Construction engineering potential safety hazard detection method and system

Abstract

The invention relates to the technical field of risk assessment in automatic safety detection, and discloses a construction engineering potential safety hazard detection method and system, and the method comprises the steps: selectively defining a detection blind area or point-by-point path planning according to the definition of data sensed by a sensor, and determining an undetected key check point set which cannot be reached; the method comprises the following steps: selecting a physical blocking state, further accumulating a basic risk weight of the physical blocking state, dynamically generating a quantitative total risk score in combination with peripheral high-risk operation types and risk exposure duration, generating a visual risk report, and pushing the visual risk report to a management terminal. The physical blocking state is coupled with the importance of structural key points and the dynamic influence of external high-risk operation, so that an invisible detection blind area is converted into a quantifiable decision basis capable of reflecting the process that risks are accumulated along with time, and the crossing from physical perception to risk semantics is realized.

Description

Technical Field

[0001] This invention relates to a method and system for detecting safety hazards in building engineering, belonging to the field of risk assessment technology in automated safety detection. Background Technology

[0002] In dynamic and complex environments such as construction projects, automated inspection equipment equipped with environmental perception sensors such as lidar is commonly used to improve the coverage and timeliness of safety supervision. Through autonomous path planning and obstacle avoidance functions, it conducts periodic inspections of preset key points. This approach has become a common technical approach to improve the level of intelligent on-site management.

[0003] However, when such automated equipment is deployed from relatively stable industrial environments to construction sites characterized by temporality and uncertainty, a contradiction arises: the high fidelity of the equipment's environmental perception versus the diminishing value of its information output in decision-making applications. On construction sites, obstacles such as temporarily laid power cables, haphazardly piled construction materials, and fixed formwork support poles coexist. Automated inspection equipment identifies and avoids all these obstacles, resulting in a large number of path-obstruction alarm events. The vast majority of these events originate from brief, harmless temporary operations, and their frequent occurrence continuously dilutes the attention of safety management personnel. The core value of alarm information—its role as a signal warning of potential danger—is thus undermined. Severely weakened by the continuous decrease in signal-to-noise ratio, managers gradually treat such alarms as routine background noise and empirically filter or ignore them in order to maintain work efficiency. This is a hidden cost that the industry generally accepts in exchange for automation coverage. The most direct approach to solving this problem is to optimize the alarm filtering algorithm, such as setting time thresholds or simple area masking. However, this approach does not address the essence of the problem. It cannot effectively distinguish between a temporary, harmless blockage and the initial stage of a long-term detection blind spot that may hide high-risk hazards, formed by temporary and permanent obstacles. Incorrect filtering may lead to the silence of the evolution of key risks, allowing risks to accumulate continuously in an invisible state.

[0004] Specifically, existing technologies have inherent limitations in the information processing chain: 1. The path obstruction information output by the system is essentially a binary physical state description, which does not carry any semantic information about the consequences and cannot reveal the potential risk level corresponding to the inspection interruption; 2. When an alarm is generated, it is unknown whether there are real hidden dangers within an area that has become a blind spot. Its risk value depends entirely on the importance of the undetected key points in the overall structure, and this core information is completely missing in existing alarm technologies; 3. Due to the lack of risk quantification in alarm information, managers cannot make effective hazard level judgments and intervention priority rankings based on the data provided by the system. Their decision-making behavior is forced to decouple from the information provided by the automated system, reverting to the traditional mode of relying on personal experience or confirmation through inefficient remote communication. Therefore, how to establish a new technical approach that, when automated inspection equipment encounters physical obstacles, can make its output information go beyond a simple physical state report, giving it accurate risk semantics, and transforming an invisible detection blind spot into a risk level assessment result that can be directly used for decision-making, is the technical problem that this invention aims to solve. Summary of the Invention

[0005] This invention provides a method and system for detecting safety hazards in building engineering. Its main purpose is to solve the problem that existing technologies cannot quantify and assess the potential risks of blind spots in automated inspections due to physical obstacles, making it difficult for managers to make effective decisions and potentially causing them to overlook high-risk hazards.

[0006] To achieve the above objectives, the present invention provides a method for detecting safety hazards in building engineering, the method comprising: Perform a step to identify a set of unreachable, uninspected critical checkpoints. This step includes selecting one of two execution paths based on the clarity of sensor-perceived data from the inspection equipment: Path 1: Under the condition that the sensor perception data is clear, use the sensor to scan the environment around the physical obstacle to generate environmental data. Combine the positioning information of the inspection equipment to calculate and determine the geometric range of the detection blind zone, and determine the key inspection points located within the geometric range of the detection blind zone in spatial coordinates as the set of uninspected key inspection points. Path Two: Under the condition that sensor data is blurred due to environmental interference, obtain a list of all critical checkpoints within the current task range, and for each critical checkpoint in the list, independently attempt to plan a safe passage path from the current location of the inspection equipment to that critical checkpoint. All critical checkpoints for which a safe passage path cannot be planned are identified as the set of uninspected critical checkpoints. Based on the identified set of uninspected critical checkpoints and the basic risk weights of each uninspected critical checkpoint obtained from the risk value table, accumulate to obtain a basic total risk score. Obtain construction operation information around the spatial location of the detection blind zone, track and accumulate the risk exposure time of the detection blind zone under the influence of high-risk operations. Based on the basic total risk score, and combined with the type of high-risk operation and the corresponding accumulated risk exposure time, generate a total risk score. Generate a visualized risk report containing the total risk score and push the visualized risk report to the management terminal.

[0007] Preferably, the step of generating the total risk score based on the basic total risk score, combined with the type of high-risk operation and the corresponding cumulative risk exposure duration, is as follows: for each high-risk operation that affects the detection blind zone, calculate the cumulative risk increment based on the risk impact factor corresponding to its operation type and its corresponding cumulative risk exposure duration; add all the calculated cumulative risk increments to the basic total risk score to obtain the updated total risk score.

[0008] Preferably, the calculation of the risk accumulation increment is based on a risk accumulation function; this risk accumulation function takes the accumulated risk exposure duration as an input variable and outputs a cumulative effect value that characterizes the nonlinear growth of risk over time; the cumulative effect value is multiplied by the risk impact factor to obtain the risk accumulation increment.

[0009] Preferably, when the detection blind zone is determined to be within the influence area of ​​multiple high-risk operations, before the step of adding all the calculated risk accumulation increments to the basic total risk score, the method further includes: querying the operation type combination rule base to analyze whether there is an interaction relationship between multiple high-risk operation types; if there is an interaction relationship, then adjusting each risk accumulation increment according to the interaction relationship before accumulation.

[0010] Preferably, if there is no interaction relationship in the job type combination rule base, the method further includes: at the edge position of the detection blind zone, using the sensor mounted on the inspection equipment to monitor the change of at least one physical parameter; simultaneously acquiring the operating characteristics of multiple high-risk jobs; identifying and quantifying the actual interaction relationship between multiple high-risk job types by analyzing the time correlation between the monitored physical parameter changes and the operating characteristics of external jobs, and using it to update the job type combination rule base.

[0011] Preferably, the monitored physical parameters include the amplitude and frequency of the micro-vibration signal, and the operating characteristics include the start-up and shutdown time and operating power of the vibration-type equipment. The steps for analyzing the time correlation between the monitored physical parameter changes and the external operating characteristics include: performing time-frequency analysis on the micro-vibration signal to extract the energy components associated with the operating frequencies of each high-risk operation; when multiple high-risk operations are running simultaneously, determining whether there is a nonlinear superposition phenomenon between their associated energy components; if there is a nonlinear superposition phenomenon, determining the interaction relationship as synergistic enhancement based on the degree of nonlinear superposition, and quantifying its enhancement coefficient.

[0012] Preferably, in path one, the sensor is a lidar, the environmental data is three-dimensional point cloud data, and the positioning information is obtained through inertial measurement unit and real-time positioning and mapping technology; high-risk operations include heavy equipment vibration operations, deep foundation pit excavation operations, and large-volume concrete pouring operations; the impact area of ​​high-risk operations is a circular area with the center of the high-risk operation as the center and the radius of the impact radius determined according to industry safety standards as the radius.

[0013] Preferably, the steps for generating a visual risk report include: highlighting the area corresponding to the set of uninspected critical checkpoints on a two-dimensional site plan using a specified color; marking the location of each uninspected critical checkpoint using a graphic symbol corresponding to the risk level within the highlighted area; and displaying the generated total risk score in numerical form at a specified location on the two-dimensional plan, along with the main high-risk operation types that contribute to the total risk score and their corresponding cumulative risk exposure duration.

[0014] Preferably, in path two, the step of attempting to plan a safe passage route involves performing A* pathfinding algorithm or fast expanding random tree algorithm calculations on a temporary local environment map constructed based on fuzzy perception data.

[0015] A safety hazard detection system for building construction projects, the system comprising: A checkpoint determination module, based on the clarity of sensor data from the inspection equipment, selects one of the following two modes to determine the set of uninspected critical checkpoints that cannot be reached: First mode: In this mode, the checkpoint determination module uses sensors to scan the surrounding environment of physical obstacles to generate environmental data, combines this data with the positioning information of the inspection equipment to calculate and determine the geometric range of the detection blind zone, and identifies critical checkpoints located within the geometric range of the detection blind zone in spatial coordinates as the set of uninspected critical checkpoints; Second mode: In this mode, the checkpoint determination module obtains a list of all critical checkpoints within the current task range, and for each critical checkpoint in the list, independently attempts to plan a safe passage path from the current location of the inspection equipment to that critical checkpoint, and identifies all critical checkpoints for which a safe passage path cannot be planned as the set of uninspected critical checkpoints. A risk assessment module is used to: obtain a basic total risk score by summing the basic risk weights of each uninspected key checkpoint obtained from the set of uninspected key checkpoints determined by the checkpoint determination module and querying the risk value table; acquire construction operation information around the spatial location of the detection blind zone, track and accumulate the risk exposure time of the detection blind zone under the influence of high-risk operations; and generate a total risk score based on the basic total risk score, combined with the type of high-risk operation and the corresponding accumulated risk exposure time. A report generation and push module is used to generate a visual risk report containing the total risk score generated by the risk assessment module, and push the visual risk report to the management terminal.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. When automated inspection equipment detects physical obstruction in the field environment, this method does not interrupt the task or merely report the path status. Instead, it spatially correlates the dynamically generated physical spatial information of the obstructed area with the structural key point information in the preset building information model in real time. Then, each uninspected key point falling into the area is converted into a quantified risk weight value based on its importance in the engineering structure and accumulated. As a result, the information output by the system changes from a simple state description of physical accessibility to structured data that directly reveals the level of potential safety risks. This allows managers to naturally shift their focus from whether the path is unobstructed to the extent of the risk caused by the lack of information when faced with inspection interruption events, avoiding empirical neglect of high-risk blind spots due to information overload.

[0017] 2. When there are diffuse media such as dust or water vapor at the construction site that interfere with optical sensors, this method avoids the difficulty of accurately reconstructing the geometric boundaries of fuzzy environmental information. It changes the logical basis of risk assessment from the overall definition of a continuous and potentially distorted blind zone range to the attempt to plan the accessibility path for a series of discrete, definite key checkpoints one by one. This approach builds the accumulation of risk values ​​on the deterministic judgment of whether each independent target point is reachable. Even if the overall quality of the sensing data deteriorates, the system can still form a reliable risk assessment result that is not affected by the accuracy of the blind zone boundary description, thus maintaining the stability of the basis for safety decisions.

[0018] 3. This method further considers that the risks of key points within the detection blind zone are not static, but may be dynamically accumulated due to the indirect impact of ongoing construction work outside the blind zone. By linking the spatial information of the blind zone generated by the inspection equipment with the spatiotemporal information of surrounding high-risk operations obtained from the site management platform, and introducing the tracking of risk exposure time, the calculation of risk score is no longer just a simple summation of the static weights of uninspected key points, but a dynamic value that reflects the accumulation process of potential damage caused by external stress over time. This transforms risk assessment from a snapshot-style judgment of the current spatial obscuration state to a process insight into the development of potential hazards over time, providing objective data support for managers to intervene in areas that seem temporarily harmless but have continuously increasing long-term risks. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for detecting safety hazards in building construction according to the present invention. Figure 2 The figure shows the results of a comparative experiment to verify the effectiveness of the dynamic risk score calculation method of the present invention. Figure 3 This is a structural schematic diagram of a building engineering safety hazard detection system according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] The present invention provides a method and system for detecting safety hazards in building engineering, which is configured as a risk perception and decision support process. The process begins by using sensors mounted on inspection equipment to collect data on the physical environment. Then, through a dual-path adaptive checkpoint accessibility analysis stage, it identifies the detection coverage gaps caused by physical obstruction. Subsequently, in a risk quantification model that integrates the static importance of the structure and the dynamic impact of external operations, a risk score that evolves over time is generated. Finally, the quantification result is transformed into a visual report to assist managers in making intervention decisions.

[0022] In dynamic environments such as construction sites, the performance of environmental perception sensors in automated inspection equipment can be affected by temporary obstacles and diffused media such as dust or water vapor, potentially leading to deviations in the assessment of the completeness of inspection coverage. To address this challenge, this invention includes a step for determining the set of uninspected key inspection points based on the clarity of sensor-sensed data. When the signal-to-noise ratio of the point cloud data returned by a sensor, such as a solid-state lidar, is higher than a preset threshold... For example, a signal-to-noise ratio greater than Under the condition that the sensor-perceived data is clear, the system executes path one. It utilizes LiDAR in conjunction with an inertial measurement unit and real-time localization and mapping (RTD) technology to scan the environment surrounding physical obstacles, generating 3D point cloud data with spatial coordinates. Combined with the location information of the inspection equipment on the global map, it uses convex hull or Alpha-shapes algorithms to delineate the geometric range of the detection blind zone caused by obstacle occlusion. This range is defined as a polygonal region in 3D space. Subsequently, the system identifies and includes all key checkpoints in the pre-loaded building information model that intersect with this polygonal region in spatial coordinates into a set, which is determined as the set of undetected key checkpoints. However, when the sensor-perceived data is blurred due to high-concentration dust scattering, i.e., the signal-to-noise ratio of the point cloud data is below a threshold... Under these conditions, execution path two is switched. In this case, the system does not attempt to reconstruct the geometric boundaries of the fuzzy perception data. Instead, it retrieves a list of all critical checkpoints within the current task area from the Building Information Modeling (BIM) database. For each critical checkpoint in the list, it independently attempts to plan a safe passage path from the current location of the inspection equipment to that checkpoint. This path planning is performed using the A* pathfinding algorithm or the Fast Expanded Random Tree algorithm on a temporary local environment map constructed based on the fuzzy perception data. If a path cannot be found within the preset maximum number of iterations that does not overlap with any other path with a density higher than a threshold... If a path collides with a point cloud cluster, it is determined that the path cannot be planned. All critical checkpoints for which a safe passage path cannot be planned are ultimately identified as the set of unchecked critical checkpoints. This dual-path adaptive approach enables the system to transform from defining a continuous geometric range to making a deterministic judgment on the reachability of a series of discrete target points based on data quality, providing a definite input for subsequent risk quantification.

[0023] Simply identifying unchecked critical checkpoints is insufficient to provide a basis for risk level ranking, as the potential failure consequences of checkpoints at different structural locations vary. Therefore, this invention introduces a procedure for quantifying basic risk based on a risk value table. This risk value table is a pre-defined table that associates building component types defined in the Building Information Model with a dimensionless basic risk weight. The data structure used for mapping, for example, the load-bearing column base components are defined. The value is 10, while in ordinary floor slab areas... The value is 3; once the system determines the set of unchecked critical checkpoints, it queries this value table and assigns the basic risk weights corresponding to each unchecked critical checkpoint in the set. The scores are accumulated to obtain the basic total risk score. Its calculation can be expressed as ,in, This represents the total number of unchecked critical checkpoints in the set; for example, if a blind spot contains a load-bearing column base with a weight of 10 and two ordinary floor slab areas with a weight of 3, then its total foundation risk score is... identified as Through this step, the information output by the system changes from a binary state description of physical accessibility to quantitative data that reveals the level of potential security risks.

[0024] The risks at construction sites are not static. Potential risks within detection blind spots accumulate dynamically due to ongoing high-risk operations that can transmit physical stress. To incorporate this cumulative effect into the risk assessment model, this invention includes a dynamic risk increment calculation module. This module first acquires construction operation information surrounding the spatial location of the detection blind spot. For example, it obtains the type, location, and start / stop times of operations such as heavy equipment vibration operations or deep foundation pit excavation through the application programming interface of the construction site management platform. Based on industry safety standards, it determines an influence radius for each type of operation and then determines whether the detection blind spot is located within this influence area. If so, the system starts a timer to track and accumulate the risk exposure time of the detection blind spot to this high-risk operation. And based on the accumulated duration Generate a total risk score based on the type of task. The specific generation steps are as follows: for each high-risk operation that affects the detection blind zone... Based on the risk impact factors corresponding to its operation type and the corresponding cumulative risk exposure duration Calculate a cumulative risk increment This calculation is based on a risk accumulation function. This function will accumulate the duration of risk exposure. As an input variable, the output is a cumulative effect value representing the non-linear growth of risk over time; for example, this function can be set as a logarithmic function. ,in, A timescale parameter (e.g., 60 minutes) is used to adjust the rate of risk growth; correspondingly, the cumulative risk increment is... The calculation shows that, ultimately, the updated total risk score is obtained by summing all the calculated cumulative risk increments with the basic total risk score. This design enables risk assessment to reflect the development of potential hazards over time.

[0025] In situations where the detection blind zone is simultaneously located within the influence area of ​​multiple high-risk operations, the interaction between these operations may not be linearly additive. To address this issue, before accumulating all calculated risk increments, the system first queries a rule base for operation type combinations to analyze whether predefined interaction relationships exist between multiple high-risk operation types. If a corresponding rule exists in the rule base, for example, a rule defining a synergistic enhancement effect when vibration operations and large-volume concrete pouring operations occur simultaneously, then the risk increments are adjusted according to this rule. For instance, the risk increment for vibration operations might be adjusted accordingly. The system multiplies the data by an enhancement factor greater than 1 before accumulating it. In cases where no corresponding interaction exists in the rule base, the system is also configured to update rules through proactive sensing. At the edge of the detection blind zone, it uses sensors on the inspection equipment, such as high-precision microelectromechanical system accelerometers, to monitor changes in physical parameters such as the amplitude and frequency of micro-vibration signals, and simultaneously acquires the operational characteristics of multiple high-risk operations, such as the start-up and shutdown times and operating power of vibration-type equipment. Through time-frequency analysis of the micro-vibration signals, such as short-time Fourier transform, it extracts the energy components associated with the operating frequencies of each high-risk operation. When multiple high-risk operations run simultaneously, the system determines whether there is a relationship between their associated energy components. In cases of nonlinear superposition, if the total energy after superposition exceeds the sum of the energies of the individual components, the interaction relationship is determined to be synergistic enhancement based on the degree of nonlinear superposition, and its enhancement coefficient is quantified to update the work type combination rule base. Finally, to effectively transmit the quantified risk results generated in the preceding steps to management personnel, the system executes a step of generating and pushing a visual risk report. This step highlights the area corresponding to the set of unchecked critical checkpoints on the site's two-dimensional plan view using a specified color, and within this highlighted area, marks the location of each unchecked critical checkpoint using a graphic icon corresponding to the risk level. Simultaneously, the generated total risk score is displayed numerically at a designated location on the two-dimensional plan view. It also displays the main high-risk job types that contribute to the total risk score and their corresponding cumulative risk exposure duration. The visualized risk report is packaged into a standard data format and pushed to the designated management personnel's terminal via a wireless communication network, thus providing managers with objective data basis for prioritizing interventions.

[0026] Example 1: The technical solution disclosed in this embodiment of the invention operates as follows in a specific industrial application scenario: In a large bridge construction project, automated inspection equipment is responsible for periodically checking the structural integrity of the main structure of the bridge pier during concrete curing. The base of the load-bearing column P-01 at the bottom of pier No. 3 is marked as a primary critical inspection point by the building information model, and its corresponding basic risk weight... The setting is 10; during a routine inspection, due to the temporary cable laying by the on-site construction team, the cable, together with the fixed steel scaffolding, formed a physical obstruction area at the bottom of pier No. 3, preventing the inspection equipment from reaching checkpoint P-01; after the path planning failed, the inspection equipment immediately activated the detection blind zone geometric range definition program based on LiDAR point cloud data, identified a set of uninspected critical checkpoints including checkpoint P-01, and calculated the basic total risk score of the blind zone according to the risk value table. The risk assessment for this blind spot did not stop there, but instead entered the external operation impact correlation analysis stage. The system, through querying the data interface of the construction site management platform, identified that a large hydraulic vibratory hammer was conducting sheet pile driving operations in the river channel adjacent to Pier 3. This operation was predefined as a high-risk operation type, and its risk impact factor was 10. The value is set to 5, and its radius of influence, as calculated, covers the area where pier No. 3 is located.

[0027] At this point, the risk assessment results based on the static spatial shielding state become the direct input for the time-cumulative risk assessment of the external dynamic stress source; the undetected critical checkpoint P-01 identified by the former provides a clear target for the latter's risk accumulation calculation, while the latter gives the former's risk assessment results a time evolution dimension; the system then starts a risk exposure timer for the detection blind zone containing point P-01, and begins to accumulate the duration of its exposure to the effects of vibration operations. As the pile driving operation continues, When 120 minutes have elapsed, the system uses the risk accumulation function. Among them, the time scale parameter Set to 60 minutes, calculate the cumulative risk increment. The system then adds this increment to the base total risk score to obtain the updated total risk score. .

[0028] The system then generated a visual risk report and pushed it to the management personnel's terminal. This report not only highlighted the blind spot at the bottom of pier No. 3 and marked the location of the uninspected critical checkpoint P-01, but also displayed the dynamically changing total risk score of 15.49 in numerical form, explicitly stating that the increased risk was attributed to the external vibration operation that had been ongoing for two hours. The information provided by this report transformed the management personnel's decision-making basis from a known point that was temporarily obscured to a critical load-bearing structure that was not being monitored under continuous vibration stress, thus prompting them to immediately assign a structural engineer to conduct a manual inspection, rather than treating it as a routine event to be postponed. This approach did not directly solve the physical obstruction, but rather changed the definition of the risk nature of the blind spot by associating and quantifying an invisible blind spot with an external dynamic risk source at the data level, transforming it from a low-priority accessibility issue into a high-priority structural safety hazard. The information received by the management terminal was no longer a binary status report about whether the physical path of the equipment was unobstructed, but a quantitative decision-making basis reflecting the process of potential structural damage accumulating over time under external stress.

[0029] Example 2: To objectively verify the effectiveness of the dynamic risk score calculation method in the technical solution of this invention, the following comparative experiment was designed and executed. The purpose of the experiment was to quantitatively verify the performance difference between the method of this invention and a method with only static risk assessment capabilities in distinguishing between potential high-risk detection blind zones caused by external dynamic stress and conventional static detection blind zones. The experiment was conducted on a 10-meter by 10-meter indoor test platform. The platform was equipped with a standardized template support rod matrix on the ground and a programmable industrial-grade vibrator to simulate the continuous low-frequency vibration generated by the vibration operation of heavy equipment. Its vibration frequency and amplitude parameters were set to be consistent with those of a... The ground vibration characteristics generated by a 30 kW vibratory compactor at a distance of 5 meters were matched. Two independent test groups were set up: a control group that only calculated the total foundation risk score, and an experimental group that implemented the complete technical solution of this invention. Two test scenarios were set up: Scenario A was a static obstruction scenario, in which a detection blind zone was formed only by a combination of temporary cables and support rods; Scenario B was a dynamic stress composite scenario, in which, based on Scenario A, an industrial-grade vibrator was activated next to the blind zone to apply continuous external vibration stress. The detection blind zones formed in both scenarios were completely consistent in geometric range and contained undetected key checkpoints, both including a basic risk weight. The main beam welding points have a weight of 8, and a general floor slab area has a weight of 3. Therefore, its total foundation risk score is... Both were set to 11; the total trial duration was set at 240 minutes. The risk scores of the experimental and control groups were recorded simultaneously, with a recording period of 30 minutes. This period was set to ensure that the risk accumulation function could be captured. The result is a technical trade-off between the non-linear growth trend and the processing load of the data recording system.

[0030] After the experiment started, in scenario A, the total risk scores reported by both the control group and the experimental group remained stable at their baseline total risk score of 11 throughout the entire 240-minute test period. However, in scenario B, the outputs of the two groups showed a significant difference. The total risk score reported by the control group also remained stable at 11, failing to reflect the application of external vibration stress. In contrast, the total risk score reported by the experimental group... The test showed a clear time-dependent increase. After the vibrator was started, the score rose from an initial 11 to 15.49 at 120 minutes of the test, and finally reached 19.05 at 240 minutes. The test data showed that the control group, lacking the ability to correlate the spatiotemporal information of the blind zone and the quantitative model of risk accumulation over time, could not distinguish between two blind zones with different potential hazard levels. In contrast, the test group, through its configured dynamic risk increment calculation module, coupled the spatial obstruction information of the blind zone with the time accumulation effect of the external vibration source at the data level, and its output total risk score... The changing trend is consistent with the potential risk accumulation process caused by the continuous external stress on the internal structure of the blind zone in scenario B; the experimental results confirm that the technical solution of the present invention can identify and quantify the risk increment caused by external dynamic factors, thereby providing objective data basis for solving the problem of empirical neglect of high-risk blind zones due to lack of information.

[0031] Example 3: This example combines Figures 1 to 3 A description of a method and system for detecting safety hazards in building construction, such as... Figure 1 As shown, the process begins with the inspection equipment initiating an environmental scan. The sensor system feeds back the clarity of the detected data to the inspection equipment. If the signal-to-noise ratio is lower than a threshold of, for example, 15dB, it returns fuzzy perception data. The inspection equipment then requests and obtains a complete list of key checkpoints from the BIM database. For each checkpoint in the list, it requests path planning from the path planning module. After the path planning module executes the algorithm, if the path planning is successful, it returns a passable path. If the path planning fails, it marks the point as an uninspected point. After iterating through all checkpoints, the inspection equipment transmits the final set of uninspected key checkpoints to the risk assessment module. This module calculates the basic total risk score and combines it with the impact of external operations to generate the final risk assessment result. Finally, the result is pushed to the management terminal.

[0032] like Figure 2 As shown in the figure, the graph uses time (minutes) as the horizontal axis and the total risk score as the vertical axis to compare the output results of the method of the present invention (experimental group) and the traditional static method (control group). The dashed line representing the traditional static method (control group) has a constant total risk score of 11 throughout the entire 240-minute test period. The solid line representing the method of the present invention (experimental group) shows that its total risk score starts from the same initial value of 11 as the control group and increases non-linearly over time, reaching 15.49 at 120 minutes and increasing to 19.05 at the end of the 240-minute test. This result intuitively confirms that the method of the present invention can effectively quantify the potential risk increment caused by external dynamic stress that accumulates over time.

[0033] like Figure 3 As shown in the diagram, the core of the detection system is at its center. This core receives sensor data from inspection equipment equipped with sensors such as LiDAR and inertial measurement units, as well as operation information from the site management platform, including high-risk operation types, locations, and start / stop times. The core system includes a checkpoint determination module, which adaptively selects the operating mode based on the clarity of the sensor data to identify the set of uninspected critical checkpoints; a risk assessment module, which integrates the static importance of the structure with the dynamic impact of external operations to calculate and generate a total risk score; and a report generation and push module, which transforms the quantified risk score into intuitive decision-making information and pushes it to designated terminals. During operation, the core system calls upon three external databases: the Building Information Model (BIM), the Risk Value Table (VFA), and the Operation Type Combination Rule Base, ultimately generating a visualized risk report and transmitting it to the management personnel's terminal.

[0034] Example 4: In the context of an upcoming automated safety inspection of a large subway transfer hub project, a systematic on-site calibration procedure is required to establish the mapping relationship between the risk assessment model in this invention and the specific engineering environment. This project employs a large-span arch structure and plans to introduce a new type of high-frequency hydraulic rock drilling rig. Therefore, general risk weights and influencing factor parameters cannot be directly applied; the risk value table and risk parameters related to the operation of the new rock drilling rig must be initialized and calibrated. The calibration process begins with the basic total risk score. The core basis for this approach is the construction of the Risk Value Table. The building information model of the key project is imported into structural analysis software. For each type of critical structural component within the model, finite element stress analysis is performed targeting the main load-bearing path. By simulating the stress distribution of the component and its contribution to the overall structural stability under 150% overload conditions of the standard design load, a quantitative structural importance coefficient is obtained. Subsequently, this coefficient is weighted and averaged with the protection level requirements for this type of component in the local building safety code. Finally, the result is normalized to a range of 1-10, serving as the basic risk weight for this component type. According to this regulation, the keystone of the large-span arch unique to this project... It was determined to be 10, while a non-load-bearing precast partition wall panel... It is then determined to be 2, thus making the risk value statement reflect the structural mechanical characteristics of this particular project.

[0035] Furthermore, for the specific high-risk operation type of the new high-frequency hydraulic rock drilling rig, its risk impact factors are analyzed. With time scale parameters Physical calibration was performed. Three-axis high-precision microelectromechanical system accelerometers were installed at radial positions of 5 meters, 10 meters, and 20 meters within the predetermined initial working area of ​​the drilling rig. The drilling rig was started and operated continuously at standard power for 10 minutes, with vibration signals recorded simultaneously at each measuring point. By analyzing the root mean square amplitude of the signals at the measuring points and comparing it with the permissible vibration limits for concrete structures in industry vibration safety standards, the risk impact factors of this operation were determined. In this calibration, the measured vibration energy at 5 meters was 1.5 times the standard limit. The value was set to 6.0; simultaneously, spectral analysis was performed on the vibration signal to obtain its main vibration frequencies, and the time scale parameter in the risk accumulation function was determined based on the fatigue life curve of concrete under sinusoidal load at the corresponding frequency in mechanics of materials. This parameter is determined to be one-twentieth of the cumulative cycle period corresponding to the initial damage point on the fatigue life curve, used to provide an engineering safety margin. In this scenario, this... The value was set at 45 minutes; finally, after calibrating the risk value table and key operational parameters, the entire risk assessment model was localized for this subway hub project; this procedure anchors the setting of all key parameters in the model to structural analysis based on engineering design documents and measured data based on the on-site physical environment; thus, the total risk score subsequently generated by the system... The numerical source of each component corresponds to the output of the aforementioned structural analysis or physical measurement steps, providing a decision-making basis that is consistent with the on-site conditions for the detection of safety hazards in this specific scenario.

[0036] Example 5: In the environment of deep urban tunnel shield construction, the complex electromagnetic shielding effect leads to unstable wireless network communication. At the same time, high humidity and alkaline water mist generated by concrete spraying operations create a compound interference to optical sensors. When the automated inspection equipment is running under these conditions, its data interface with the site management platform will be interrupted periodically, making it impossible to obtain the spatiotemporal information of high-risk operations in the surrounding area through the application programming interface. To deal with this situation, the method of the present invention is configured to enable a backup procedure for operation type inference and risk assessment based on local multimodal perception information.

[0037] Under this procedure, when the inspection equipment forms a detection blind spot due to path obstruction, and its interface call to the construction site management platform fails to return a successful status within a preset timeout threshold, the system will activate a backup local sensor array, which includes a high-sensitivity acoustic sensor and a temperature and humidity sensor. The system performs a Fast Fourier Transform on the real-time acquired audio stream and matches its spectral characteristics with a pre-set local high-risk operation acoustic feature model library. When an acoustic feature matching the concrete spraying operation is identified, the system associates the operation type with the current detection blind spot. Then, the system calls an exponential function based on a chemical reaction kinetic model corresponding to the chemical curing operation type from a risk accumulation function library. This function library is used to calculate the cumulative risk increment. It is built offline by conducting physicochemical experiments on the curing process of different construction materials, fitting empirical curves of their key performance indicators over time, and pre-setting the mathematical expression of the curve in the library. Through this backup procedure, the risk assessment process can continue to complete the calculation based on local environmental perception data and apply the cumulative model that matches the type of work even under communication interruption conditions.

[0038] In another subsequent application scenario, within a detection blind spot located in the same area of ​​the tunnel, during the risk assessment period after its formation, the system identified ongoing concrete spraying operations through local acoustic sensors and almost simultaneously detected newly added high-frequency hydraulic rock drilling rig operation signals through a high-precision microelectromechanical system accelerometer. The system then queried the operation type combination rule library. Since the library did not have pre-set interaction rules for these two operations, it initially linearly superimposed the risk accumulation increments calculated by the two operations. At the same time, the online learning module used to update the operation type combination rule library, after performing time-frequency analysis on the micro-vibration signals collected during this period, identified the energy component associated with the operating frequency of the rock drilling rig. Its amplitude, under the background of continuous high humidity and acoustic white noise generated by the spraying operation, exhibited a nonlinear superposition phenomenon exceeding 15% of the independent operating state. Based on this, the system quantified the synergistic enhancement coefficient as 1.15 and generated a rule update log pending review. This log, in addition to pushing a visual risk report containing the current total risk score, was also submitted to the management terminal. After obtaining manual confirmation, this newly discovered interaction relationship was solidified into the rule library.

[0039] Example 6: To adapt the technical solution of this invention to the physical characteristics of a specific building environment, a pre-deployment system self-test and sensing parameter calibration procedure must be performed before it is formally deployed in a large hospital building containing various materials such as high-reflectivity glass curtain walls and diffuse-reflective concrete walls. This procedure is used to verify the positioning accuracy of the inspection equipment and to set operating parameters for the equipment sensors that are adapted to the optical and geometric characteristics of the site environment. The procedure first initiates a positioning system accuracy verification step. The inspection equipment is instructed to autonomously run one cycle along a baseline route that has been pre-calibrated by a high-precision static 3D scanner and includes a closed-loop path. This route covers various typical ground and wall materials. After the operation is completed, the system compares the terminal pose calculated by the inspection equipment through real-time positioning and mapping technology with the actual pose of that point in the pre-calibrated baseline map to calculate its absolute positioning error. Only when the error is less than the preset task execution threshold of 5 cm is the equipment authorized to execute subsequent inspection tasks. If it exceeds this threshold, the task is terminated and a prompt is made to recalibrate the positioning sensors such as the inertial measurement unit.

[0040] After the positioning accuracy verification is passed, the system performs adaptive adjustment of the sensing parameters on the same baseline route. During the movement, the device continuously records the signal-to-noise ratio (SNR) of the point cloud data returned by the LiDAR in different material areas. The system selects the two data segments with the highest and lowest SNR, and sets the median of the average SNR values ​​of these two segments as the threshold used to distinguish between clear and blurry sensor data in this task. Simultaneously, the system analyzes the minimum effective point cloud density of point cloud clusters identified as physical obstacles in low signal-to-noise ratio regions, and sets 70% of this density as the non-passage threshold used in the path planning algorithm to determine the presence of obstacles. Through this procedure, the inspection system completes the performance verification of its positioning and perception subsystem before officially starting the safety inspection, and loads key operating parameters corresponding to the physical characteristics of the current working environment for subsequent inspection tasks.

[0041] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0042] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting safety hazards in building construction projects, characterized in that, The method includes: Perform a step to identify a set of unreachable, uninspected critical checkpoints. This step includes selecting one of two execution paths based on the clarity of sensor-perceived data from the inspection equipment: Path 1: Under the condition that the sensor perception data is clear, use the sensor to scan the environment around the physical obstacle to generate environmental data. Combine the positioning information of the inspection equipment to calculate and determine the geometric range of the detection blind zone, and determine the key inspection points located within the geometric range of the detection blind zone in spatial coordinates as the set of uninspected key inspection points. Path Two: Under the condition that sensor data is blurred due to environmental interference, obtain a list of all critical checkpoints within the current task range, and for each critical checkpoint in the list, independently attempt to plan a safe passage path from the current location of the inspection equipment to that critical checkpoint. All critical checkpoints for which a safe passage path cannot be planned are identified as the set of uninspected critical checkpoints. Based on the identified set of uninspected critical checkpoints and the basic risk weights of each uninspected critical checkpoint obtained from the risk value table, accumulate to obtain a basic total risk score. Obtain construction operation information around the spatial location of the detection blind zone, track and accumulate the risk exposure time of the detection blind zone under the influence of high-risk operations. Based on the basic total risk score, and combined with the type of high-risk operation and the corresponding accumulated risk exposure time, generate a total risk score. Generate a visualized risk report containing the total risk score and push the visualized risk report to the management terminal.

2. The method for detecting safety hazards in building engineering according to claim 1, characterized in that, The steps to generate the total risk score based on the basic total risk score, combined with the type of high-risk operation and the corresponding cumulative risk exposure duration, are as follows: For each high-risk operation that affects the detection blind zone, calculate the cumulative risk increment based on the risk impact factor corresponding to its operation type and its corresponding cumulative risk exposure duration. The calculated cumulative risk increments are summed with the base total risk score to obtain the updated total risk score.

3. The method for detecting safety hazards in building engineering according to claim 2, characterized in that, The calculation of the cumulative risk increment is based on a risk accumulation function; this risk accumulation function takes the cumulative risk exposure duration as an input variable and outputs a cumulative effect value that characterizes the nonlinear growth of risk over time. Multiply the cumulative effect value by the risk impact factor to obtain the cumulative risk increment.

4. The method for detecting safety hazards in building engineering according to claim 2, characterized in that, When it is determined that the detection blind zone is simultaneously located within the influence area of ​​multiple high-risk operations, before the step of adding all the calculated cumulative risk increments to the basic total risk score, the following steps are also included: querying the operation type combination rule base to analyze whether there is an interaction relationship between multiple high-risk operation types; if there is an interaction relationship, then the cumulative risk increments are adjusted before accumulation based on the interaction relationship.

5. The method for detecting safety hazards in building engineering according to claim 4, characterized in that, If there is no interaction relationship in the rule base for combining job types, the method further includes: at the edge of the detection blind zone, using sensors mounted on the inspection equipment to monitor the change of at least one physical parameter; simultaneously acquiring the operational characteristics of multiple high-risk jobs; and identifying and quantifying the actual interaction relationship between multiple high-risk job types by analyzing the temporal correlation between the monitored changes in physical parameters and the operational characteristics of external jobs.

6. The method for detecting safety hazards in building engineering according to claim 5, characterized in that, The monitored physical parameters include the amplitude and frequency of the micro-vibration signal, and the operational characteristics include the start-up and shutdown time and operating power of the vibration-type equipment. The steps for analyzing the time correlation between the monitored physical parameter changes and the external operational characteristics include: performing time-frequency analysis on the micro-vibration signal to extract the energy components associated with the operating frequencies of each high-risk operation; when multiple high-risk operations are running simultaneously, determining whether there is a nonlinear superposition phenomenon between their associated energy components; if there is a nonlinear superposition phenomenon, determining the interaction relationship as synergistic enhancement based on the degree of nonlinear superposition, and quantifying its enhancement coefficient.

7. The method for detecting safety hazards in building engineering according to claim 1, characterized in that, In Path 1, the sensor is LiDAR, the environmental data is 3D point cloud data, and the positioning information is obtained through inertial measurement unit and real-time positioning and mapping technology. High-risk operations include heavy equipment vibration operations, deep foundation pit excavation operations, and large-volume concrete pouring operations. The impact area of ​​high-risk operations is a circular area with the center of the high-risk operation as the center and the radius of the impact radius determined according to industry safety standards as the radius.

8. The method for detecting safety hazards in building engineering according to claim 1, characterized in that, The steps for generating a visual risk report include: highlighting the area corresponding to the set of unchecked critical checkpoints on the two-dimensional site plan using a specified color; marking the location of each unchecked critical checkpoint using graphic icons corresponding to the risk level within the highlighted area; and displaying the generated total risk score in numerical form at a specified location on the two-dimensional plan, along with the main high-risk operation types that contribute to the total risk score and their corresponding cumulative risk exposure duration.

9. The method for detecting safety hazards in building engineering according to claim 1, characterized in that, In Path 2, the step of attempting to plan a safe passage route involves performing calculations using the A* pathfinding algorithm or the fast expanding random tree algorithm on a temporary local environment map constructed based on fuzzy perception data.

10. A safety hazard detection system for building engineering, characterized in that, The system includes: A checkpoint determination module, based on the clarity of sensor data from the inspection equipment, selects one of the following two modes to determine the set of uninspected critical checkpoints that cannot be reached: First mode: In this mode, the checkpoint determination module uses sensors to scan the surrounding environment of physical obstacles to generate environmental data, combines this data with the positioning information of the inspection equipment to calculate and determine the geometric range of the detection blind zone, and identifies critical checkpoints located within the geometric range of the detection blind zone in spatial coordinates as the set of uninspected critical checkpoints; Second mode: In this mode, the checkpoint determination module obtains a list of all critical checkpoints within the current task range, and for each critical checkpoint in the list, independently attempts to plan a safe passage path from the current location of the inspection equipment to that critical checkpoint, and identifies all critical checkpoints for which a safe passage path cannot be planned as the set of uninspected critical checkpoints. A risk assessment module is used to: obtain a basic total risk score by summing the basic risk weights of each uninspected key checkpoint obtained from the set of uninspected key checkpoints determined by the checkpoint determination module and querying the risk value table; acquire construction operation information around the spatial location of the detection blind zone, track and accumulate the risk exposure time of the detection blind zone under the influence of high-risk operations; and generate a total risk score based on the basic total risk score, combined with the type of high-risk operation and the corresponding accumulated risk exposure time. A report generation and push module is used to generate a visual risk report containing the total risk score generated by the risk assessment module, and push the visual risk report to the management terminal.

Images