dimensional scene.

[0086] Collect GIS map data, environmental data, and obtain in real time the orientation parameters of the work objects and the running parameters of the moving objects.

[0087] A three-dimensional scene was constructed using digital twin technology, dividing the area into construction zone and road zone.

[0088] It provides accurate real-time environmental mapping and visualization, supports subsequent risk analysis, enhances the perception capability of the construction area, reduces human monitoring errors, and improves the overall safety foundation.

[0089] The intelligent analysis module analyzes the state of the work object based on orientation parameters and sets equivalent points in the 3D scene. It then combines environmental data and operational parameters to calculate the weight coefficients of the moving object.

[0090] Analyze the status of the work objects and set equivalent points. Calculate the weighting coefficients of moving objects by combining historical image data, including analyzing parameters such as the standard deviation of speed and the standard deviation of distance during the following period, and use formulas to quantify vehicle risk factors.

[0091] By using intelligent algorithms to identify potential risk sources, the accuracy of vehicle instability assessment can be improved, assisting in early warning decisions and reducing the misjudgment rate of dynamic traffic interference with construction.

[0092] The risk assessment module sets detection points for the equivalent point and analyzes the parameters of the moving object that has entered. It calculates the risk coefficient of the equivalent point and the risk index of the work object by combining the weighting coefficient, marks the work object and sets the safety point.

[0093] For each equivalent point, a detection point is set, the parameters of the moving object entering the detection point are analyzed, and the risk coefficient of the equivalent point and the risk index of the working object are calculated in combination with the status of the working object.

[0094] After marking high-risk work objects, safety points are dynamically generated by iteratively setting planning points in the 3D scene.

[0095] Dynamically quantify on-site risk levels, provide risk avoidance guidance, achieve proactive protection, shorten emergency response time, and minimize the probability of collision accidents.

[0096] The early warning module emits an audible and visual warning by having a safety helmet on, and the voice prompts the marked work object to immediately move to a safe point to avoid danger.

[0097] When the risk assessment module marks a high-risk work object, it automatically triggers an audible and visual warning function for the work object wearing a safety helmet, prompting personnel to move to a preset safe point. The warning command is updated in real time based on risk coefficient calculation.

[0098] It provides immediate, visual and audible warning signals to guide workers to avoid risks efficiently, enhances personnel safety compliance, and thus directly intervenes in potential accidents, reducing construction injury rates and downtime losses.

[0099] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0100] Advantages of full-domain dynamic perception: Integrating GIS maps, multi-source environmental data (monitoring video / BIM model / 3D scan), and real-time sensor data, a high-precision dynamic 3D twin scene is constructed, breaking through the limitations of traditional static monitoring. Through dual real-time mapping of vehicle operation parameters (mass, speed, direction) and worker location parameters, a comprehensive digital reconstruction of the physical space of the construction area and road area is achieved.

[0101] Advantages of Risk Quantification Modeling: The solution introduces a two-layer risk assessment architecture of "equivalent point-detection point". On the one hand, it intelligently sets the kinetic energy transfer coefficient based on the operational status (station / off-ground) to accurately distinguish the difference between ground impact damage and high-altitude energy dissipation. On the other hand, it dynamically quantifies the hazard level of moving objects through behavioral analysis weighting coefficients such as vehicle following distance and speed stability. This two-dimensional model significantly improves the physical accuracy of risk assessment.

[0102] Advantages of intelligent risk avoidance decision-making: Constructing a proactive risk avoidance mechanism based on "risk coefficient → safe point". Based on real-time vehicle entry parameters of equivalent points, combined with a spatiotemporal decay function, dynamic risk calculation is achieved, and the optimal risk avoidance position is automatically generated—through a circular area recursive iterative algorithm, low-risk planning points are intelligently selected as target safe points, overcoming the limitations of passive alarms.

[0103] Advantages of human-machine collaborative early warning: Establishing a closed-loop control system between the head-mounted device and the central system. When the risk index exceeds the threshold, it directly triggers an audible and visual warning for the worker's safety helmet. Combined with the preset spatial coordinates of safety points in the 3D scene, it achieves a "risk warning - evacuation guidance" response within seconds, significantly improving emergency decision-making efficiency and personnel survival rate.

[0104] Advantages of multimodal data fusion: It breaks through the bottleneck of single data source, integrates multimodal data such as historical video feature analysis (following behavior statistics), real-time image processing (vehicle trajectory capture), sensor network (quality / location acquisition), and spatial geometric calculation (vector angle fitting), and constructs a risk assessment model with spatiotemporal continuity to enhance the system's adaptability in complex traffic environments. Attached Figure Description

[0105] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0106] Figure 1This is a flowchart illustrating the intelligent sensing-based traffic engineering construction management method of the present invention.

[0107] Figure 2 This is a schematic diagram of the structure of the intelligent sensing-based traffic engineering construction management system of the present invention. Detailed Implementation

[0108] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0109] Example 1: Please refer to Figure 1 This invention provides a traffic engineering construction management method based on intelligent sensing, comprising:

[0110] S100 collects GIS maps and environmental data, as well as the orientation parameters of objects working within the construction area and the running parameters of moving objects within the road area, thereby constructing a three-dimensional scene.

[0111] Environmental data includes image data, BIM data, and scan data; image data refers to surveillance videos of construction areas and road areas.

[0112] BIM data includes 3D models of various facilities and equipment used for engineering operations within the construction area. Scan data refers to 3D physical space data of the construction area collected through multi-source sensing devices.

[0113] In the specific implementation process, GIS map provides a global geographic coordinate system, BIM data injects semantic information of engineering facilities (such as equipment size and material properties), and laser scanning point cloud fills in spatial details. The three are integrated with centimeter-level accuracy through spatial registration algorithms (such as ICP iterative nearest point).

[0114] A construction zone refers to a specific area pre-demarcated during the implementation of traffic engineering projects to ensure the safety of engineering operations. A road area refers to the portion of the road coverage area whose average distance from the boundary of the construction zone is less than a threshold.

[0115] The work object refers to the workers performing operations within the construction area. Orientation parameters refer to the spatial position of the human body, which is collected in real time by sensors on the safety helmets worn by the workers.

[0116] The moving object refers to a vehicle traveling within the road area. Operating parameters include the vehicle's mass, position, speed, and direction. Mass is acquired by a piezoelectric thin-film sensor, while position, speed, and direction are obtained through real-time image data analysis.

[0117] In the specific implementation process, the helmet sensor (UWB positioning chip) updates the personnel coordinates at a frequency of 10Hz, and the vehicle operation parameters are extracted from the monitoring video through the target tracking algorithm (such as DeepSORT) to form a dynamic object trajectory data stream.

[0118] Digital twin technology is used to construct a three-dimensional scene based on GIS maps, scanned data, and BIM data, and construction areas and road areas are divided at corresponding locations in the three-dimensional scene.

[0119] In the specific implementation process, a 3D engine (such as Unity / Unreal) is used to reconstruct the terrain mesh based on the scan data, and parametric components (such as the LOD model of scaffolding steel pipes) in the BIM model library are overlaid. The construction area (red semi-transparent) and the road area (blue wireframe) are distinguished by shader programming to achieve the visualization and labeling of risk areas.

[0120] The relative position of the work object within the construction area is analyzed using orientation parameters and image data, and then mapped in real time. The relative position of the moving object within the road area is analyzed based on image data, and then mapped in real time.

[0121] By fusing multi-source data to construct a real-time mapped digital twin scenario, a high-precision spatial foundation is provided for dynamic risk analysis.

[0122] S200: Analyze the state of the work object based on the orientation parameters and set equivalent points in the 3D scene. Combine environmental data and operational parameters to analyze and calculate the weight coefficients of the moving object. Specifically, this includes:

[0123] S201. Analyze the three-dimensional coordinates corresponding to the spatial positions of all work objects in the three-dimensional scene. ,Will Work objects with values ​​greater than the threshold are set to the off-ground state, while other work objects are set to the stationed state.

[0124] In the specific implementation process, the threshold for the off-ground state is set based on the statistics of facility and equipment height (such as the average platform height of engineering vehicles is 1.5m). The coordinates of the support surface are extracted by point cloud segmentation. If the distance between the point cloud of a person's feet and the support surface is greater than the threshold, it is judged as an off-ground state.

[0125] S202, Transfer the two-dimensional coordinates of the work objects in the site status. The equivalent point is the two-dimensional coordinate of each position where the facilities and equipment to which the work object is attached contact the ground. These are respectively used as the positions of their equivalent points.

[0126] S203. Analyze the running parameters of all moving objects in the 3D scene, and combine historical image data to analyze following distance and following speed. Preset standard speed. and standard spacing The weight coefficients of each moving object are calculated. Specifically, this includes:

[0127] S2031. Obtain historical image data and extract images containing moving objects. Video clips, calculating moving objects average speed and the average speed of all moving objects. .

[0128] S2032, Moving objects The video clip is divided into different following time periods based on the condition that the vehicle is driving directly behind other moving objects and the distance between them is less than a threshold. Within each following time period, different following time periods are evenly distributed. A specific point in time.

[0129] S2033. Analyze moving objects at each time point. The instantaneous velocity and the instantaneous time distance between the vehicle and the moving object in front are used to calculate the standard deviation of the instantaneous velocity at all time points within each following period. and instantaneous time distance standard deviation .

[0130] In practice, spatiotemporal slicing technology is used to segment the video stream, and the rate of change and standard deviation of vehicle spacing are calculated based on optical flow. Reflects speed fluctuations (sudden braking / acceleration). It indicates the degree of following closely behind the vehicle.

[0131] S2034. Calculate the moving objects within each following time period. average speed and the average distance between the moving object in front. Substitute into the formula to calculate the moving object Weighting coefficients :

[0132] ;

[0133] In the formula, It is a constant greater than 1. This represents the number of times the vehicle will accompany the passenger during all scheduled times.

[0134] The system quantifies the risk level of vehicle behavior by identifying abnormally speeding vehicles by comparing the deviation of individual speeds from the overall traffic flow speed. It also combines the instantaneous speed fluctuations and spacing during the following period to reflect the vehicle's driving stability.

[0135] In practice, the dynamic weighting of historical data and preset safety benchmarks can accurately identify high-risk vehicles (such as those that brake frequently or closely follow the vehicle in front), providing key input for subsequent collision risk assessment.

[0136] S2035, and so on, extract video segments for each moving object and calculate weight coefficients.

[0137] In the specific implementation process, the system accurately identifies the working posture of personnel, establishes vehicle behavior risk indicators, and solves the problem of heterogeneous data correlation between "people and vehicles" in traditional methods.

[0138] S300: Set different detection points for equivalent points and analyze the parameters of moving objects entering the detection points. Combine the weighting coefficients to calculate the risk coefficient of the corresponding equivalent point and the risk index of the work object, mark the work object, and set safety points. Specifically, this includes:

[0139] S301. Analyze the positions of all equivalent points in the 3D scene, and set a location for each equivalent point within the road area. Each detection point has a reference velocity and reference direction determined by analyzing historical image data. Specifically, this includes:

[0140] S3011, evenly distributed within the road area Using the sampling point as the starting point, a vector is first established according to the lane direction. Then according to the equivalent point Establish a vector for position and direction Calculate vector and The included angle .

[0141] S3012. Analyze the relationship between each sampling point and the equivalent point. Location distance Therefore, priority coefficients are calculated, and sampling points are arranged in ascending order of priority coefficients. The first sampling point is selected. Each sampling point is used as an equivalent point. The detection point.

[0142] Priority coefficient The formula is:

[0143] ;

[0144] In the formula, It is a constant greater than 1. Distance weights The included angle weight.

[0145] In the specific implementation process, the priority of the deployment of detection points is determined, and the spatial relationship between road sampling points and construction equivalent points is comprehensively evaluated. The priority coefficient gives higher weight to distance factors (prioritizing points that are closer), while introducing directional angle correction (consistency between the vehicle's driving direction and the orientation of the equivalent point) to ensure that the layout of detection points not only covers potential conflict areas but also conforms to the natural trend of vehicle movement trajectories, thereby optimizing the efficiency of risk monitoring.

[0146] S3013. Extract the image frame where the detection point is triggered from the historical image data, and analyze the speed and direction of the moving object when it reaches the detection point in each image frame. Triggering refers to the image frame captured at the instant the moving object enters the detection point.

[0147] S3014. Calculate the average velocity across all image frames at the same detection point, and use it as the reference velocity for that detection point. Fit the direction vector across all image frames at the same detection point, and use it as the reference direction for that detection point.

[0148] Since all direction vectors originate from the same detection point, the fitting methods include single-step fitting and successive fitting.

[0149] A single fitting requires a reference direction vector. The average angle between the fitting and all other direction vectors is calculated to obtain the fitted reference direction.

[0150] Successive fitting involves constructing an angle between two direction vectors in consecutive time and taking the median vector, then continuing to construct an angle between the vectors and subsequent direction vectors and taking the median vector, and so on, until all iterations are completed to obtain the fitted reference direction.

[0151] S3015, and so on, sets for each equivalent point respectively. The system identifies several detection points and analyzes the reference velocity and reference direction for each point.

[0152] S302. Analyze the parameters of each moving object entering the detection point through image data analysis, set the kinetic energy transfer coefficient according to the status of the work object corresponding to the equivalent point, and calculate the risk coefficient of each equivalent point in combination with the weighting coefficient. Specifically, this includes:

[0153] S3021. Analyze the current image data and capture the entry equivalent point. The moving objects at each detection point are analyzed, including their weight coefficients, instantaneous speed, instantaneous direction, and time of entry.

[0154] The detection point uses electronic fence technology. When the vehicle's center of gravity enters a virtual circle with a radius of 1.5m, it triggers a snapshot with a millisecond-level response.

[0155] S3022 is the equivalent point. Set the kinetic energy transfer coefficient for the corresponding work object. The state is analyzed. In the stationary state, the kinetic energy transfer coefficient is 1, and in the off-ground state, the kinetic energy transfer coefficient is less than 1 and greater than 0.

[0156] The kinetic energy transfer coefficient is a core parameter characterizing the proportion of kinetic energy effectively transferred from the moving object (the offending vehicle) to the working object (the worker) during a collision.

[0157] In practice, when the object being worked on is on a rigid ground, the feet / support points form a displacement constraint boundary, and the impact energy of the vehicle is directly transmitted to the incompressible foundation through the human body, causing the energy reflectivity to approach zero.

[0158] At this point, the human body is almost a rigid coupling medium, and the kinetic energy conversion efficiency reaches its theoretical maximum value.

[0159] In the specific implementation process, when the object of the operation is located on the facility or equipment, there is a non-holonomic constraint between the operator and the equipment. The collision induces the throwing motion of the human body, and part of the kinetic energy is converted into potential energy (increased height above the ground) and rotational kinetic energy. The effective impact energy is dissipated in multiple stages.

[0160] Scaffolding components, lifting platforms, and other equipment can be considered as a composite damping system (including elastic deformation of the steel frame and friction of connecting parts), with some kinetic energy being converted into structural vibration energy and dissipated.

[0161] In the specific implementation process, when the scaffolding is in a state where the work object is off the ground, a preset kinetic energy transfer coefficient is used. =0.7 (30% of the kinetic energy is absorbed by the deformation of the member). The kinetic energy transfer coefficient is preset during the operation of the lifting platform. =0.5 (hydraulic damping dissipates 50% of kinetic energy).

[0162] The kinetic energy transfer coefficient is a bridge variable that converts the risk of vehicle impact into the degree of damage to the workpiece. Its core logic lies in:

[0163] Collision system stiffness matching: When working on the ground, the vehicle, human body and foundation form an infinitely rigid system, while when working off the ground, it is reduced to a finitely rigid system.

[0164] Differences in energy form conversion: rigid connections transmit pure kinetic energy, while flexible connections convert some of it into sound energy, heat energy, and potential energy.

[0165] S3023. Count the number of moving objects entering at each detection point. Get the current time Substitute into the formula to calculate the equivalent point risk factor :

[0166] ;

[0167] In the formula, The time decay coefficient, and The first The detection point was the first The time and instantaneous speed at which a moving object enters. The preset standard duration, It is a constant. For the first Reference velocity of each detection point.

[0168] and The first The first detection point The weighting coefficients and quality of each moving object. This represents the average weighting coefficient of all moving objects within the road area.

[0169] For the first The detection point was the first The instantaneous direction of the moving object entering is the same as that of the first moving object. The angle between the reference directions of each detection point, For the first The detection point was the first The instantaneous direction and equivalent point of a moving object entering. The angle between the position and direction.

[0170] The risk coefficient is used to dynamically calculate the real-time threat level of vehicle collisions to construction sites. The formula incorporates multi-dimensional parameters:

[0171] Time sensitivity: The impact of recently arrived vehicles is greater than that of earlier events.

[0172] Kinetic energy transfer efficiency: The collision energy conversion ratio is adjusted according to the working height (high-altitude operations reduce injury due to equipment buffering).

[0173] Behavioral hazards: speeding, deviating from the intended direction, or vehicles with high weighting coefficients significantly amplify the risk.

[0174] Spatial correlation: The smaller the angle between the vehicle's driving direction and the equivalent point's orientation, the higher the probability of a collision.

[0175] By normalizing the data to offset the interference of environmental variables, continuous risk values ​​are output to drive early warning decisions.

[0176] Similarly, calculate the risk coefficient for each equivalent point.

[0177] S303. Calculate the risk index of the work object based on the risk coefficient, mark work objects with a risk index higher than the threshold, and set safety points around the marked work objects. Specifically, this includes:

[0178] S3031. The average risk coefficient of all equivalent points under the work object is used as the risk index, and a risk index greater than the threshold is marked. The task object.

[0179] S3032, Set Distance Using the position of each equivalent point under the marked work object as the center, the distance is... Divide the 3D scene into circular regions with a radius.

[0180] S3033. Set a two-dimensional coordinate axis at the center of each circular area, and take the four points where the boundary of the circular area intersects with the coordinate axis as planning points, and calculate the risk coefficient of each planning point.

[0181] S3034. Determine whether the risk coefficient of all planning points is greater than the threshold. If the result is negative, then the planning point with the lowest risk coefficient is taken as the safe point.

[0182] S3035 If the result is yes, then divide the area into circular regions with the position of each planning point as the center, and continue to set planning points until the conditions in step S3034 are met. Then stop setting planning points and take the planning point with the lowest risk coefficient as the safe point.

[0183] In practice, the algorithm intelligently plans safe haven locations for high-risk workers. It delineates a circular safety zone with an equivalent point as the center and a fixed distance as the radius, prioritizing the location with the lowest risk coefficient among the boundary intersection points.

[0184] If all locations exceed the threshold, the search range is expanded incrementally, iterating until a location meeting the safety criteria is found. This rule ensures the shortest possible escape path and consistently avoids real-time risk hotspots.

[0185] Predicting collision probabilities in the spatiotemporal dimensions and automatically planning the optimal avoidance path enables "risk-driven" proactive protection.

[0186] The S400 system automatically issues an audible and visual warning when marking the work object's head and prompts the user to move to a safe point for safety.

[0187] In practice, the safety helmet has a built-in LoRa module that receives warning commands and controls the three-color LED to flash (alternating between red and yellow) and the buzzer frequency to change (80-100dB) via PWM. The visibility distance is greater than 50m in strong light.

[0188] Risk assessment results drive precise individual early warnings, shortening emergency response delays.

[0189] Example 2: Please refer to Figure 2The present invention also provides a traffic engineering construction management system based on intelligent perception, including an environmental perception module, an intelligent analysis module, a risk assessment module, and an early warning module.

[0190] The environmental perception module is used to collect GIS maps and environmental data, the orientation parameters of objects working in the construction area, the running parameters of moving objects in the road area, and to construct a three-dimensional scene.

[0191] Collect GIS map data, environmental data (such as image data, BIM 3D model data and multi-source scan data), and acquire in real time the orientation parameters of the work object (such as the spatial position captured by the helmet sensor) and the running parameters of the moving object (such as the vehicle mass collected by the piezoelectric film sensor, and the position, speed and direction analyzed by real-time image analysis).

[0192] In the specific implementation process, digital twin technology is used to construct a three-dimensional scene and divide the construction area (the specific range of engineering operations) and the road area (the boundary distance is less than the threshold).

[0193] It provides accurate real-time environmental mapping and visualization, supports subsequent risk analysis, enhances the perception capability of the construction area, reduces human monitoring errors, and improves the overall safety foundation.

[0194] The intelligent analysis module analyzes the state of the work object based on orientation parameters and sets equivalent points in the 3D scene. It then combines environmental data and operational parameters to calculate the weight coefficients of the moving object.

[0195] Analyze the status of the work object (e.g., whether it is stationed or off-site) and set equivalent points (calculated based on location coordinates). Combine historical image data to calculate the weight coefficients of moving objects, including analyzing parameters such as the standard deviation of speed and the standard deviation of distance during the following period, and use formulas to quantify vehicle risk factors.

[0196] In the specific implementation process, intelligent algorithms are used to identify potential risk sources, improve the accuracy of vehicle instability assessment, assist in early warning decision-making, and reduce the misjudgment rate of dynamic traffic interference with construction.

[0197] The risk assessment module sets detection points for the equivalent point and analyzes the parameters of the moving object that has entered. It calculates the risk coefficient of the equivalent point and the risk index of the work object by combining the weighting coefficient, marks the work object and sets the safety point.

[0198] For each equivalent point, a detection point is set (the priority coefficient is calculated based on the distance and angle between the sampling points). The parameters of the moving object entering the detection point (such as instantaneous speed, direction and weight coefficient) are analyzed. The risk coefficient of the equivalent point and the risk index of the working object are calculated in combination with the status of the working object.

[0199] In the specific implementation process, after marking high-risk operation objects, safety points are dynamically generated by iteratively setting planning points in the three-dimensional scene (such as dividing a circular area with a distance L as the radius, and the point with the minimum risk coefficient is the safety point).

[0200] Dynamically quantify on-site risk levels, provide risk avoidance guidance, achieve proactive protection, shorten emergency response time, and minimize the probability of collision accidents.

[0201] The early warning module emits an audible and visual warning by having a safety helmet on, and the voice prompts the marked work object to immediately move to a safe point to avoid danger.

[0202] In practice, when the risk assessment module marks a high-risk work object, it automatically triggers an audible and visual warning function for the work object wearing a safety helmet, prompting personnel to move to a preset safe point. The warning command is updated in real time based on risk coefficient calculation.

[0203] It provides immediate, visual and audible warning signals to guide workers to avoid risks efficiently, enhances personnel safety compliance, and thus directly intervenes in potential accidents, reducing construction injury rates and downtime losses.

[0204] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0205] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent perception-based traffic engineering construction management, characterized in that: The method comprises: S100, collecting GIS map and environmental data, and orientation parameters of work objects in the construction area and running parameters of mobile objects in the road area, thereby constructing a three-dimensional scene; S200, analyzing the state of the work object according to the orientation parameter, and setting an equivalent point in the three-dimensional scene; combining the environmental data and the running parameter to analyze and calculate the weight coefficient of the mobile object; S300, setting different detection points for the equivalent point and analyzing the parameters of the mobile object driving into the detection point, combining the weight coefficient to calculate the risk coefficient of the corresponding equivalent point and the risk index of the work object, marking the work object and setting a safety point; S400, automatically issuing an audible and visual warning for the safety helmet of the work object, and prompting the mobile object to move to the safety point for risk avoidance. 2.The smart sensing based traffic engineering construction management method of claim 1, wherein: In S100, the environmental data includes image data, BIM data and scanning data; the image data refers to the monitoring video of the construction area and the road area; The BIM data includes three-dimensional models of various facilities and equipment used for engineering work in the construction area; the scanning data refers to the three-dimensional data of the physical space in the construction area collected by multi-source sensing devices; The construction area refers to a specific range pre-defined to ensure the safety of engineering work during the implementation of the traffic engineering; the road area refers to the coverage of the part of the road with an average distance from the boundary of the construction area less than a threshold value; The work object refers to the worker working in the construction area; the orientation parameter refers to the spatial position of the human body, which is collected by the sensor on the safety helmet worn by the worker; The mobile object refers to the vehicle driving in the road area; the running parameter includes the mass, position, speed and direction of the vehicle, the mass is collected by the piezoelectric film sensor, and the position, speed and direction are obtained by analyzing the real-time image data; A three-dimensional scene is constructed according to the GIS map, scanning data and BIM data by using digital twinning technology, and the construction area and the road area are divided at the corresponding positions in the three-dimensional scene; The relative position of the work object in the construction area is analyzed by the orientation parameter and the image data, and is mapped in real time; the relative position of the mobile object in the road area is analyzed according to the image data, and is mapped in real time. 3.The smart sensing based traffic engineering construction management method of claim 2, wherein: S200 includes: S201、analyze the three-dimensional coordinates corresponding to the spatial positions of all work objects in the three-dimensional scene , the work objects with values greater than the threshold are set to off-ground state, and other work objects are set to on-ground state S202, Transfer the two-dimensional coordinates of the work objects in the site status. The equivalent point is the two-dimensional coordinate of each position where the facilities and equipment to which the work object is attached contact the ground. These are respectively used as the positions of their equivalent points; S203, analyze the running parameters of all moving objects in the three-dimensional scene, analyze the following vehicle distance and following vehicle speed in combination with historical image data; preset standard speed and standard distance , and calculate the weight coefficient of each moving object. 4.The smart sensing based traffic engineering construction management method of claim 3, wherein: S203 includes: S2031、acquire historical image data, intercept a video clip containing a moving object S2032、calculate the average speed of the moving object S2033、calculate the average speed of all moving objects ;​ S2032, with the moving object The different following periods are divided in the video clip under the condition that the other moving object is right behind the vehicle and the front-rear distance is less than a threshold value, and a time point is uniformly set in each following period. ​ S2033、analyze the instantaneous speed of the moving object at each time point, and the instantaneous time distance between the moving object and the front moving object respectively calculate the instantaneous speed standard deviation and the instantaneous time distance standard deviation of all time points in each following period and the instantaneous time distance ; S2034. Calculate the moving objects within each following time period. average speed and the average distance between the moving object in front. Substitute into the formula to calculate the moving object Weighting coefficients : ； wherein is a constant greater than 1, is the number of all following periods; S2035, and so on, respectively, the video clips are intercepted for each mobile object, and the weight coefficient is calculated. 5.The smart sensing based traffic engineering construction management method of claim 3, wherein: S300 includes: S301、analyze the positions of all equivalent points in the three-dimensional scene, set a probe point for each equivalent point in the road area, and analyze the reference speed and reference direction of each probe point according to historical image data; S302, the parameters of each mobile object driving into the detection point are analyzed by the image data, the kinetic energy transfer coefficient is set according to the state of the work object corresponding to the equivalent point, and the risk coefficient of each equivalent point is calculated by combining the weight coefficient; S303, the risk index of the work object is calculated according to the risk coefficient, the work object with a risk index higher than a threshold value is marked, and a safety point is set around the marked work object. 6.The intelligent perception based traffic engineering construction management method of claim 5, wherein: S301 includes: S3011、In the road area, uniformly set a sampling point, taking the sampling point position as a starting point, first establishing a vector according to the lane direction , then establishing a vector according to the position direction of the equivalent point , , calculating the included angle between the vector and ;​ S3012、analysis of each sampling point and the equivalent point The position distance Thus, the priority coefficient is calculated, the sampling points are arranged in order from small to large priority coefficient, and the first Sampling points are selected as the detection points of the equivalent point ​ S3013, the image frames in which the detection point is triggered are intercepted in the historical image data, and the speed and direction of the mobile object driving to the detection point in each image frame are analyzed; S3014, the average speed of all image frames under the same detection point is calculated as the reference speed of the corresponding detection point; the direction vectors of all image frames under the same detection point are fitted as the reference direction of the corresponding detection point; S3015、Set up the reference speed and the reference direction of each probe point respectively by analogy probe points, and analyze to obtain the reference speed and the reference direction of each probe point. 7.The smart sensing based traffic engineering construction management method of claim 6, wherein: In S3012, the priority coefficient The formula is: ； wherein is a constant greater than 1, is a distance weight, is an angle weight. 8.The smart sensing based traffic engineering construction management method of claim 6, wherein: S302 includes: S3021、analyze the current image data, and capture the entry equivalent point The moving object of each detection point, the weight coefficient of the moving object, and the instantaneous speed, instantaneous direction, and time of entry are analyzed. S3022, equivalent point Corresponding to the work object set dynamic energy transfer coefficient And analysis state, the value of dynamic energy transfer coefficient is 1 in the state of residence, and the value of dynamic energy transfer coefficient is less than 1 and greater than 0 in the state of leaving the ground. S3023, count the number of moving objects entering each probe point , obtain the current time , substitute the formula to calculate the equivalent point risk coefficient : ； In the formula, is a time decay coefficient, and is the time and instantaneous speed when the th detection point is entered by the th moving object, respectively; is a preset standard time length, is a constant; is the reference speed of the th detection point; and The first The first detection point The weighting coefficients and quality of each moving object. This is the average weighting coefficient of all moving objects within the road area; For the first The detection point was the first The instantaneous direction of the moving object entering is the same as that of the first moving object. The angle between the reference directions of each detection point, For the first The detection point was the first The instantaneous direction and equivalent point of a moving object entering. The angle between the position and direction; Similarly, the risk coefficient of each equivalent point is calculated. 9.The smart sensing based traffic engineering construction management method of claim 8, wherein: S303 comprises: S3031、Calculate the average value of the risk coefficients of all equivalent points under the job object as a risk index, and mark the risk index greater than the threshold value the job object; S3032, set distance The position of each equivalent point under the marked job object is taken as the center of a circle, and the distance is taken as the radius to divide a circular area in the three-dimensional scene. S3033, the position of each circular region center is set as the coordinate axis of two dimensions, the four point positions where the boundary of the circular region intersects with the coordinate axis are set as planning points, and the risk coefficient of each planning point is calculated respectively; S3034、judge whether the risk coefficient of all planning points is greater than the threshold value , if the result is no, the planning point with the minimum risk coefficient is taken as the safety point; S3035, if the result is yes, the circular region is divided with the position of each planning point as the center, the planning point is continuously set until the condition in S3034 is met, then the setting of the planning point is stopped, and the planning point with the minimum risk coefficient is set as the safety point.

10. A smart sensing based traffic engineering construction management system characterized in that: The system comprises an environment perception module, an intelligent analysis module, a risk assessment module and a warning prompt module; The environment perception module is used for collecting GIS maps and environmental data, orientation parameters of work objects in a construction area, and operation parameters of mobile objects in a road area, and constructing a three-dimensional scene; The intelligent analysis module analyzes the state of the work object according to the orientation parameters, and sets equivalent points in the three-dimensional scene; the weight coefficient of the mobile object is analyzed and calculated in combination with the environmental data and the operation parameters; The risk assessment module sets a detection point for the equivalent point and analyzes the parameters of the mobile object entering, calculates the risk coefficient of the equivalent point and the risk index of the work object in combination with the weight coefficient, marks the work object and sets a safety point; The warning prompt module issues an audible and visual warning through a safety helmet, and a voice prompt marks the work object to immediately move to the safety point for risk avoidance.

Images