A method for precast component hoisting process safety risk assessment

Abstract

The present application belongs to the technical field of hoisting, and discloses a safety risk assessment method for a prefabricated component hoisting process; comprising the following steps: S1: real-time acquisition of component parameters, operation parameters and environmental parameters of the prefabricated component in each process unit node. The present application sets up a dangerous process unit node, takes reducing the accident probability as the core target, establishes a fine adjustment rule of the hoisting route and hoisting speed, realizes risk control from the root causes such as reducing the collision risk, suppressing the component swing and reducing the operation impact, simultaneously, through an adjustment coefficient prediction model, accurately predicts the reduction effect of different adjustment schemes on the accident probability, realizes the quantitative optimization of the control scheme, ensures that the risk level after adjustment is reduced to within the safety threshold, and establishes a safety-first construction period balancing mechanism to solve the problem of construction period increase caused by risk control, and only the hoisting speed optimization is carried out within the safety red line of the historical low-risk node.

Description

Technical Field

[0001] This invention relates to the field of hoisting technology, and more specifically, to a method for safety risk assessment in the hoisting process of prefabricated components. Background Technology

[0002] With the rapid development of my country's construction industrialization, prefabricated buildings have become the core direction for the industry's transformation and upgrading. As a core construction process, the hoisting of prefabricated components directly determines the project's construction efficiency and safety management level. This operation is characterized by high component weight, frequent high-altitude work, highly dynamic working conditions, and complex risk factors. Once a component falls, equipment overturns, or a collision occurs, it will cause serious casualties, property damage, and project delays, making it a key high-risk management link in the construction field.

[0003] Current safety management of precast component hoisting still faces numerous challenges: In terms of risk assessment, existing technologies largely rely on the experience and judgment of management personnel, or only set threshold alarms for a few single parameters such as component weight and wind speed. This fails to cover all risk factors across components, operations, environment, personnel, and equipment, and cannot achieve dynamic quantitative risk assessment, frequently resulting in over-management of low-risk activities and insufficient management of high-risk activities. In terms of risk warning, existing solutions mostly focus on alarm handling after an accident occurs, failing to accurately predict the probability of accidents during hoisting, thus missing the best opportunity for proactive risk prevention and control, and making it difficult to curb accidents at their source. In terms of intelligent management, existing risk assessment technologies are severely disconnected from on-site management measures. Assessment results cannot be directly translated into feasible hoisting parameter optimization solutions. Even when some solutions propose adjustments to hoisting speed and routes, a quantitative correlation between management measures and risk reduction effects is not established, making it impossible to predict the effectiveness of management. Summary of the Invention

[0004] To address the problems in the background art, this invention proposes a method for safety risk assessment in the hoisting process of prefabricated components.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for safety risk assessment in the hoisting process of precast components, comprising the following steps: S1: Real-time acquisition of component parameters, operation parameters and environmental parameters of prefabricated components at each process unit node, and normalization of the parameters; S2: Generate component risk scores, operational risk scores, and environmental risk scores for each process unit node based on the processed parameters. Then, calculate the comprehensive risk score for each process unit node based on the component risk scores, operational risk scores, and environmental risk scores. Finally, calculate the rate of change of the comprehensive risk score based on the comprehensive risk score. S3: Classify each process unit node according to the comprehensive risk score and the rate of change of the comprehensive risk score, and take corresponding measures for different levels; S4: Obtain historical accident probability data and construct an accident probability prediction model. Use the accident probability prediction model to predict the accident probability of each high-risk node. Mark the dangerous process unit node according to the predicted accident probability and readjust the hoisting route and hoisting speed of the planned dangerous process unit node. S5: Obtain historical adjustment coefficient data and build an adjustment coefficient prediction model. Use the adjustment coefficient prediction model to predict the adjustment coefficient of the dangerous process unit node. Calculate the reduction value of the probability of accident occurrence based on the predicted adjustment coefficient, and decide on the corresponding hoisting route and hoisting speed. S6: Calculate and determine whether all hoisting operations can be completed within the specified hoisting time based on the determined hoisting route and hoisting speed. Adjust the hoisting speed of other process unit nodes according to the number of prefabricated components to be hoisted and the specified hoisting time.

[0006] Furthermore, the process unit nodes of the entire precast component hoisting process include the pre-hoisting preparation node, hoisting operation node, aerial transfer node, positioning and adjustment node, and installation and fixing node. Among them, the nodes in which the precast component is in the hoisting state are the hoisting operation node, aerial transfer node, and positioning and adjustment node. The relevant parameters of these three nodes are collected, including component parameters, operation parameters, and environmental parameters. The construction parameters include the weight, size, center of gravity offset, and lifting point position deviation of the prefabricated components; the operation parameters include the lifting sling angle deviation, lifting acceleration, and component swing amplitude; and the environmental parameters include wind speed, visibility, and obstacle density. The above parameters are normalized.

[0007] Furthermore, the process of generating component risk scores, operational risk scores, and environmental risk scores for each process unit node based on the processed parameters includes: Component risk scoring :

[0008] In the formula, , , and These are weighting coefficients, obtained through training based on historical data; Operational risk score :

[0009] In the formula, , and These are weighting coefficients, obtained through training based on historical data; Environmental risk scoring :

[0010] In the formula, , and These are weighting coefficients, obtained through training based on historical data.

[0011] Furthermore, the process of calculating the comprehensive risk score for each process unit node based on component risk scores, operational risk scores, and environmental risk scores includes: Overall risk score P:

[0012] In the formula, , and These are weighting coefficients, obtained through training based on historical data; The process of calculating the rate of change of the comprehensive risk score through the comprehensive risk score includes: The assessment cycle is set, and the process unit node corresponding to the precast component currently being hoisted is taken as the object. The comprehensive risk score value of five consecutive assessment cycles within the node is continuously collected according to the assessment cycle. First, the difference in comprehensive risk score between two adjacent assessment cycles is calculated. The score of the previous assessment cycle is subtracted from the score of the later assessment cycle to obtain the score change value within each assessment cycle interval. Then, the score change values ​​of all assessment cycle intervals are added together to obtain the total change in score within the node. Finally, the total change is divided by the total time span of these five assessment cycles to obtain the comprehensive risk score change rate per unit time.

[0013] Furthermore, the process of classifying each process unit node based on the comprehensive risk score and the rate of change of the comprehensive risk score, and taking corresponding measures for different levels, includes: Based on the rate of change of the comprehensive risk score, the risk level of each process unit node is classified. The specific classification process is as follows: a first risk threshold and a second risk threshold are set in advance, and a risk change rate warning threshold is set at the same time. First, the comprehensive risk score of the node is compared with the two risk thresholds. When the comprehensive risk score is less than the first risk threshold, regardless of the rate of change of the score, the node is marked as a low-risk node and routine control measures are implemented, namely, continuously collecting and monitoring risk data, providing routine safety reminders to operators, and not interfering with the normal hoisting operation process. When the comprehensive risk score is greater than or equal to the first risk threshold and less than or equal to the second risk threshold, the change rate of the comprehensive risk score is further examined. If the change rate is less than the risk change rate warning threshold, the node is marked as a medium-risk node and enhanced control measures are implemented, namely, increasing the frequency of risk data collection, providing real-time warnings for risk indicators exceeding the standard, arranging on-site safety officers to supervise, and reminding workers to operate in a standardized manner. If the change rate is greater than or equal to the risk change rate warning threshold, the node is directly upgraded and marked as a high-risk node. When the comprehensive risk score exceeds the second risk threshold, regardless of the rate of change of the score, the node is directly marked as a high-risk node, and emergency control measures are implemented. The crane operator is immediately given an order to suspend the lifting operation, the equipment operation authority is locked, the specific indicators that caused the risk to exceed the standard are identified at the same time, and special rectification requirements are pushed to the on-site management personnel. After the rectification is completed, the risk score is re-verified, and the lifting operation can only be resumed after the risk is confirmed to have dropped to a safe range.

[0014] Furthermore, the process of acquiring historical accident occurrence probability data and constructing an accident occurrence probability prediction model, and then using this model to predict the accident occurrence probability at each high-risk node, includes: The probability of an accident occurring refers to the probability value of a safety accident occurring at a high-risk node; Factors influencing the probability of accidents include: current overall risk score, current overall score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and historical accident frequency. Obtain historical accident probability data for a single high-risk node in different lifting tasks. The historical accident probability data includes the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, historical accident frequency, and historical accident probability of the single high-risk node in different lifting tasks. Based on the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, frequency of historical accidents and corresponding historical accident probabilities of high-risk nodes in different historical accident probability data, an accident probability prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, taking the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and frequency of historical accidents from different historical accident occurrence probability data in the first training set as the input data of the first convolutional neural network, and taking the corresponding historical accident occurrence probability in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the accident occurrence probability prediction model. The current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, crane aging index, and historical accident frequency of each high-risk node are input into the accident probability prediction model to obtain the predicted accident probability for each high-risk node.

[0015] Furthermore, the process of marking hazardous process unit nodes based on the predicted probability of accident occurrence includes: Pre-set an accident probability threshold. The predicted accident probability is obtained once in each evaluation cycle. The predicted accident probability output by high-risk nodes in real time is continuously compared with the accident probability threshold. When the predicted accident probability at a certain moment is less than or equal to the accident probability threshold, the node is determined to be in a safe and controllable state. No additional intervention is required, and continuous monitoring is sufficient. When the predicted probability of an accident at a certain moment exceeds the accident probability threshold, the process unit node is immediately marked as a dangerous process unit node, and an audible and visual warning is triggered simultaneously. Risk warning information is pushed to on-site operators and managers, and the subsequent hoisting route and speed adjustment process is initiated to reduce the probability of an accident through proactive control measures. The process of readjusting the hoisting route and speed for hazardous process unit nodes includes: When a process unit node is marked as a hazardous process unit node, the current hoisting operation will be immediately suspended. At the same time, based on the current risk factors, component condition, and working environment, the hoisting route and hoisting speed for that node will be readjusted. The specific process is as follows: In terms of route planning, the complete three-dimensional scene data of the current hoisting operation space is first obtained through three-dimensional laser scanning. Combined with the real-time position of the component, the target positioning position, the distribution of identified obstacles, and wind speed and direction data, the hoisting route is replanned with the core principles of "minimizing the swing amplitude of the component, maximizing the safe distance from obstacles, and avoiding areas affected by strong winds". For aerial transfer nodes, the shortest straight transfer route is prioritized to reduce the combined actions of crane rotation and luffing. At the same time, it is ensured that the safe distance between the component and surrounding obstacles is not less than 2 meters. For positioning and adjustment nodes, the vertical slow positioning route is prioritized to reduce horizontal adjustment actions and avoid collisions between the component and installed components or edge protection. Regarding hoisting speed adjustment, the speed adjustment coefficient is first calculated and determined based on the real-time predicted probability of accidents. The calculation method for the speed adjustment coefficient is as follows: Step 1: Calculate the base coefficient The calculation formula is: = / ; In the formula, The preset accident probability threshold, The base coefficient is used to predict the probability of an accident occurring in real time for the current node. The range of values ​​is 0 < ≤1; Step 2: Determine if the core risk factors exceed the standard. Perform quantitative judgment based on the normalized parameters collected in real time at the current node: when the normalized lifting acceleration A ≥ 0.6, it is determined that the lifting acceleration exceeds the standard; when the normalized component swing amplitude B ≥ 0.7, it is determined that the component swing amplitude is large. Step 3: Based on the results of the exceedance judgment, the basic coefficient is quantitatively corrected. The specific mathematical model for correction is as follows: If the problem is solely due to excessive lifting acceleration, calculate the acceleration correction factor. The calculation formula is: =1 0.4×A; In the formula, A is the normalized lifting acceleration collected in real time at the current node. The value range is (0, 1]; the corrected speed adjustment coefficient K= × ; If the only cause is a large component sway amplitude, calculate the sway correction factor. The calculation formula is: =1 0.3×B; In the formula, B is the normalized component swing amplitude collected in real time by the current node. The value range is (0, 1]; Corrected speed adjustment coefficient K= × ; If both the lifting acceleration and component sway amplitude are determined to be excessive, a comprehensive correction factor k is calculated. The formula is: k = × ; Corrected speed adjustment coefficient K= ×k; Step 4: Boundary constraints for speed adjustment coefficient. The corrected speed adjustment coefficient K must satisfy 0.2≤K≤0.5; The final hoisting speed is equal to the original hoisting speed multiplied by the corrected speed adjustment coefficient K. The adjusted hoisting mechanism speed shall not exceed 40% of the original speed, and the slewing and luffing mechanism speed shall not exceed 30% of the original speed. At the same time, a minimum stable speed threshold is set to avoid the components from swaying due to excessively low speed.

[0016] Furthermore, the process of acquiring historical adjustment coefficient data and constructing an adjustment coefficient prediction model, and then using this model to predict the adjustment coefficients for hazardous process unit nodes, includes: The adjustment coefficient refers to the degree of change in the probability of an accident at a hazardous process unit node after adjusting the hoisting route and hoisting speed. Specifically, it is the ratio of the predicted accident probability after adjustment to the predicted accident probability before adjustment. The value range of the adjustment coefficient is 0-1. Factors affecting the adjustment coefficient include: the adjusted hoisting speed, the average safety distance, the complexity of the action, the change in the windward area, and the operational response redundancy time. Obtain historical adjustment coefficient data for a single hazardous process unit node in different lifting tasks. The historical adjustment coefficient data includes the adjusted lifting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and historical adjustment coefficient of the single hazardous process unit node in different lifting tasks. Based on the adjusted hoisting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and corresponding historical adjustment coefficients of the corresponding dangerous process unit nodes in different historical adjustment coefficient data, an adjustment coefficient prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time in the different historical adjustment coefficient data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical adjustment coefficients in the second training set are used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the adjustment coefficient prediction model. The adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time of each hazardous process unit node are input into the adjustment coefficient prediction model to obtain the predicted adjustment coefficient for each hazardous process unit node.

[0017] Furthermore, the process of calculating the reduction in the probability of an accident based on the prediction adjustment coefficient, and then deciding on the corresponding hoisting route and hoisting speed, includes: After generating multiple alternative hoisting routes and hoisting speed adjustment schemes for hazardous process unit nodes, for each scheme, the prediction adjustment coefficient corresponding to the scheme is predicted by the adjustment coefficient prediction model. The predicted accident probability before adjustment is multiplied by the prediction adjustment coefficient to calculate the predicted accident probability value after the scheme is adjusted. Then, the predicted accident probability after each set of schemes is compared with the preset probability threshold. Invalid schemes that are still greater than the accident probability threshold after adjustment are eliminated. For valid schemes whose predicted probability after adjustment is less than or equal to the accident probability threshold, the scheme with the smallest adjustment coefficient and the largest reduction in accident probability is selected first. Finally, the optimal hoisting route and hoisting speed scheme is determined, and an execution command is issued to the crane control system to complete the hoisting operation of this node according to the determined scheme.

[0018] Furthermore, the process of calculating and determining whether all lifting operations can be completed within the specified lifting time based on the determined lifting route and lifting speed includes: After completing the hoisting operation of the dangerous process unit node according to the determined optimal adjustment plan, the actual increase in hoisting time due to the adjustment of the route and speed will be calculated based on the adjusted hoisting route length and hoisting speed. Then, the number of unfinished process nodes of the prefabricated components hoisted this time, as well as the total number of prefabricated components to be hoisted in the entire hoisting operation plan, the standard hoisting time of each component, and the hoisting completion time set in advance for the project will be calculated. Based on this, the current cumulative hoisting time, the remaining available hoisting time, and the total standard time of the remaining hoisting operations will be calculated to determine whether the hoisting operation can be completed within the specified time according to the current hoisting pace. If it is found that the cumulative increase in hoisting time is within the redundancy range of the remaining available time, and all hoisting operations can be completed within the specified time, the system will not adjust the hoisting plans for other components or other nodes, and will only continue to perform normal hoisting operations in accordance with risk control requirements. If the accumulated increase in time exceeds the redundancy limit, and the current pace makes it impossible to complete all hoisting operations within the specified time, the system will, while ensuring safety, optimize and adjust the speed of subsequent hoisting nodes for other precast components. The specific process is as follows: First, identify the process unit nodes among the components to be hoisted that have consistently had a comprehensive risk score below the first risk threshold and an accident probability far below the accident probability threshold during the historical hoisting of similar components. For these process unit nodes, under the premise of not exceeding the maximum hoisting speed specified in the hoisting safety regulations and not changing the original safe hoisting route, reasonably optimize the hoisting speed of the stable operation section. The optimized speed should not exceed 80% of the maximum speed allowed by the regulations. At the same time, ensure that the predicted accident probability of the optimized node is still far below the accident probability threshold and will not cause new safety risks. By optimizing the speed of multiple low-risk nodes, the hoisting time of a single component is reduced to make up for the increased time of the previously dangerous nodes. For each node after speed adjustment, the comprehensive risk score and predicted accident probability will be recalculated to ensure that the accident probability of all nodes does not exceed the accident probability threshold. If no adjustable node that meets the safety requirements can be found, a schedule warning will be immediately sent to the project management personnel, indicating that additional hoisting equipment and adjustments to the hoisting operation plan are needed to avoid exceeding the safety red line in order to meet the schedule.

[0019] The technical effects and advantages of the present invention regarding a method for safety risk assessment in the hoisting process of precast components are as follows: (1) By setting dangerous process unit nodes, with the core objective of reducing the probability of accidents, a refined adjustment rule for hoisting route and hoisting speed was established. Risk control was achieved from the root causes such as reducing collision risk, suppressing component swing, and reducing operational impact. At the same time, by adjusting the coefficient prediction model, the effect of different adjustment schemes on the reduction of accident probability was accurately predicted, and the quantitative optimization of control schemes was achieved. This ensured that the risk level after adjustment was reduced to within the safety threshold, avoiding the waste of construction period caused by blind control. In response to the problem of increased construction period caused by risk control, this invention established a safety-first construction period balancing mechanism, and only optimized the hoisting speed within the safety red line of historical low-risk nodes.

[0020] (2) By targeting the three core high-risk nodes of lifting operation, aerial transfer and positioning adjustment of precast components in a suspended state, three core risk parameters of components, operation and environment were accurately selected, and a standardized parameter normalization processing method was established. Through a hierarchical weighted scoring system, the risk of each node was quantitatively represented. At the same time, by combining the comprehensive risk score and the score change rate, the static risk level and the dynamic risk evolution trend were dually assessed. Compared with the traditional fixed threshold alarm, the risk deterioration trend can be identified earlier, and the targeting of risk classification and control is significantly improved. By accurately predicting the probability of node accidents through multi-dimensional feature input, the accurate decision-making basis for risk prevention and control is provided. It can effectively identify safety hazards in the lifting process in advance, greatly reduce the probability of safety accidents such as component falling and collision, and ensure the inherent safety of lifting operations. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0022] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 1 A method for safety risk assessment of precast component hoisting procedures includes the following steps: S1: Data acquisition module: Real-time acquisition of component parameters, operation parameters and environmental parameters of prefabricated components at each process unit node, and normalization of the parameters; S2: Risk Assessment Module: Generates component risk score, operational risk score and environmental risk score for each process unit node based on the processed parameters, and then calculates the comprehensive risk score for each process unit node based on the component risk score, operational risk score and environmental risk score, and calculates the comprehensive risk score change rate through the comprehensive risk score. S3: Risk Classification Module: Classifies each process unit node according to the comprehensive risk score and the rate of change of the comprehensive risk score, and takes corresponding measures for different levels; S4: Accident Probability Prediction Module: Acquires historical accident probability data and constructs an accident probability prediction model. Uses the accident probability prediction model to predict the accident probability of each high-risk node. Marks dangerous process unit nodes based on the predicted accident probability and readjusts the hoisting route and hoisting speed of the planned dangerous process unit nodes. S5: Adjustment coefficient prediction module: acquires historical adjustment coefficient data and builds an adjustment coefficient prediction model, uses the adjustment coefficient prediction model to predict the adjustment coefficient of dangerous process unit nodes, calculates the reduction value of accident probability based on the predicted adjustment coefficient, and makes decisions on selecting the corresponding hoisting route and hoisting speed. S6: Lifting Operation Planning and Adjustment Module: Calculates and determines whether all lifting operations can be completed within the specified lifting time based on the determined lifting route and lifting speed, and adjusts the lifting speed of other process unit nodes according to the number of prefabricated components to be lifted and the specified lifting time.

[0024] It should be further explained that, in the specific implementation process, the process unit nodes of the entire process of prefabricated component hoisting include the pre-hoisting preparation node, hoisting operation node, aerial transfer node, positioning and adjustment node, and installation and fixing node. Among them, the nodes in which the prefabricated component is in the hoisting state are the hoisting operation node, aerial transfer node, and positioning and adjustment node. The relevant parameters of these three nodes are collected, including component parameters, operation parameters, and environmental parameters. The construction parameters include the weight, size, center of gravity offset, and lifting point position deviation of the prefabricated components; the operation parameters include the lifting sling angle deviation, lifting acceleration, and component swing amplitude; and the environmental parameters include wind speed, visibility, and obstacle density. The above parameters are normalized. The theoretical weight is extracted from the BIM design model of the precast component, and the actual weight is obtained by reviewing the weighing and acceptance records when the component arrives on site. The normalized weight M is obtained by subtracting the lower limit of the weight from the actual measured weight of the precast component, and then dividing by the difference between the upper limit and the lower limit of the weight. The lower limit of weight is 0 (no load means no risk), and the upper limit of weight is the rated maximum lifting capacity of the crane used in this hoisting under the current working conditions, which is 20 tons. Extract the length, width, and height design dimensions from the component BIM design drawings. Verify the actual outer contour dimensions of the component by on-site laser scanning and steel tape measure. First, subtract the lower limit value of the dimension from the maximum measured contour dimension of the precast component, and then divide by the difference between the upper limit value and the lower limit value to obtain the normalized dimension C. Among them, the lower limit of the size is 0 (there is no related risk if there is no component volume), and the upper limit of the size is taken as the maximum allowable outer contour size of the component under the normal hoisting conditions of this project, which is 6 meters. The theoretical center of gravity coordinates of the component are calculated by BIM model. Combined with the actual size deviation of the component and the actual position of the embedded lifting parts, the horizontal offset distance between the actual center of gravity of the component and the geometric center of the lifting point is obtained by mechanical calculation. First, the lower limit of the center of gravity offset is subtracted from the measured center of gravity offset, and then divided by the difference between the upper limit of the center of gravity offset and the lower limit of the center of gravity offset to obtain the normalized center of gravity offset Z. Among them, the lower limit of the center of gravity offset is 0mm (the center of gravity is completely coincident with the center of the lifting point, and there is no risk of offset), and the upper limit of the center of gravity offset is the specified maximum allowable offset limit of 50mm. The actual installation position of the embedded lifting parts of the component was obtained by on-site laser scanning and steel tape measure measurement. The position of each lifting point was compared with the lifting point position marked on the design drawings. The position deviation value of each lifting point was calculated. The maximum deviation value was taken as the calculation index. First, the lower limit of the deviation was subtracted from the measured maximum deviation value of the lifting point. Then, the difference between the upper limit of the deviation value and the lower limit of the deviation value was divided to obtain the normalized lifting point position deviation D. Among them, the lower limit of deviation is 0mm (the position of the lifting point is completely consistent with the design and there is no risk of deviation), and the upper limit of deviation is the maximum allowable deviation of the lifting point installation specified in the standard, which is 30mm. By using attitude sensors installed on the lifting slings, the actual angle between the two main slings is collected in real time. First, the lower limit of the angle is subtracted from the absolute deviation of the actual angle from 60°, and then divided by the difference between the upper and lower limits of the angle to obtain the normalized lifting sling angle deviation J. The optimal included angle is 60°. First, the absolute difference between the actual included angle and 60° is calculated as the basis for deviation calculation. Then, the lower limit of safety is determined to be 0° (which is completely consistent with the optimal included angle and has no deviation risk). The upper limit of risk is 30°. The vertical acceleration values ​​during the lifting process of the component are collected in real time by a triaxial accelerometer installed on the crane hook. The maximum acceleration during the lifting process is taken as the calculation index. First, the lower limit of acceleration is subtracted from the measured maximum lifting acceleration, and then divided by the difference between the upper and lower limit of acceleration to obtain the normalized lifting acceleration A. The lower limit of acceleration is 0 m / s² (smooth lifting, no acceleration impact), and the upper limit of acceleration is 0.5 m / s². By using attitude sensors and laser rangefinders installed on the components, the maximum horizontal swing distance relative to the vertical stationary position of the components is collected in real time during the hoisting process. The maximum swing amplitude during the operation is taken as the calculation index. First, the lower limit of the swing amplitude is subtracted from the measured maximum swing amplitude of the components, and then divided by the difference between the upper limit and the lower limit of the swing amplitude to obtain the normalized component swing amplitude B. The lower limit of the swing amplitude is 0m (the component is completely stationary and there is no risk of swinging), and the upper limit of the swing amplitude is 0.8m. By using a mobile weather station set up at the hoisting operation point, the instantaneous wind speed at the hoisting height of the component is collected in real time. The average wind speed within 1 minute is taken as the calculation index. First, the measured average wind speed is subtracted from the lower limit of the wind speed, and then divided by the difference between the upper limit of the wind speed and the lower limit of the wind speed to obtain the normalized wind speed V. The lower limit for wind speed is 0 m / s (no wind, no wind risk), and the upper limit for wind speed is 10.8 m / s. The visibility values ​​of the work area are collected in real time by the visibility meter of the on-site weather station. The image clarity analysis of the on-site visual recognition equipment can also be used to assist in the verification. The real-time visibility value is taken as the calculation index. First, the lower limit of visibility is subtracted from the measured visibility value. Then, it is divided by the difference between the upper limit of visibility and the lower limit of visibility. Finally, the above calculation result is subtracted from 1 to obtain the normalized visibility N. The minimum visibility limit is 30m (if visibility is below 30m, hoisting operations are completely impossible, reaching the highest risk), and the maximum visibility limit is 1000m (if visibility is above 1000m, the line of sight is completely unobstructed, and there is no risk). By using 3D laser scanning of the construction site, the quantity, size and spatial distribution of all obstacles (installed components, scaffolding, edge protection, construction machinery, temporary facilities, etc.) within the hoisting operation space are obtained. The volume ratio of obstacles within the space covered by the hoisting route is calculated to obtain the obstacle density index. First, the measured obstacle density is subtracted from the lower limit of density, and then divided by the difference between the upper limit of density and the lower limit of density to obtain the normalized density Q. The lower limit of density is 0 (no obstacles in the work space and no risk of collision), and the upper limit of density is 0.6.

[0025] It should be further explained that, in the specific implementation process, the process of generating component risk scores, operational risk scores, and environmental risk scores for each process unit node based on the processed parameters includes: Component risk scoring :

[0026] In the formula, , , and These are weighting coefficients, derived from historical data training, and set to 0.2, 0.15, 0.35, and 0.3 respectively. If the component parameters for a certain process unit node are: the actual measured weight of the precast component is 8 tons, the actual measured size is 3.6 meters, the actual offset between the actual center of gravity and the lifting point center is 15 mm, and the actual maximum deviation of the lifting point position is 9 mm, then the component risk score for this process unit node is... It is 0.365; Operational risk score :

[0027] In the formula, , and These are weighting coefficients, derived from historical data training, and set to 0.3, 0.25, and 0.45 respectively; If the operational parameters of a certain process unit node are: the actual included angle of the sling is 75°, the actual lifting acceleration is 0.15 m / s², and the actual horizontal swing amplitude is 0.24 m, then the operational risk score for this process unit node is... It is 0.36; Environmental risk scoring :

[0028] In the formula, , and These are weighting coefficients, derived from historical data training, and set to 0.5, 0.35, and 0.15 respectively. If the environmental parameters for a certain process unit node are: average wind speed at the measured hoisting height is 4.32 m / s, measured on-site visibility is 418 m, and measured obstacle density is 0.12, then the environmental risk score for that process unit node is... It is 0.44.

[0029] It should be further explained that, in the specific implementation process, the process of calculating the comprehensive risk score of each process unit node based on component risk score, operational risk score, and environmental risk score includes: Overall risk score P:

[0030] In the formula, , and These are weighting coefficients, derived from historical data training, and set to 0.35, 0.4, and 0.25 respectively; If the component risk score of a certain process unit node The operational risk score is 0.365. The environmental risk score is 0.36. If the value is 0.44, then the comprehensive risk score P for this process unit node is 0.382; The process of calculating the rate of change of the comprehensive risk score through the comprehensive risk score includes: The assessment cycle is set, and the process unit node corresponding to the precast component currently being hoisted is taken as the object. The comprehensive risk score value of five consecutive assessment cycles within the node is continuously collected according to the assessment cycle. First, the difference in comprehensive risk score between two adjacent assessment cycles is calculated. The score of the previous assessment cycle is subtracted from the score of the later assessment cycle to obtain the score change value within each assessment cycle interval. Then, the score change values ​​of all assessment cycle intervals are added together to obtain the total change in score within the node. Finally, the total change is divided by the total time span of these five assessment cycles to obtain the comprehensive risk score change rate per unit time.

[0031] It should be further explained that, in the specific implementation process, the process of classifying each process unit node according to the comprehensive risk score and the rate of change of the comprehensive risk score, and taking corresponding measures for different levels, includes: Based on the rate of change of the comprehensive risk score, the risk level of each process unit node is classified. The specific classification process is as follows: a first risk threshold and a second risk threshold are set in advance, and a risk change rate warning threshold is set at the same time. First, the comprehensive risk score of the node is compared with the two risk thresholds. When the comprehensive risk score is less than the first risk threshold, regardless of the rate of change of the score, the node is marked as a low-risk node and routine control measures are implemented, namely, continuously collecting and monitoring risk data, providing routine safety reminders to operators, and not interfering with the normal hoisting operation process. When the comprehensive risk score is greater than or equal to the first risk threshold and less than or equal to the second risk threshold, the change rate of the comprehensive risk score is further examined. If the change rate is less than the risk change rate warning threshold, the node is marked as a medium-risk node and enhanced control measures are implemented, namely, increasing the frequency of risk data collection, providing real-time warnings for risk indicators exceeding the standard, arranging on-site safety officers to supervise, and reminding workers to operate in a standardized manner. If the change rate is greater than or equal to the risk change rate warning threshold, the node is directly upgraded and marked as a high-risk node. When the comprehensive risk score exceeds the second risk threshold, regardless of the rate of change of the score, the node is directly marked as a high-risk node, and emergency control measures are implemented. The crane operator is immediately given an order to suspend the lifting operation, the equipment operation authority is locked, the specific indicators that caused the risk to exceed the standard are identified at the same time, and special rectification requirements are pushed to the on-site management personnel. After the rectification is completed, the risk score is re-verified, and the lifting operation can only be resumed after the risk is confirmed to have dropped to a safe range.

[0032] It should be further explained that, in the specific implementation process, the process of acquiring historical accident occurrence probability data and constructing an accident occurrence probability prediction model, and then using the accident occurrence probability prediction model to predict the accident occurrence probability of each high-risk node, includes: The probability of an accident occurring refers to the probability value of a safety accident occurring at a high-risk node; Factors influencing the probability of accidents include: current overall risk score, current overall score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and historical accident frequency. The extreme value of the sub-item risk score is the highest value extracted from the three sub-item risk scores of the current component. The experience index of the crane operators and signalmen currently in operation is obtained through the on-site personnel management system. The aging index of the crane equipment is obtained through the equipment records of the crane. Extract historical operation data with the same component type and process node as the current one from the project's historical hoisting database and the industry hoisting accident database, and count the frequency of safety accidents that occurred in them; Obtain historical accident probability data for a single high-risk node in different lifting tasks. The historical accident probability data includes the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, historical accident frequency, and historical accident probability of the single high-risk node in different lifting tasks. Based on the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, frequency of historical accidents and corresponding historical accident probabilities of high-risk nodes in different historical accident probability data, an accident probability prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, taking the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and frequency of historical accidents from different historical accident occurrence probability data in the first training set as the input data of the first convolutional neural network, and taking the corresponding historical accident occurrence probability in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is a one-dimensional convolutional neural network adapted to one-dimensional structured feature input, and the specific network structure is set as follows: The network consists of five layers: an input layer, a convolutional feature extraction layer, a pooling layer, a fully connected layer, and an output layer. The input layer has 6 neurons, matching the dimensions of the 6 input feature parameters. The input data is a feature vector normalized to the [0, 1] interval using min-max normalization. The convolutional feature extraction layer consists of two consecutive one-dimensional convolutional layers. The first convolutional layer has 32 filters, a kernel size of 3, a stride of 1, and equal-length padding. The second convolutional layer has 16 filters, a kernel size of 3, a stride of 1, and equal-length padding. The two convolutional layers use the ReLU function to extract non-linear relationships between features. The pooling layer is set to a one-dimensional global average pooling layer to reduce the dimensionality of the features extracted by convolution, retaining core features while reducing the computational cost of the model. There are two fully connected layers: the first fully connected layer has 32 neurons and the second fully connected layer has 8 neurons. The activation function of both fully connected layers is the ReLU function. The output layer has 1 neuron and uses the Sigmoid function to map the output value to the interval [0, 1], corresponding to the predicted probability of the accident. The specific hyperparameter settings for model training are as follows: the optimizer uses the Adam adaptive optimizer, the initial learning rate is set to 0.001, the learning rate decay strategy is to decay the learning rate to 0.9 every 10 epochs, the batch size is set to 32, the maximum training epochs are set to 200, and the loss function is the mean squared error (MSE) function, which is adapted to the training requirements of the regression prediction task. The following measures were taken to prevent overfitting during training: Dropout layers were added after each of the two fully connected layers, with a random inactivation rate of 0.3. During training, some neurons were randomly inactivated to reduce the model's dependence on the training data for overfitting. At the same time, L2 regularization was added with a regularization coefficient of 0.001 to constrain the network's weight parameters and avoid overfitting caused by excessive weights. In addition, an early stopping mechanism was enabled during training. When the loss value on the validation set did not decrease for 15 consecutive epochs, training was terminated early to retain the optimal weight parameters. The training of the model is terminated when any of the following conditions are met: first, the preset maximum number of iterations (200 epochs) is reached; second, the early stopping mechanism is triggered and the loss on the validation set does not decrease for 15 consecutive epochs; and third, the loss on the training set decreases to below 0.005 and tends to stabilize. The preset first test error threshold is set to 0.03. This threshold is determined based on the accuracy requirements of hoisting safety management. When the model's mean absolute error (MAE) on the test set is less than or equal to 0.03, it means that the deviation between the model's predicted value and the actual probability of an accident does not exceed 3%, which can meet the accuracy requirements of on-site safety early warning. If the model verification error exceeds this threshold, it will be retrained by expanding the training samples and adjusting the network hyperparameters until the error threshold requirements are met. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the accident occurrence probability prediction model. The current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and historical accident frequency of each high-risk node are input into the accident probability prediction model to obtain the predicted accident probability of each high-risk node. In an embodiment of the present invention, the predicted accident probability of all high-risk nodes is obtained through an accident probability prediction model. The predicted accident probability is related to the current comprehensive risk score, the current comprehensive score change rate, the extreme value of the sub-risk score, the operator experience index, the lifting equipment aging index, and the frequency of historical accidents. The current comprehensive risk score directly affects the probability of an accident occurring. The higher the current comprehensive risk score, the greater the base probability of an accident occurring, and the greater the predicted probability of an accident occurring. Therefore, the current comprehensive risk score is positively correlated with the probability of an accident occurring. The rate of change of the current comprehensive risk score directly affects the probability of an accident. The higher the rate of change of the current comprehensive risk score, the higher the possibility of the risk rapidly rising to the accident threshold, and the greater the probability of an accident. Therefore, the rate of change of the current comprehensive risk score is positively correlated with the probability of an accident. The extreme values ​​of the sub-risk scores directly affect the probability of an accident. The higher the extreme value of the sub-risk score, the higher the probability of an accident caused by a single-dimensional risk exceeding the safety threshold, and the greater the probability of an accident. Therefore, the extreme value of the sub-risk score is positively correlated with the probability of an accident. The size of the operator's experience index directly affects the probability of an accident. The larger the operator's experience index, the more experienced the operator is, the more standardized the operation is, the stronger the ability to deal with sudden risks, and the lower the probability of an accident. Therefore, the operator's experience index is negatively correlated with the probability of an accident. The size of the aging index of lifting equipment directly affects the probability of predicting accidents. The larger the aging index, the more serious the aging of the equipment, the higher the probability of failure, and the higher the probability of predicting accidents. Therefore, the aging index of lifting equipment is positively correlated with the probability of accidents. The frequency of historical accidents directly affects the probability of predicted accidents. The higher the frequency of historical accidents, the higher the inherent risk of this type of operation, and the higher the probability of predicted accidents. Therefore, the frequency of historical accidents and the probability of accidents are positively correlated.

[0033] It should be further explained that, in the specific implementation process, the process of marking hazardous process unit nodes based on the predicted probability of accident occurrence includes: Pre-set an accident probability threshold. The predicted accident probability is obtained once in each evaluation cycle. The predicted accident probability output by high-risk nodes in real time is continuously compared with the accident probability threshold. When the predicted accident probability at a certain moment is less than or equal to the accident probability threshold, the node is determined to be in a safe and controllable state. No additional intervention is required, and continuous monitoring is sufficient. When the predicted probability of an accident at a certain moment exceeds the accident probability threshold, the process unit node is immediately marked as a dangerous process unit node, and an audible and visual warning is triggered simultaneously. Risk warning information is pushed to on-site operators and managers, and the subsequent hoisting route and speed adjustment process is initiated to reduce the probability of an accident through proactive control measures. The process of readjusting the hoisting route and speed for hazardous process unit nodes includes: When a process unit node is marked as a hazardous process unit node, the current hoisting operation will be immediately suspended. At the same time, based on the current risk factors, component condition, and working environment, the hoisting route and hoisting speed for that node will be readjusted. The specific process is as follows: In terms of route planning, the complete three-dimensional scene data of the current hoisting operation space is first obtained through three-dimensional laser scanning. Combined with the real-time position of the component, the target positioning position, the distribution of identified obstacles, and wind speed and direction data, the hoisting route is replanned with the core principles of "minimizing the swing amplitude of the component, maximizing the safe distance from obstacles, and avoiding areas affected by strong winds". For aerial transfer nodes, the shortest straight transfer route is prioritized to reduce the combined actions of crane rotation and luffing. At the same time, it is ensured that the safe distance between the component and surrounding obstacles is not less than 2 meters. For positioning and adjustment nodes, the vertical slow positioning route is prioritized to reduce horizontal adjustment actions and avoid collisions between the component and installed components or edge protection. Regarding hoisting speed adjustment, the speed adjustment coefficient is first calculated and determined based on the real-time predicted probability of accidents. The calculation method for the speed adjustment coefficient is as follows: Step 1: Calculate the base coefficient The calculation formula is: = / ; In the formula, The preset accident probability threshold, The base coefficient is used to predict the probability of an accident occurring in real time for the current node. The range of values ​​is 0 < ≤1; Step 2: Determine if the core risk factors exceed the standard. Perform quantitative judgment based on the normalized parameters collected in real time at the current node: when the normalized lifting acceleration A ≥ 0.6, it is determined that the lifting acceleration exceeds the standard; when the normalized component swing amplitude B ≥ 0.7, it is determined that the component swing amplitude is large. Step 3: Based on the results of the exceedance judgment, the basic coefficient is quantitatively corrected. The specific mathematical model for correction is as follows: If the problem is solely due to excessive lifting acceleration, calculate the acceleration correction factor. The calculation formula is: =1 0.4×A; In the formula, A is the normalized lifting acceleration collected in real time at the current node. The value range is (0, 1]; the corrected speed adjustment coefficient K= × ; If the only cause is a large component sway amplitude, calculate the sway correction factor. The calculation formula is: =1 0.3×B; In the formula, B is the normalized component swing amplitude collected in real time by the current node. The value range is (0, 1]; Corrected speed adjustment coefficient K= × ; If both the lifting acceleration and component sway amplitude are determined to be excessive, a comprehensive correction factor k is calculated. The formula is: k = × ; Corrected speed adjustment coefficient K= ×k; Step 4: Boundary constraints for speed adjustment coefficient. The corrected speed adjustment coefficient K must satisfy 0.2≤K≤0.5; The final hoisting speed is equal to the original hoisting speed multiplied by the corrected speed adjustment coefficient K. The adjusted hoisting mechanism speed shall not exceed 40% of the original speed, and the slewing and luffing mechanism speed shall not exceed 30% of the original speed. At the same time, a minimum stable speed threshold is set to avoid the components from swaying due to excessively low speed.

[0034] It should be further explained that, in the specific implementation process, the process of acquiring historical adjustment coefficient data and constructing an adjustment coefficient prediction model, and then using the adjustment coefficient prediction model to predict the adjustment coefficient of hazardous process unit nodes, includes: The adjustment coefficient refers to the degree of change in the probability of an accident at a hazardous process unit node after adjusting the hoisting route and hoisting speed. Specifically, it is the ratio of the predicted accident probability after adjustment to the predicted accident probability before adjustment. The value range of the adjustment coefficient is 0-1. Factors affecting the adjustment coefficient include: the adjusted hoisting speed, the average safety distance, the complexity of the action, the change in the windward area, and the operational response redundancy time. The adjusted hoisting speed is determined by the process described above; The average safe distance refers to the average safe distance of the adjusted hoisting route. It is calculated by taking the average of the minimum safe distances between the component and all surrounding obstacles during the operation of the component along the route through the replanned hoisting route. Motion complexity refers to the motion complexity of the adjusted hoisting route. It is obtained by statistically analyzing the number of combined actions (slewing, luffing, and hoisting) and the number of start-stop operations that the crane needs to perform in the adjusted hoisting route. Change in windward area refers to the change in windward area after the route is adjusted. Based on the adjusted hoisting route, combined with the component's posture and real-time wind direction, the average windward area of ​​the component during its movement along the route is calculated. Operational response redundancy time refers to the operational response redundancy time of the operator. Based on the adjusted hoisting speed, it is calculated to determine the longest response time for the operator to deal with emergencies during the movement of the component along the route. Obtain historical adjustment coefficient data for a single hazardous process unit node in different lifting tasks. The historical adjustment coefficient data includes the adjusted lifting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and historical adjustment coefficient of the single hazardous process unit node in different lifting tasks. Based on the adjusted hoisting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and corresponding historical adjustment coefficients of the corresponding dangerous process unit nodes in different historical adjustment coefficient data, an adjustment coefficient prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time in the different historical adjustment coefficient data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical adjustment coefficients in the second training set are used as the output data of the second convolutional neural network. The second convolutional neural network is a one-dimensional convolutional neural network adapted to one-dimensional structured feature input. The specific network structure is set as follows: the network consists of five layers: an input layer, a convolutional feature extraction layer, a pooling layer, a fully connected layer, and an output layer. The input layer has 5 neurons, matching the dimensions of the 5 input feature parameters. The input data is a feature vector normalized to the [0, 1] interval by min-max. The convolutional feature extraction layer consists of two consecutive one-dimensional convolutional layers. The first convolutional layer has 24 filters, a kernel size of 2, a stride of 1, and equal-length padding. The second convolutional layer has 12 filters. The convolutional kernel size is 2, the stride is 1, and the padding method is equal-length padding. The activation function of both convolutional layers is the ReLU function, which is used to extract the nonlinear influence of different features on the adjustment coefficient. The pooling layer is set as a one-dimensional global average pooling layer to reduce the dimensionality of the features extracted by convolution and reduce the number of model parameters. There are 2 fully connected layers. The number of neurons in the first fully connected layer is 24, and the number of neurons in the second fully connected layer is 6. The activation function of both fully connected layers is the ReLU function. The output layer has 1 neuron and uses the Sigmoid function to map the output value to the interval [0, 1], corresponding to the predicted adjustment coefficient. The specific hyperparameter settings for model training are as follows: the optimizer uses the Adam adaptive optimizer, the initial learning rate is set to 0.0005, the learning rate decay strategy is to decay the learning rate to 0.92 every 8 epochs, the batch size is set to 16, the maximum training epochs are set to 150, and the loss function is the mean squared error (MSE) function, which is adapted to the regression prediction requirements of adjusting the coefficients. The following measures were taken to prevent overfitting during training: Dropout layers were added after each of the two fully connected layers, with a random dropout rate of 0.25 to avoid model overfitting. L2 regularization was also added with a regularization coefficient of 0.0005 to constrain the distribution range of network weight parameters. An early stopping mechanism was enabled during training: when the loss value of the validation set did not decrease for 10 consecutive epochs, training was terminated early to retain the model weights with the lowest loss on the validation set and avoid overfitting. The training of the model is terminated when any of the following conditions are met: first, the preset maximum number of iterations (150 epochs) is reached; second, the early stopping mechanism is triggered and the loss on the validation set does not decrease for 10 consecutive epochs; and third, the loss on the training set decreases to below 0.003 and tends to stabilize. The preset second test error threshold is set to 0.02. This threshold is determined based on the control accuracy requirements of hoisting parameter adjustment. When the model's mean absolute error (MAE) on the test set is less than or equal to 0.02, it means that the prediction deviation of the adjustment coefficient does not exceed 2%, which can accurately predict the risk reduction effect of control measures and meet the accuracy requirements of on-site solution decision-making. If the model verification error exceeds this threshold, it is retrained by expanding the simulation and on-site test samples and optimizing the network hyperparameters until the error threshold requirement is met. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the adjustment coefficient prediction model. The adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time of each hazardous process unit node are input into the adjustment coefficient prediction model to obtain the predicted adjustment coefficient for each hazardous process unit node. In an embodiment of the present invention, the predicted adjustment coefficients of all hazardous process unit nodes are obtained through the adjustment coefficient prediction model. The predicted adjustment coefficients are related to the adjusted hoisting speed, average safety distance, action complexity, windward area change, and operation response redundancy time. The adjusted hoisting speed directly affects the magnitude of the prediction adjustment coefficient. The lower the adjusted hoisting speed, the better the stability of the component, the greater the reduction in the probability of accidents, and the smaller the prediction adjustment coefficient. Therefore, the adjusted hoisting speed and the adjustment coefficient are positively correlated. The size of the mean safety distance directly affects the size of the prediction adjustment coefficient. The larger the mean safety distance, the lower the risk of the component colliding with the obstacle, the greater the reduction in the probability of the accident, and the smaller the prediction adjustment coefficient. Therefore, the mean safety distance and the adjustment coefficient are negatively correlated. The complexity of the motion directly affects the magnitude of the prediction adjustment coefficient. The lower the motion complexity, the simpler and smoother the hoisting operation, the lower the probability of component swaying and operational errors, and the smaller the prediction adjustment coefficient. Therefore, motion complexity and adjustment coefficient are positively correlated. The magnitude of the change in windward area directly affects the magnitude of the prediction adjustment coefficient. The smaller the change in windward area, the smaller the sway amplitude of the component affected by the wind, the lower the probability of an accident, and the smaller the prediction adjustment coefficient. Therefore, the change in windward area is positively correlated with the adjustment coefficient. The length of the operational response redundancy time directly affects the magnitude of the prediction adjustment coefficient. The longer the operational response redundancy time, the more time the operators have to deal with sudden risks, the lower the probability of accidents caused by operational errors, and the smaller the prediction adjustment coefficient. Therefore, the operational response redundancy time and the adjustment coefficient are negatively correlated.

[0035] It should be further explained that, in the specific implementation process, the process of calculating the reduction in the probability of accident occurrence based on the prediction adjustment coefficient and deciding on the corresponding hoisting route and hoisting speed includes: After generating multiple alternative hoisting routes and hoisting speed adjustment schemes for hazardous process unit nodes, for each scheme, the prediction adjustment coefficient corresponding to the scheme is predicted by the adjustment coefficient prediction model. The predicted accident probability before adjustment is multiplied by the prediction adjustment coefficient to calculate the predicted accident probability value after the scheme is adjusted. Then, the predicted accident probability after each set of schemes is compared with the preset probability threshold. Invalid schemes that are still greater than the accident probability threshold after adjustment are eliminated. For valid schemes whose predicted probability after adjustment is less than or equal to the accident probability threshold, the scheme with the smallest adjustment coefficient and the largest reduction in accident probability is selected first. Finally, the optimal hoisting route and hoisting speed scheme is determined, and an execution command is issued to the crane control system to complete the hoisting operation of this node according to the determined scheme.

[0036] It should be further explained that, in the specific implementation process, the process of calculating and determining whether all hoisting operations can be completed within the specified time based on the determined hoisting route and hoisting speed includes: After completing the hoisting operation of the dangerous process unit node according to the determined optimal adjustment plan, the actual increase in hoisting time due to the adjustment of the route and speed will be calculated based on the adjusted hoisting route length and hoisting speed. Then, the number of unfinished process nodes of the prefabricated components hoisted this time, as well as the total number of prefabricated components to be hoisted in the entire hoisting operation plan, the standard hoisting time of each component, and the hoisting completion time set in advance for the project will be calculated. Based on this, the current cumulative hoisting time, the remaining available hoisting time, and the total standard time of the remaining hoisting operations will be calculated to determine whether the hoisting operation can be completed within the specified time according to the current hoisting pace. If it is found that the cumulative increase in hoisting time is within the redundancy range of the remaining available time, and all hoisting operations can be completed within the specified time, the system will not adjust the hoisting plans for other components or other nodes, and will only continue to perform normal hoisting operations in accordance with risk control requirements. If the accumulated increase in time exceeds the redundancy limit, and the current pace makes it impossible to complete all hoisting operations within the specified time, the system will, while ensuring safety, optimize and adjust the speed of subsequent hoisting nodes for other precast components. The specific process is as follows: First, identify the process unit nodes among the components to be hoisted that have consistently had a comprehensive risk score below the first risk threshold and an accident probability far below the accident probability threshold during the historical hoisting of similar components. For these process unit nodes, under the premise of not exceeding the maximum hoisting speed specified in the hoisting safety regulations and not changing the original safe hoisting route, reasonably optimize the hoisting speed of the stable operation section. The optimized speed should not exceed 80% of the maximum speed allowed by the regulations. At the same time, ensure that the predicted accident probability of the optimized node is still far below the accident probability threshold and will not cause new safety risks. By optimizing the speed of multiple low-risk nodes, the hoisting time of a single component is reduced to make up for the increased time of the previously dangerous nodes. For each node after speed adjustment, the comprehensive risk score and predicted accident probability will be recalculated to ensure that the accident probability of all nodes does not exceed the accident probability threshold. If no adjustable node that meets the safety requirements can be found, a schedule warning will be immediately sent to the project management personnel, indicating that additional hoisting equipment and adjustments to the hoisting operation plan are needed to avoid exceeding the safety red line in order to meet the schedule.

[0037] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0038] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. 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 safety risk assessment in the hoisting process of precast components, characterized in that, Includes the following steps: S1: Real-time acquisition of component parameters, operation parameters and environmental parameters of prefabricated components at each process unit node, and normalization of the parameters; S2: Generate component risk scores, operational risk scores, and environmental risk scores for each process unit node based on the processed parameters. Then, calculate the comprehensive risk score for each process unit node based on the component risk scores, operational risk scores, and environmental risk scores. Finally, calculate the rate of change of the comprehensive risk score based on the comprehensive risk score. S3: Classify each process unit node according to the comprehensive risk score and the rate of change of the comprehensive risk score, and take corresponding measures for different levels; S4: Obtain historical accident probability data and construct an accident probability prediction model. Use the accident probability prediction model to predict the accident probability of each high-risk node, including: inputting the current comprehensive risk score, current comprehensive score change rate, extreme value of sub-risk score, operator experience index, lifting equipment aging index and historical accident frequency of each high-risk node into the accident probability prediction model to obtain the predicted accident probability of each high-risk node. Mark the dangerous process unit node according to the predicted accident probability and readjust the hoisting route and hoisting speed of the dangerous process unit node. S5: Obtain historical adjustment coefficient data and construct an adjustment coefficient prediction model. Use the adjustment coefficient prediction model to predict the predicted adjustment coefficient of hazardous process unit nodes. This includes: inputting the adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time of each hazardous process unit node into the adjustment coefficient prediction model to obtain the predicted adjustment coefficient of each hazardous process unit node. Calculate the reduction in accident probability based on the predicted adjustment coefficient and decide on the corresponding hoisting route and hoisting speed. The adjustment coefficient refers to the degree of change in the accident probability of a hazardous process unit node after adjusting the hoisting route and hoisting speed; specifically, it is the ratio of the predicted accident probability after adjustment to the predicted accident probability before adjustment. S6: Calculate and determine whether all hoisting operations can be completed within the specified hoisting time based on the determined hoisting route and hoisting speed. Adjust the hoisting speed of other process unit nodes according to the number of prefabricated components to be hoisted and the specified hoisting time.

2. The method for safety risk assessment of precast component hoisting process according to claim 1, characterized in that, The process unit nodes of the entire precast component hoisting process include the pre-hoisting preparation node, hoisting operation node, aerial transfer node, positioning and adjustment node, and installation and fixing node. Among them, the nodes in which the precast component is in the hoisting state are the hoisting operation node, aerial transfer node, and positioning and adjustment node. The relevant parameters of these three nodes are collected, including component parameters, operation parameters, and environmental parameters. The component parameters include the weight, dimensions, center of gravity offset, and lifting point position deviation of the prefabricated component; the operation parameters include the lifting sling angle deviation, lifting acceleration, and component swing amplitude; and the environmental parameters include wind speed, visibility, and obstacle density. The above parameters are normalized.

3. The method for safety risk assessment of precast component hoisting process according to claim 2, characterized in that, The process of generating component risk scores, operational risk scores, and environmental risk scores for each process unit node based on the processed parameters includes: Component risk scoring : In the formula, , , and The weighting coefficients are derived from historical data training; M is the normalized weight, C is the normalized dimension, Z is the normalized center of gravity offset, and D is the normalized lifting point position deviation. Operational risk score : In the formula, , and , where is the weighting coefficient, derived from historical data training; J is the normalized lifting sling angle deviation, A is the normalized lifting acceleration, and B is the normalized component swing amplitude. Environmental risk scoring : In the formula, , and , where is the weighting coefficient, obtained from training based on historical data; V is the normalized wind speed, N is the normalized visibility, and Q is the normalized obstacle density.

4. The method for safety risk assessment of precast component hoisting process according to claim 3, characterized in that, The process of calculating the comprehensive risk score for each process unit node based on component risk score, operational risk score, and environmental risk score includes: Overall risk score P: In the formula, , and These are weighting coefficients, obtained through training based on historical data; The process of calculating the rate of change of the comprehensive risk score through the comprehensive risk score includes: The assessment cycle is set, and the process unit node corresponding to the precast component currently being hoisted is taken as the object. The comprehensive risk score value of five consecutive assessment cycles within the node is continuously collected according to the assessment cycle. First, the difference in comprehensive risk score between two adjacent assessment cycles is calculated. The score of the previous assessment cycle is subtracted from the score of the later assessment cycle to obtain the score change value within each assessment cycle interval. Then, the score change values ​​of all assessment cycle intervals are added together to obtain the total change in score within the node. Finally, the total change is divided by the total time span of these five assessment cycles to obtain the comprehensive risk score change rate per unit time.

5. The method for safety risk assessment of precast component hoisting process according to claim 4, characterized in that, The process of classifying each process unit node based on the comprehensive risk score and the rate of change of the comprehensive risk score, and taking corresponding measures for different levels, includes: Based on the rate of change of the comprehensive risk score, the risk level of each process unit node is classified. The specific classification process is as follows: a first risk threshold and a second risk threshold are set in advance, and a risk change rate warning threshold is set at the same time. First, the comprehensive risk score of the node is compared with the two risk thresholds. When the comprehensive risk score is less than the first risk threshold, regardless of the rate of change of the score, the node is marked as a low-risk node and routine control measures are implemented, namely, continuously collecting and monitoring risk data, providing routine safety reminders to operators, and not interfering with the normal hoisting operation process. When the comprehensive risk score is greater than or equal to the first risk threshold and less than or equal to the second risk threshold, the change rate of the comprehensive risk score is further examined. If the change rate is less than the risk change rate warning threshold, the node is marked as a medium-risk node and enhanced control measures are implemented, namely, increasing the frequency of risk data collection, providing real-time warnings for risk indicators exceeding the standard, arranging on-site safety officers to supervise, and reminding workers to operate in a standardized manner. If the change rate is greater than or equal to the risk change rate warning threshold, the node is directly upgraded and marked as a high-risk node. When the comprehensive risk score exceeds the second risk threshold, regardless of the rate of change of the score, the node is directly marked as a high-risk node, and emergency control measures are implemented. The crane operator is immediately given an order to suspend the lifting operation, the equipment operation authority is locked, the specific indicators that caused the risk to exceed the standard are identified at the same time, and special rectification requirements are pushed to the on-site management personnel. After the rectification is completed, the risk score is re-verified, and the lifting operation can only be resumed after the risk is confirmed to have dropped to a safe range.

6. The method for safety risk assessment of precast component hoisting process according to claim 5, characterized in that, The process of acquiring historical accident probability data and constructing an accident probability prediction model, and then using this model to predict the accident probability at each high-risk node, includes: The probability of an accident occurring refers to the probability value of a safety accident occurring at a high-risk node; Factors influencing the probability of accidents include: current overall risk score, current overall score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and historical accident frequency. Obtain historical accident probability data for a single high-risk node in different lifting tasks. The historical accident probability data includes the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, historical accident frequency, and historical accident probability of the single high-risk node in different lifting tasks. Based on the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, frequency of historical accidents and corresponding historical accident probabilities of high-risk nodes in different historical accident probability data, an accident probability prediction set is generated and divided into the first training set and the first test set. Construct a first convolutional neural network, taking the current comprehensive risk score, current comprehensive score change rate, extreme values ​​of sub-risk scores, operator experience index, lifting equipment aging index, and frequency of historical accidents from different historical accident occurrence probability data in the first training set as the input data of the first convolutional neural network, and taking the corresponding historical accident occurrence probability in the first training set as the output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network whose output is less than or equal to the preset first test error threshold is used as the accident occurrence probability prediction model.

7. The method for safety risk assessment of precast component hoisting process according to claim 6, characterized in that, The process of marking hazardous process unit nodes based on the predicted probability of accident occurrence includes: Pre-set an accident probability threshold. The predicted accident probability is obtained once in each evaluation cycle. The predicted accident probability output by high-risk nodes in real time is continuously compared with the accident probability threshold. When the predicted accident probability at a certain moment is less than or equal to the accident probability threshold, the node is determined to be in a safe and controllable state. No additional intervention is required, and continuous monitoring is sufficient. When the predicted probability of an accident at a certain moment exceeds the accident probability threshold, the process unit node is immediately marked as a dangerous process unit node, and an audible and visual warning is triggered simultaneously. Risk warning information is pushed to on-site operators and managers, and the subsequent hoisting route and speed adjustment process is initiated to reduce the probability of an accident through proactive control measures. The process of readjusting the hoisting route and speed for hazardous process unit nodes includes: When a process unit node is marked as a hazardous process unit node, the current hoisting operation will be immediately suspended. At the same time, based on the current risk factors, component condition, and working environment, the hoisting route and hoisting speed for that node will be readjusted. The specific process is as follows: In terms of route planning, the complete three-dimensional scene data of the current hoisting operation space is first obtained through three-dimensional laser scanning. Combined with the real-time position of the component, the target positioning position, the distribution of identified obstacles, and wind speed and direction data, the hoisting route is replanned with the core principles of "minimizing the swing amplitude of the component, maximizing the safe distance from obstacles, and avoiding areas affected by strong winds". For aerial transfer nodes, the shortest straight transfer route is prioritized to reduce the combined actions of crane rotation and luffing. At the same time, it is ensured that the safe distance between the component and surrounding obstacles is not less than 2 meters. For positioning and adjustment nodes, the vertical slow positioning route is prioritized to reduce horizontal adjustment actions and avoid collisions between the component and installed components or edge protection. Regarding hoisting speed adjustment, the speed adjustment coefficient is first calculated and determined based on the real-time predicted probability of accidents. The calculation method for the speed adjustment coefficient is as follows: Step 1: Calculate the base coefficient The calculation formula is: = / ； In the formula, The preset accident probability threshold, The base coefficient is used to predict the probability of an accident occurring in real time for the current node. The range of values ​​is 0 < ≤1; Step 2: Determine if the core risk factors exceed the standard. Perform quantitative judgment based on the normalized parameters collected in real time at the current node: when the normalized lifting acceleration A ≥ 0.6, it is determined that the lifting acceleration exceeds the standard; when the normalized component swing amplitude B ≥ 0.7, it is determined that the component swing amplitude is large. Step 3: Based on the results of the exceedance judgment, the basic coefficient is quantitatively corrected. The specific mathematical model for correction is as follows: If the problem is solely due to excessive lifting acceleration, calculate the acceleration correction factor. The calculation formula is: =1 0.4×A； In the formula, A is the normalized lifting acceleration collected in real time at the current node. The value range is (0, 1]; the corrected speed adjustment coefficient K= × ; If the only cause is a large component sway amplitude, calculate the sway correction factor. The calculation formula is: =1 0.3×B； In the formula, B is the normalized component swing amplitude collected in real time by the current node. The value range is (0, 1]; Corrected speed adjustment coefficient K= × ; If both the lifting acceleration and component sway amplitude are determined to be excessive, a comprehensive correction factor k is calculated. The formula is: k = × ; Corrected speed adjustment coefficient K= ×k; Step 4: Boundary constraints for speed adjustment coefficient. The corrected speed adjustment coefficient K must satisfy 0.2≤K≤0.5; The final hoisting speed is equal to the original hoisting speed multiplied by the corrected speed adjustment coefficient K. The adjusted hoisting mechanism speed shall not exceed 40% of the original speed, and the slewing and luffing mechanism speed shall not exceed 30% of the original speed. At the same time, a minimum stable speed threshold is set to avoid the components from swaying due to excessively low speed.

8. The method for safety risk assessment of precast component hoisting process according to claim 7, characterized in that, The process of acquiring historical adjustment coefficient data and constructing an adjustment coefficient prediction model, and then using this model to predict the adjustment coefficients of hazardous process unit nodes, includes: The adjustment factor ranges from 0 to 1; Factors affecting the adjustment coefficient include: the adjusted hoisting speed, the average safety distance, the complexity of the action, the change in the windward area, and the operational response redundancy time. Acquire historical adjustment coefficient data for a single hazardous process unit node in different lifting tasks. The historical adjustment coefficient data includes the adjusted lifting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and historical adjustment coefficient of the single hazardous process unit node in different lifting tasks. Based on the adjusted hoisting speed, average safety distance, action complexity, windward area change, operation response redundancy time, and corresponding historical adjustment coefficients of the corresponding dangerous process unit nodes in different historical adjustment coefficient data, an adjustment coefficient prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The adjusted hoisting speed, average safety distance, motion complexity, windward area change, and operation response redundancy time in the different historical adjustment coefficient data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical adjustment coefficients in the second training set are used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network whose output is less than or equal to the preset second test error threshold is used as the adjustment coefficient prediction model.

9. The method for safety risk assessment of precast component hoisting process according to claim 8, characterized in that, The process of calculating the reduction in the probability of an accident based on the prediction adjustment coefficient, and then deciding on the corresponding hoisting route and hoisting speed, includes: After generating multiple alternative hoisting routes and hoisting speed adjustment schemes for hazardous process unit nodes, for each adjustment scheme, the prediction adjustment coefficient corresponding to the adjustment scheme is predicted by the adjustment coefficient prediction model. The predicted accident probability before adjustment is multiplied by the prediction adjustment coefficient to calculate the predicted accident probability value after the scheme is adjusted. Then, the predicted accident probability after each set of adjustment schemes is compared with the preset probability threshold. Invalid schemes that are still greater than the accident probability threshold after adjustment are eliminated. For valid schemes that are less than or equal to the accident probability threshold after adjustment, the scheme with the smallest adjustment coefficient and the largest reduction in accident probability is selected first. Finally, the optimal hoisting route and hoisting speed scheme is determined, and an execution command is issued to the crane control system to complete the hoisting operation of this node according to the determined scheme.

10. The method for safety risk assessment of precast component hoisting process according to claim 9, characterized in that, The process of calculating and determining whether the entire hoisting operation can be completed within the specified time based on the determined hoisting route and hoisting speed includes: After completing the hoisting operation of the dangerous process unit node according to the determined optimal adjustment plan, the actual increase in hoisting time due to the adjustment of the route and speed will be calculated based on the adjusted hoisting route length and hoisting speed. Then, the number of unfinished process nodes of the prefabricated components hoisted this time, as well as the total number of prefabricated components to be hoisted in the entire hoisting operation plan, the standard hoisting time of each component, and the hoisting completion time set in advance for the project will be calculated. Based on this, the current cumulative hoisting time, the remaining available hoisting time, and the total standard time of the remaining hoisting operations will be calculated to determine whether the hoisting operation can be completed within the specified time according to the current hoisting pace. If it is found that the cumulative increase in hoisting time is within the redundancy range of the remaining available time, and all hoisting operations can be completed within the specified time, the system will not adjust the hoisting plans for other components or other nodes, and will only continue to perform normal hoisting operations in accordance with risk control requirements. If the accumulated increase in time exceeds the redundancy limit, and the current pace makes it impossible to complete all hoisting operations within the specified time, the system will, while ensuring safety, optimize and adjust the speed of subsequent hoisting nodes for other precast components. The specific process is as follows: First, identify the process unit nodes among the components to be hoisted that have consistently had a comprehensive risk score below the first risk threshold and an accident probability far below the accident probability threshold during the historical hoisting of similar components. For these process unit nodes, under the premise of not exceeding the maximum hoisting speed specified in the hoisting safety regulations and not changing the original safe hoisting route, reasonably optimize the hoisting speed of the stable operation section. The optimized speed should not exceed 80% of the maximum speed allowed by the regulations. At the same time, ensure that the predicted accident probability of the optimized node is lower than the accident probability threshold and will not cause new safety risks. By optimizing the speed of multiple low-risk nodes, the hoisting time of a single component is reduced to make up for the increased time of the previously dangerous nodes. For each node after speed adjustment, the comprehensive risk score and predicted accident probability will be recalculated to ensure that the accident probability of all nodes does not exceed the accident probability threshold. If no adjustable node that meets the safety requirements can be found, a schedule warning will be immediately sent to the project management personnel, indicating that it is necessary to add hoisting equipment and adjust the hoisting operation plan to avoid exceeding the safety red line in order to meet the schedule.

Images