Aerial work anti-falling pipe control device and method

Abstract

The application provides a high-altitude operation anti-falling pipe control device and method, and belongs to the technical field of high-altitude operation safety protection. The technical scheme is as follows: multi-source anti-falling state data is acquired and preprocessed to generate preprocessed anti-falling state data; the current effective hanging point is determined and a dynamic falling risk coefficient is constructed according to the preprocessed anti-falling state data; the falling margin state is generated by combining the dynamic falling risk coefficient and the preprocessed anti-falling state data, and the pre-locking demand state is generated when the preset risk condition is reached; the heterogeneous back-check data is called to construct the pre-execution back-check state, and the real falling confirmation state, the high-risk disturbance state or the back-check abnormal state is generated; the graded anti-falling control action is output according to the above, and the post-execution back-check is performed after the action is executed, and the state identifier is written into the anti-falling event record. The application has the beneficial effects that the real falling, the operation disturbance and the data abnormality can be distinguished, the risk of false locking is reduced, and the anti-falling judgment accuracy, the action recoverability and the event traceability are improved.

Description

Technical Field

[0001] This invention relates to the field of safety protection technology for high-altitude operations, and in particular to a high-altitude operation fall prevention control device and method. Background Technology

[0002] With the increasing complexity of work scenarios such as building construction, steel structure installation, bridge maintenance, power operation and maintenance, and high-altitude equipment maintenance, the safety management of high-altitude operations is gradually evolving from relying solely on mechanical protective devices such as safety belts, safety ropes, and fall arresters to intelligent fall arrest systems that integrate sensor detection, edge computing, status recognition, and platform-based supervision. Existing intelligent fall arrest systems typically monitor the hook-locking status, movement status, and abnormal falling trends of workers through hook locking detection, personnel height detection, posture detection, rope tension detection, or fall arrester action feedback. When an anomaly is detected, alarms, locking, or platform prompts are output to improve the safety response capability during high-altitude operations.

[0003] However, existing fall protection and control technologies for high-altitude operations still have significant shortcomings. First, current solutions primarily rely on whether the hook is locked, whether the worker's height has decreased, or whether the safety rope tension has changed abruptly. They lack comprehensive modeling of the load-bearing relationship between the current effective anchor point, rope length, worker connection point height, and equipment extension margin, making it difficult to accurately quantify the actual fall margin for workers at different anchor point heights and under rope slack conditions. Second, existing posture detection is typically used only as an independent alarm condition, failing to form a coordinated judgment with the fall risk coefficient, slack margin, and anchor point height. This leads to normal disturbances such as bending over, stepping, hook switching, and brief squatting being easily misjudged as fall risks, and there may be a lack of early intervention when posture instability occurs simultaneously with anchor point descent. Third, existing active fall protection controls often directly alarm or lock after a single threshold is triggered, lacking a pre-execution confirmation mechanism based on heterogeneous backtesting data. This makes it difficult to distinguish between actual falls, high-risk disturbances, and abnormal data states, easily resulting in false or missed lockouts. Fourth, existing technologies are insufficient for verifying the effects of actions and tracing events after execution. They cannot form a closed-loop record of whether actions meet standards, whether equipment configuration needs to be updated, and whether abnormal events recur. This makes it difficult to continuously adjust fall protection parameters based on the actual work area, attachment point type, and equipment status. Summary of the Invention

[0004] The purpose of this invention is to provide a high-altitude operation fall prevention and control device and method that can distinguish between real falls, operational disturbances, and data anomalies, reduce the risk of false locking, and improve the accuracy of fall prevention judgment, the recoverability of actions, and the traceability of events.

[0005] This invention is achieved through the following measures: 1. A method for preventing falls during high-altitude operations, characterized by comprising the following steps: S1. Acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data. S2. Determine the current effective attachment point based on the preprocessed fall protection status data, and construct a dynamic fall risk coefficient based on the current effective attachment point; S3. Generate a fall margin state based on the dynamic fall risk coefficient and the pre-processed fall protection status data, and generate a pre-lockdown requirement state when the fall margin state reaches the preset risk condition. S4. When the pre-lock-up requirement state is generated, heterogeneous back-check data is called from the pre-processed fall protection state data to construct the pre-execution back-check state, and one of the following is generated based on the pre-execution back-check state: real fall confirmation state, high-risk disturbance state, and back-check abnormal state. S5. Based on the fall margin status and the pre-execution check status, output the corresponding graded fall protection control action; S6. After the graded fall protection control action is executed, fall protection status data is collected again and a post-execution check is performed. A status identifier is generated based on the post-execution check result and the status identifier is written into the fall protection event record.

[0006] The invention also has the following specific features: Step S1 includes: Acquire multi-source fall protection status data corresponding to workers, fall protection equipment, candidate attachment points, and work areas; The time reference of the multi-source fall protection status data is unified, and the data is divided into continuous data processing windows according to a preset sampling period; Based on the data attributes, the multi-source fall protection status data is divided into pre-calibration data, quasi-static configuration data, and dynamic fall protection status data. The pre-calibration data and quasi-static configuration data serve as spatial reference data in the current effective attachment point determination process, while the dynamic fall protection status data is used to generate the corresponding window feature set. Perform data quality assessment on the window feature set, and add window data anomaly markers based on the assessment results; The precalibrated data, quasi-static configuration data, window feature set, and corresponding window data anomaly markers are output as the preprocessed fall protection status data.

[0007] Step S2 includes: Extract candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, real-time rope extension length of the fall arrester, worker connection point height, and window data anomaly markers from the preprocessed fall arrest status data; Based on the candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, and window data anomaly markers, the candidate attachment points are screened for validity. The current valid attachment point or no valid attachment point status is determined by combining the real-time rope extension length of the fall arrester with the spatial distance between the worker's connection point and the candidate attachment point. Based on the current effective anchor points, determine the slack margin between the worker's connection point and the current effective anchor points, the relationship between the anchor point heights, the worker's vertical movement status, and the configuration parameters of the fall protection equipment; Based on the slack margin, the relationship between the hanging point heights, the vertical movement state of the personnel, and the configuration parameters of the fall protection equipment, the predicted potential fall displacement and the effective load-bearing rope length are determined, and a dynamic fall risk coefficient is constructed based on the predicted potential fall displacement and the effective load-bearing rope length. The dynamic fall risk coefficient of the current data processing window is compared with the dynamic fall risk coefficient of the previous data processing window to obtain the rate of change of the dynamic fall risk coefficient. The dynamic fall risk coefficient and the rate of change of the dynamic fall risk coefficient are then output to subsequent steps.

[0008] Step S3 includes: The personnel posture data is retrieved from the preprocessed fall protection status data, and the IMU posture state is generated based on the personnel posture data. The IMU posture state includes the torso tilt angle, torso tilt angular velocity, vertical acceleration rate of change, and dynamic stability margin. An instability trend state is generated based on at least one of the following: trunk tilt angle, trunk tilt angular velocity, rate of change of vertical acceleration, and dynamic stability margin. The deviation of the predicted center of mass position from the foot support domain boundary is determined based on the personnel posture data, and a dynamic stability margin over-limit ratio is generated based on the deviation. The second coefficient threshold is corrected based on the dynamic stability margin over-limit ratio to obtain the corrected second coefficient threshold. The IMU attitude state, instability trend state, dynamic stability margin overshoot ratio, and the corrected second coefficient threshold are used to generate the fall margin state.

[0009] Step S3 also includes: Based on the dynamic fall risk coefficient, the rate of change of the dynamic fall risk coefficient, the slack margin, the current effective attachment point, the state without an effective attachment point, the IMU attitude state, the instability trend state, and the window data anomaly marker, a normal fall margin state or a fall margin state of concern is generated. When at least one of the dynamic fall risk coefficient, dynamic fall risk coefficient change rate, slack margin, no effective anchor point state, instability trend state and dynamic stability margin exceeding limit ratio meets the preset risk conditions, a pre-lock-up requirement state is generated. A sliding time window voting confirmation is performed on the data processing windows that meet the preset risk conditions. When the proportion of the number of data processing windows that meet the preset risk conditions reaches a preset proportion threshold, the fall margin state is confirmed to have met the preset risk conditions. When generating the pre-locked-out requirement state, a corresponding trigger type identifier is generated according to the trigger path that triggers the preset risk condition, and the pre-locked-out requirement state and the trigger type identifier are output to the pre-execution backcheck.

[0010] In step S4, when the pre-lockout requirement state is generated, heterogeneous back-check data is retrieved from the preprocessed fall protection state data to construct the pre-execution back-check state, including: When the pre-lockout requirement state is generated, the pre-execution back-check process is started, and heterogeneous back-check data is called from the pre-processed fall protection status data. The heterogeneous backtesting data includes at least two of the following: the fall arrester rope release speed, the worker connection point height descent rate, the safety rope tension abrupt change characteristics, the redundant hook force abrupt change characteristics, the rope impact acceleration characteristics, and the fall arrester execution unit status feedback. The pre-execution back-check channel is selected based on the trigger type identifier, or the back-check weight of the heterogeneous back-check data is adjusted. The back-check weight is determined by adjusting the preset threshold of the corresponding back-check condition or adjusting the contribution coefficient of the corresponding pre-execution back-check channel. Extract the return inspection features corresponding to the heterogeneous return inspection data within the preset return inspection time window, and construct the pre-execution return inspection state based on the return inspection features; For high-frequency data, the back-check feature includes at least one of window peak value, window maximum change rate, and duration of mutation feature; for low-frequency data or event-driven data, the back-check feature includes the most recent valid state value or event occurrence timestamp within the preset back-check time window.

[0011] Step S4 generates one of the following states based on the pre-execution re-check status: actual fall confirmation status, high-risk disturbance status, and re-check anomaly status: The validity of the data channels involved in the pre-execution review is assessed to determine the number of valid review channels. The number of channels that meet the corresponding return inspection conditions is determined based on whether each valid return inspection channel meets the return inspection conditions. Based on whether the number of valid return check channels has reached the minimum number of valid return check channels, and the relationship between the number of return check conditions met and the minimum number of conditions met for actual fall confirmation and the minimum number of conditions met for high-risk disturbance state, a real fall confirmation state, a high-risk disturbance state, or a return check abnormal state is generated. Specifically, when the pre-execution re-check status meets the actual fall confirmation condition, the actual fall confirmation status is generated; when the pre-execution re-check status meets the high-risk disturbance condition but does not meet the actual fall confirmation condition, the high-risk disturbance status is generated; when the number of effective re-check channels is insufficient, or when the pre-execution re-check status does not meet the actual fall confirmation condition and the high-risk disturbance condition, the re-check abnormal status is generated. The abnormal return check status includes insufficient return check data status or return check failure status, and is accompanied by a sub-status identifier to distinguish between the insufficient return check data status and the return check failure status.

[0012] Step S5 includes: Receive the pre-execution check status and the fall margin status, and determine the graded fall protection control action based on the correspondence between the pre-execution check status and the fall margin status; When the pre-execution check status includes a confirmed actual fall status, a safety locking action is output. When the pre-execution check state includes a high-risk disturbance state, a reversible pre-tightening action, a resistance increase action, or a low-level early warning action is output. When the pre-execution check status includes a check abnormal status, an abnormal handling action, a reversible pre-tightening action, or a low-level warning action is output according to the sub-status identifier of the check abnormal status. When the pre-execution review status does not include the actual fall confirmation status, high-risk disturbance status, and review abnormal status, and the fall margin status is the fall margin of concern status, a low-level warning action is output or the current fall prevention execution unit status is maintained. When outputting the graded fall protection control action, an action control instruction is generated. The action control instruction includes the action type, target fall protection execution unit, trigger status, trigger type identifier, execution hold condition, and release condition.

[0013] Step S6 includes: The output time of the action control command is used as the reference time for the start of the post-execution review. The fall protection status data is re-acquired within the preset post-execution review time window, and the post-execution review result is generated based on the re-acquired fall protection status data. Based on the post-execution check results, determine whether the graded fall protection control action has reached the corresponding target state, and generate the corresponding status identifier; Based on the post-execution check results, determine whether to generate an equipment configuration update identifier or parameter offset update information; At least one of the graded fall protection control actions, action control commands, post-execution check results, status identifiers, equipment configuration update identifiers, and parameter offset update information is associated with the current effective anchor point, dynamic fall risk coefficient, fall margin status, pre-lockout requirement status, trigger type identifier, and pre-execution check status information and written into the fall protection event record. A high-altitude operation fall prevention and control device, characterized in that it includes a controller, the controller being equipped with a data preprocessing module, an effective attachment point determination module, a fall margin generation module, a pre-execution check module, a graded control module, and a post-execution check module; The data preprocessing module is used to acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data. The effective attachment point determination module is used to determine the current effective attachment point based on the preprocessed fall protection status data, and to construct a dynamic fall risk coefficient based on the current effective attachment point. The fall margin generation module is used to generate a fall margin state based on the dynamic fall risk coefficient and the pre-processed fall protection status data, and to generate a pre-lockdown requirement state when the fall margin state reaches a preset risk condition. The pre-execution back-check module is used to call heterogeneous back-check data from the preprocessed fall protection status data when the pre-lock-up requirement state is generated, construct the pre-execution back-check state, and generate one of the following states based on the pre-execution back-check state: real fall confirmation state, high-risk disturbance state, and back-check abnormal state. The graded control module is used to output corresponding graded fall protection control actions based on the fall margin status and the pre-execution check status. The post-execution check module is used to re-collect fall protection status data and perform post-execution check after the graded fall protection control action is executed, generate a status identifier based on the post-execution check result, and write the status identifier into the fall protection event record.

[0014] The beneficial effects of this invention are: it can distinguish between real falls, operational disturbances and data anomalies, reduce the risk of false locking, and improve the accuracy of fall protection judgment, the recoverability of actions and the traceability of events; This invention forms a complete fall protection management closed loop through multi-source fall protection status data preprocessing, current effective anchor point determination, dynamic fall risk coefficient construction, fall margin status judgment, pre-execution heterogeneous back-check, graded fall protection control action output, and post-execution back-check recording. This solution can heterogeneously confirm the pre-locking requirement status before safety locking actions, reducing false locking caused by single sensor anomalies, normal attitude disturbances, or misjudgments of hook status. Simultaneously, through status identification, fall protection event records, equipment configuration update identification, and parameter bias update information, it enables verification of fall protection action effectiveness and subsequent parameter correction, improving the accuracy, recoverability, and traceability of fall protection management in high-altitude operations. Attached Figure Description

[0015] Figure 1 This is an overall flowchart of the high-altitude operation fall prevention and control method provided in Embodiment 1 of the present invention. Detailed Implementation

[0016] To clearly illustrate the technical features of this solution, the following detailed implementation method will be used to explain the solution.

[0017] Example 1 See Figure 1 This embodiment provides a method for fall prevention and control in high-altitude operations, applicable to high-altitude operation scenarios such as building construction, steel structure installation, bridge maintenance, curtain wall construction, power maintenance, and equipment maintenance at height. The high-altitude operation scenario includes candidate anchor points and a work area for worker movement. In some scenarios, lifeline nodes or fall arrestor anchor points can also be used as the source of candidate anchor point configuration. Candidate anchor points have identifiable locations and heights, which can be determined based on at least one of the following: a preset anchor point coordinate table, a site structural model, a BIM model, positioning data, or anchor point electronic tag data.

[0018] In this embodiment, the worker wears fall protection equipment, which may include a safety belt, hook, fall arrestor, and safety rope. The safety belt forms the worker's attachment point, the hook connects to a candidate attachment point and creates a hook-locked state, and the fall arrestor generates the real-time rope extension length and speed. The fall protection execution unit enters a free rope extension state, a light tension state, a resistance-increasing state, or a safety-locked state according to the graded fall protection control actions. It should be noted that the fall protection equipment is only used to illustrate the data source and action object of this embodiment and does not limit the invention to using a specific structural form of safety belt, hook, fall arrestor, or safety rope.

[0019] Furthermore, the data acquisition unit is used to collect multi-source fall arrest status data. This multi-source fall arrest status data includes at least the worker's connection point height, candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, redundant hook force data, real-time rope extension length of the fall arrester, rope extension speed of the fall arrester, personnel posture data, rope impact acceleration data, fall arrest execution unit status, and work area identification. The data acquisition unit can be installed in fall arrest equipment, work area equipment, or on-site edge computing equipment.

[0020] Furthermore, the edge controller receives multi-source fall protection status data and executes the high-altitude operation fall protection control method of this embodiment. The edge controller preprocesses the multi-source fall protection status data, determines the current effective anchor points, constructs a dynamic fall risk coefficient, generates a fall margin state and a pre-lockdown requirement state, and after generating the pre-lockdown requirement state, calls heterogeneous back-check data to construct a pre-execution back-check state. Then, based on the pre-execution back-check state and the fall margin state, it outputs graded fall protection control actions. The on-site safety management platform can receive fall protection event records, status identifiers, equipment configuration update identifiers, and parameter bias update information uploaded by the edge controller, and can be used to configure and manage fall protection equipment parameters and work area parameters.

[0021] It should be noted that the aforementioned fall protection equipment, data acquisition unit, edge controller, and on-site safety management platform are only used to illustrate the implementation environment and data source of the method in this embodiment, and do not limit the present invention to a specific hardware structure, a specific sensor type, or a specific platform deployment method. The core technology of this embodiment lies in determining the current effective attachment point, constructing a dynamic fall risk coefficient, performing heterogeneous back-checks before execution, outputting graded fall protection control actions, and recording back-checks after execution based on multi-source fall protection status data, rather than limiting the specific mechanical structure of the fall protection equipment.

[0022] Based on the above implementation environment, the high-altitude operation fall prevention and control method in this embodiment specifically includes the following steps: S1. Acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data; specifically including: During high-altitude operations, multi-source fall protection status data is acquired through the fall protection equipment worn by workers, candidate attachment point marking devices in the work area, and detection units on the fall arresters. This multi-source fall protection status data includes at least the worker's connection point height, candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, redundant hook force data, real-time rope extension length of the fall arrester, rope extension speed of the fall arrester, personnel posture data, rope impact acceleration data, fall arrester execution unit status, and work area markings.

[0023] It should be noted that the personnel posture data is preferably collected by an IMU sensor worn by the operator's torso, including at least three-axis acceleration, three-axis angular velocity, and attitude angle, and is used to generate the IMU attitude state in subsequent steps. In this embodiment, the time base unification, data processing window division, window feature set generation, and window data anomaly marking are preferred implementations of preprocessing. Other preprocessing methods that can achieve multi-source data time alignment, feature extraction, and data quality marking are not excluded.

[0024] Furthermore, after receiving multi-source fall protection status data, the edge controller unifies the time base for data from different sources, with different sampling frequencies, and different triggering methods. Specifically, it adds a collection timestamp to each data source and divides the multi-source fall protection status data into consecutive data processing windows according to a preset sampling period. Each data processing window is represented as:

[0025] in, For the first A data processing window, For the first The start time of each data processing window The sampling period is preset. For periodically sampled data, data whose acquisition timestamps are within the data processing window are assigned to the corresponding data processing window; for event-driven data such as hook locking status and fall arrestor status, data are assigned to the corresponding data processing window according to the event occurrence timestamp; when the event occurrence time is located at the boundary of two adjacent data processing windows, data is assigned to the data processing window whose starting time is the boundary according to the left-closed-right-open rule; for state variables that persist across windows, the trailing edge state of the previous data processing window is used as the starting reference state of the next data processing window.

[0026] Furthermore, based on data attributes, multi-source fall protection status data is divided into pre-calibration data, quasi-static configuration data, and dynamic fall protection status data. Among these, candidate attachment point locations, candidate attachment point heights, and work area identifiers are used as pre-calibration data or quasi-static configuration data. The edge controller directly uses these as spatial reference data in the current effective attachment point determination process without performing window aggregation processing. In the absence of detected candidate attachment point configuration update events, the pre-calibration data and quasi-static configuration data maintain their preset values ​​within each data processing window.

[0027] Furthermore, the worker's connection point height, safety rope tension, redundant hook force data, real-time rope extension length of the fall arrester, rope extension speed of the fall arrester, personnel posture data, and rope impact acceleration data are used as dynamic fall arrest status data. The edge controller generates a corresponding window feature set for each type of dynamic fall arrest status data within the same data processing window. The window feature set includes at least stationary feature values ​​and abrupt change feature values. For the first... The first data processing window The window feature set of dynamic fall arrest status data is represented as follows:

[0028] in, For the first The first data processing window Window feature set of dynamic fall arrest status data, For stationary eigenvalues, For the peak value of the window, The maximum rate of change of the window. The duration of the mutation feature.

[0029] It should also be noted that stationary feature values ​​are used as input values ​​in subsequent S2 and S3, while window peak value, maximum rate of change of the window, and duration of abrupt change feature values ​​are used as the basis for judging the pre-execution check conditions in subsequent S4. For continuous slow variables among the worker connection point height, real-time rope extension length of the fall arrester, and safety rope tension value, the mean of effective sampling points within the window is preferred for stationary feature values; when the number of effective sampling points is insufficient, the effective sampling value closest to the data processing window is used; when effective sampling values ​​exist in adjacent time periods and the data change meets the continuity condition, the interpolation result is used. For state variables in the hook locking state and fall arrester execution unit state, the last effective state value within the data processing window is used; if there is no state change event within the data processing window, the following state of the previous data processing window is used.

[0030] Furthermore, the edge controller performs data quality assessment on the window feature set and adds window data anomaly markers based on the assessment results. When the change in the height of the worker's connection point within the same data processing window or between adjacent data processing windows exceeds the upper limit of normal human movement, and the worker's posture data does not show a corresponding posture change, it is marked as height anomaly data; when the hook is locked but the safety rope tension value remains below the effective load-bearing lower limit, it is marked as unloaded hooking state; when the real-time rope extension length of the fall arrester and the spatial distance between the worker's connection point and the candidate hooking point do not meet the preset consistency conditions, it is marked as rope length anomaly data; when the safety rope tension value shows a momentary spike without corresponding rope extension changes, height changes, or posture changes, it is marked as tension anomaly data.

[0031] Furthermore, when a certain type of dynamic fall protection status data within a data processing window meets any of the above-mentioned abnormal conditions, a window data abnormality marker is added to the window feature set corresponding to that data processing window; when the abnormal conditions are not met, the window feature set is marked as normal data. The window data abnormality marker is output along with the corresponding window feature set to subsequent steps and serves as the basis for judging the reliability of data when determining the current valid anchor point in S2, generating the fall margin status in S3, and constructing the pre-execution review status in S4.

[0032] It should also be noted that window data anomaly markers belong to the data layer and are used to characterize the reliability of a certain type of data within a single data processing window; anomaly handling identifiers belong to the system layer status identifiers, and the two are at different levels. The preprocessed fall protection status data includes at least pre-calibration data, quasi-static configuration data, window feature sets of various dynamic fall protection status data, and corresponding window data anomaly markers, thus providing a unified data source for subsequently determining the current effective anchor point, constructing dynamic fall risk coefficients, generating fall margin status, and calling heterogeneous backtesting data.

[0033] S2. Determine the current effective attachment points based on the pre-processed fall protection status data, and construct a dynamic fall risk coefficient based on the current effective attachment points; specifically including: Within the current data processing window, the edge controller retrieves candidate attachment point locations, candidate attachment point heights, hook locking status, safety rope tension values, redundant hook force data, real-time rope extension length of the fall arrester, worker connection point height, worker posture data, and corresponding window data anomaly flags from the pre-processed fall arrest status data.

[0034] It should be noted that the candidate attachment point position and candidate attachment point height are pre-calibrated data or quasi-static configuration data, serving as spatial reference data for determining the current effective attachment point; the hook locking status, safety rope tension value, real-time rope extension length of the fall arrestor, and worker connection point height are judged using stable feature values ​​from the corresponding window feature set; redundant hook force data are retained along with redundant re-inspection attachment point information and used to judge the conditions of sudden changes in redundant hook force during subsequent pre-execution re-inspection.

[0035] Furthermore, the edge controller performs validity screening on the candidate attachment points. When the hook locking state corresponding to a candidate attachment point is in an effective locking state, and the candidate attachment point has an identifiable candidate attachment point location, while the safety rope tension value of the rope path corresponding to the candidate attachment point is within the preset effective load-bearing range, and the corresponding window feature set is not marked by S1 as an unloaded attachment state or other abnormal data affecting the validity of the attachment point, the candidate attachment point is regarded as an attachment point to be confirmed.

[0036] Furthermore, the edge controller determines the three-dimensional spatial distance between the operator's connection point and the location of the anchor point to be confirmed, and performs a spatial distance consistency check with the real-time rope extension length of the fall arrester. When the difference between the three-dimensional spatial distance and the real-time rope extension length of the fall arrester is within the preset length error range, or when the real-time rope extension length of the fall arrester is greater than the three-dimensional spatial distance and can form a slack margin, the anchor point to be confirmed is determined to meet the spatial distance consistency condition; when the real-time rope extension length of the fall arrester is less than the three-dimensional spatial distance and the difference exceeds the preset length error range, the corresponding rope extension length data is marked as rope length abnormal data. Rope length abnormal data belongs to a type of window data abnormality marking in S1, and the anchor point to be confirmed is not determined as the current valid anchor point.

[0037] Furthermore, when only one anchor point meets both the validity screening criteria and the spatial distance consistency criteria, that anchor point is designated as the current valid anchor point. When multiple anchor points simultaneously meet both the validity screening criteria and the spatial distance consistency criteria, each anchor point is used as a temporary valid anchor point. The corresponding risk coefficient estimate is calculated according to the construction method of the dynamic fall risk coefficient. Combined with whether the corresponding rope route is not marked as unloaded by S1 within the continuous data processing window, one anchor point is determined as the current valid anchor point, and the remaining anchor points that meet the criteria are used as redundant check anchor points. When there are no anchor points that meet both the validity screening criteria and the spatial distance consistency criteria, a state of no valid anchor point is generated, and this state is used as a high-risk input in the subsequent fall margin state generation process.

[0038] Furthermore, after determining the current effective anchor point, the edge controller constructs a dynamic fall risk coefficient based on this anchor point. The edge controller calculates the slack margin based on the real-time rope extension length of the fall arrester and the three-dimensional spatial distance between the worker's connection point and the current effective anchor point.

[0039] in, This is the relaxation margin. The real-time rope length of the fall arrestor. This represents the three-dimensional spatial distance between the worker's connection point and the current effective anchor point. The slack allowance characterizes the initial free fall space that may occur when the safety rope is untensioned.

[0040] Furthermore, the edge controller calculates the hanging point height risk item based on the relative relationship between the worker's connection point height and the current effective hanging point height:

[0041] in, For high-risk items, The height of the connection point for the workers. The current effective attachment point height, This is a risk correction coefficient for the hanging point. In the basic implementation, the preset configuration parameters are used. When the work area is configured with a risk level for attachment points, a rope deflection level, or a attachment point type level, Determine based on the equipment configuration table or the work area configuration table.

[0042] Furthermore, the edge controller calculates and predicts potential fall displacement based on slack margin, attachment point height risk, changes in worker connection point height, worker posture data, and preset execution response time.

[0043] in, To predict potential fall displacement, The vertical descent speed at the worker's connection point. For the vertical acceleration of the worker's connection point, The preset execution response time is used. The vertical descent speed and vertical acceleration are not data directly output by S1, but rather derived quantities calculated by the edge controller in S2 based on the changes in the height of the worker's connection point and the worker's posture data in adjacent data processing windows.

[0044] Furthermore, the edge controller determines the effective load-bearing rope length based on the real-time rope extension length of the fall arrester and the configuration parameters of the fall arrest equipment.

[0045] in, To effectively bear the rope length, For buffer stretch compensation, For seat belt extension compensation, To compensate for the braking margin of the fall arrester; and Based on the product technical specifications, type test parameters, or equipment configuration table of the fall protection equipment, the edge controller reads the current valid status of the equipment configuration table before calling the aforementioned compensation parameters; if there is an equipment configuration update identifier generated by the subsequent post-execution check process, it re-determines the configuration table according to the updated equipment configuration table. and Then it will be included in the calculation.

[0046] Furthermore, the edge controller constructs a dynamic fall risk coefficient based on the predicted potential fall displacement and the effective load-bearing rope length:

[0047] in, The dynamic fall risk factor. This is a correction factor used to avoid instability in calculations due to an excessively small denominator. It should also be noted that the edge controller compares the dynamic fall risk coefficient of the current data processing window with the dynamic fall risk coefficient of the previous data processing window to obtain the rate of change of the dynamic fall risk coefficient, and outputs this rate of change along with the dynamic fall risk coefficient to S3. When there is no valid dynamic fall risk coefficient from the previous data processing window, the rate of change of the dynamic fall risk coefficient is set to zero, and this initial rate of change is not used as a trigger for monitoring fall margin status or pre-lockdown requirement status.

[0048] Furthermore, the dynamic fall risk coefficient is obtained by taking the current effective anchor point as the benchmark, and comprehensively considering the slack margin, anchor point height risk, personnel vertical movement trend, fall protection equipment configuration parameters, and fall arrestor braking margin. Therefore, S2 converts the preprocessed fall protection status data output by S1 into a dynamic fall risk coefficient and its rate of change, providing quantitative input for S3 to generate the fall margin state and pre-lockup requirement state.

[0049] S3. Generate a fall margin state based on the dynamic fall risk coefficient and pre-processed fall protection status data, and generate a pre-lockdown requirement state when the fall margin state reaches the preset risk condition; specifically including: Within the current data processing window, the edge controller receives the dynamic fall risk coefficient, dynamic fall risk coefficient change rate, slack margin, current effective anchor point, and no effective anchor point status from the output of S2, and calls the personnel posture data, window feature set, and window data anomaly marker from the preprocessed fall protection status data in S1 to generate the fall margin status.

[0050] The fall margin status characterizes the fall risk margin for current workers and includes at least a normal fall margin status and a fall margin of concern status. When the fall margin status reaches a preset risk condition, the edge controller generates a pre-lockdown requirement status. It should be noted that the pre-lockdown requirement status is a candidate intervention status for entering the pre-execution check, not the same as the actual fall confirmation status, and does not directly trigger a safety locking action.

[0051] Furthermore, the edge controller generates IMU attitude state based on the person's posture data. The IMU attitude state includes torso tilt angle, torso tilt angular velocity, vertical acceleration rate of change, and dynamic stability margin. Among them, the torso tilt angle and torso tilt angular velocity are determined based on the three-axis acceleration, three-axis angular velocity, and attitude angle, while the vertical acceleration rate of change is determined based on the vertical acceleration change within the adjacent data processing window.

[0052] Furthermore, the edge controller generates an unstable trend state when at least one of the following conditions is met: First, the torso tilt angle exceeds the preset tilt angle threshold and continues to reach the first time window; Second, the torso tilt angular velocity exceeds the preset angular velocity threshold; Third, the vertical acceleration exhibits short-term weightlessness characteristics and the rate of change of attitude angle exceeds the preset instability threshold; Fourth, the dynamic stability margin exceeds the preset threshold and continues to reach the second time window.

[0053] Furthermore, in one embodiment, the edge controller simplifies the worker's body into an inverted pendulum model, uses the location of the torso IMU as an approximate center of mass position, and estimates the deviation of the center of mass position from the foot support domain boundary based on the worker's posture data to obtain the dynamic stability margin overshoot ratio. The foot support domain is determined based on at least one of a static standing support region, a gait support region, and a preset work posture support region.

[0054] The dynamic stability margin overshoot ratio is the ratio between the distance the predicted centroid position exceeds the foot support domain boundary and the feature size of the support domain. When the predicted centroid position is within the foot support domain, the dynamic stability margin overshoot ratio is zero. When the predicted centroid position exceeds the foot support domain boundary, the dynamic stability margin overshoot ratio increases with the distance exceeded.

[0055] Furthermore, the edge controller corrects the second coefficient threshold based on the proportion of dynamic stability margin exceeding the limit, resulting in the corrected second coefficient threshold:

[0056] in, This is the corrected second coefficient threshold. The second coefficient threshold, This is the dynamic stability margin threshold correction coefficient. Furthermore, the dynamic stability margin threshold correction coefficient is determined as follows:

[0057] in, This is the upper limit of the threshold correction coefficient. These are the dynamic stability margin mapping coefficients. This represents the percentage of dynamic stability margin exceeding the limit. Through the above processing, the IMU attitude state is incorporated into the correction of the dynamic fall risk assessment threshold.

[0058] Furthermore, the edge controller generates a fall margin state based on the dynamic fall risk coefficient, the rate of change of the dynamic fall risk coefficient, the slack margin, the current effective attachment point, the state without an effective attachment point, the IMU attitude state, the instability trend state, and the window data anomaly markers.

[0059] When the dynamic fall risk coefficient is lower than the first coefficient threshold, the rate of change of the dynamic fall risk coefficient does not continue to rise, the slack margin does not continue to increase, there is no state without effective anchor points, and no unstable trend is generated, the edge controller generates a normal fall margin state.

[0060] When the dynamic fall risk coefficient reaches the first coefficient threshold but not the corrected second coefficient threshold, or when the rate of change of the dynamic fall risk coefficient shows an upward trend within the continuous data processing window, or when the slack margin continues to increase, or when an unstable trend is generated, the edge controller generates a fall margin state of concern.

[0061] Furthermore, when the fall margin condition reaches a preset risk condition, the edge controller generates a pre-lockdown requirement state; the preset risk condition includes at least one of the following: First, the dynamic fall risk coefficient reaches the revised second coefficient threshold and generates an unstable trend state; Second, the dynamic fall risk coefficient did not reach the corrected second coefficient threshold, but the rate of change of the dynamic fall risk coefficient exceeded the rate of change threshold within the continuous data processing window, and the relaxation margin continued to increase. Third, the dynamic fall risk coefficient has reached the third coefficient threshold; Fourth, the rate of descent of the worker's connection point exceeds the preset descent rate threshold, and the rate of change of attitude angle changes abruptly. Fifth, the dynamic stability margin exceeds the limit by a certain percentage, and the unstable trend continues for a certain period of time. Sixth, S2 generates a state with no valid attachment points, and the data source related to the state with no valid attachment points is not marked as abnormal by the window data anomaly flag.

[0062] It should be noted that when any of the aforementioned preset risk conditions are met, the edge controller only generates a pre-lockdown requirement state and outputs it to S4 for pre-execution review. Pre-lockdown requirement states generated by different trigger paths are accompanied by corresponding trigger type identifiers, which are also output to S4 along with the pre-lockdown requirement state. Trigger type identifiers include at least one of risk coefficient trigger identifiers, attitude trigger identifiers, and hang point anomaly trigger identifiers, used to assist S4 in selecting the pre-execution review channel or adjusting the weights of heterogeneous review data.

[0063] Furthermore, to avoid instantaneous disturbances in a single data processing window triggering a pre-lockdown requirement, the edge controller employs a sliding time window voting mechanism to confirm preset risk conditions. Specifically, within a preset number of consecutive data processing windows, the number of data processing windows that meet the preset risk conditions is counted, and the total number of valid windows is used as the denominator for calculation; when a data processing window is determined to be unusable due to an anomaly flag, that data processing window is not included in the total number of valid windows.

[0064] When the proportion of data processing windows that meet the preset risk conditions reaches the preset proportion threshold, the fall margin status is confirmed to meet the preset risk conditions, and a pre-lockdown requirement status is generated; if the total number of valid windows is less than the preset minimum number of valid windows, the pre-lockdown requirement status is not triggered, and the current fall margin status is maintained or an anomaly handling flag related to insufficient data is generated.

[0065] It should also be noted that when there is no valid dynamic fall risk coefficient in the previous data processing window in S2, the rate of change of the dynamic fall risk coefficient is set to zero, and this initial rate of change is not used as the trigger for focusing on the fall margin state or the pre-lockdown demand state.

[0066] Furthermore, when the window data anomaly marker indicates that the data involved in the judgment has abnormal height, rope length, tension, or is not in a load-bearing attachment state, the edge controller does not directly generate a pre-lockdown requirement state based on the abnormal data. Instead, it re-judges the fall margin state by combining the data sources that are not marked as abnormal.

[0067] When the available data source is insufficient to support the fall margin status judgment, an exception handling flag related to insufficient main judgment data is generated. If a valid dynamic fall risk coefficient or a valid fall margin status still exists, the exception handling flag is output to S4 along with the valid data. If a valid dynamic fall risk coefficient does not exist and a fall margin status cannot be generated, a pre-lockdown requirement status is not generated, and the exception handling flag is output to the subsequent exception handling process. The exception handling flag and the exception handling flag generated in S1 due to multiple consecutive data processing windows anomalies belong to the same system-level flag system, but they are accompanied by different triggering reason information.

[0068] S4. When generating a pre-lockdown requirement state, heterogeneous back-check data is retrieved from the pre-processed fall protection state data to construct a pre-execution back-check state. Based on the pre-execution back-check state, one of the following is generated: a true fall confirmation state, a high-risk disturbance state, or a back-check anomaly state; specifically including: When S3 generates a pre-lockout requirement state, the edge controller does not directly output a safety locking action. Instead, it initiates a pre-execution check process at the moment the pre-lockout requirement state is generated. The pre-execution check process is used to confirm whether the pre-lockout requirement state corresponds to an actual fall before the graded fall protection control actions are executed.

[0069] Furthermore, the edge controller retrieves heterogeneous feedback data from the preprocessed fall arrest status data output by S1. This heterogeneous feedback data includes at least two categories from the fall arrester rope exit speed, worker connection point height descent rate, safety rope tension abrupt change characteristics, redundant hook force abrupt change characteristics, rope impact acceleration characteristics, and fall arrest actuator status feedback. Specifically, the redundant hook force data is associated with the redundant feedback attachment point information determined by S2 and used to determine the redundant hook force abrupt change conditions.

[0070] It should be noted that the heterogeneous back-check data uses physical measurement principles that are at least partially different from those used to generate the pre-lockout requirement state; the data used to generate the pre-lockout requirement state includes at least the dynamic fall risk coefficient, slack margin, and IMU attitude state. The pre-execution back-check process cross-confirms the actual fall characteristics through rope release speed, rope force, impact acceleration, redundant hook force, and fall arrestor status feedback.

[0071] Furthermore, the edge controller selects the pre-execution check channel or adjusts the weights of heterogeneous check data based on the trigger type identifier output by S3. When the trigger type identifier is a risk factor trigger identifier, the check weights for the fall arrester rope release speed, the worker's connection point height descent rate, and the safety rope tension mutation characteristics are increased; when the trigger type identifier is an attitude trigger identifier, the check weights for the rope impact acceleration characteristics and the safety rope tension mutation characteristics are increased; when the trigger type identifier is an abort point anomaly trigger identifier, the check weights for the redundant hook force mutation characteristics and the hook locking state consistency characteristics are increased. The trigger type identifier is only used to assist in the selection of the pre-execution check channel or weight configuration and is not used alone as the basis for generating the actual fall confirmation status.

[0072] It should also be noted that adjusting the weights of heterogeneous backcheck data includes at least one of adjusting the preset threshold of the corresponding backcheck condition and adjusting the contribution coefficient of the corresponding pre-execution backcheck channel in the number of backcheck conditions met; wherein, adjusting the preset threshold means lowering or raising the trigger threshold of the corresponding backcheck condition, and adjusting the contribution coefficient means adding a contribution coefficient to the satisfaction result of the corresponding pre-execution backcheck channel. In a preferred embodiment, the edge controller achieves weight configuration by adjusting the preset threshold of the corresponding backcheck condition, while keeping the counting rule for the number of backcheck conditions met unchanged.

[0073] Furthermore, the pre-execution review is completed within a preset review time window. The preset review time window is determined based on the response time of the fall arrest execution unit, the processing cycle of the edge controller, and the sampling frequency of the data sources participating in the review. The edge controller uses the moment when the pre-locked-out demand state is generated as the reference moment for the start of the review and extracts heterogeneous review data from the preset time range before and after that moment. For high-frequency data, the window peak value, the maximum rate of change of the window, or the duration of the abrupt change feature is used as the review feature. For low-frequency data or event-driven data, the most recent valid state value or the timestamp of the event occurrence within the preset review time window is used as the review feature.

[0074] Furthermore, the edge controller performs a validity check on the data channels involved in the pre-execution backcheck. For the first... A pre-execution backcheck channel is considered a valid backcheck channel if it contains a valid sampling point, a valid event state, or a window value that can be determined by adjacent valid data within a preset backcheck time window, and the corresponding data is not marked as unavailable by the window data anomaly flag.

[0075] in, To determine the effective number of return inspection channels, The total number of data channels participating in the pre-execution review. For the first The validity identifier of the first pre-execution check channel; when the first When each pre-execution check channel is a valid check channel. ,otherwise .

[0076] Furthermore, the edge controller determines the number of channels that meet the corresponding return-to-check conditions based on whether each valid return-to-check channel meets the corresponding return-to-check conditions:

[0077] in, To meet the number of return inspection conditions, For the first An indicator to determine whether the pre-execution check channel meets the corresponding check conditions; when the first... When each pre-execution check channel meets the corresponding check conditions ,otherwise .

[0078] Furthermore, the re-inspection conditions include at least two of the following: the worker's connection point height descent rate exceeds the preset height descent threshold; the fall arrestor's rope extension speed exceeds the preset rope extension speed threshold; the maximum rate of change of the safety rope tension value within a preset short time window exceeds the tension mutation threshold; the maximum rate of change of the redundant hook force data within a window exceeds the redundant hook force mutation threshold; the peak value of the rope impact acceleration data within a window reaches the impact pulse threshold, or the duration of the mutation characteristic reaches the preset impact duration; the fall arrestor's status feedback shows that the rope status is abnormal, or the fall arrestor is in an unexpected state.

[0079] Furthermore, when the pre-lockout requirement is met, and the number of valid return check channels is satisfied:

[0080] And the number of return check conditions is satisfied:

[0081] At that time, the edge controller generates a true fall confirmation status. Among these, To minimize the number of valid return check channels, The minimum number of conditions required to confirm a real fall; the real fall confirmation status serves as the basis for the S5 output of the safety locking action.

[0082] Furthermore, when the pre-lockout requirement is met, and the number of valid return check channels is satisfied:

[0083] However, the number of conditions for re-checking is satisfied: At that time, the edge controller generates a high-risk disturbance state. Among them, The minimum number of conditions required to satisfy a high-risk disturbance state, and A high-risk disturbance state is used to characterize a current state that exhibits some characteristics of a fall risk, but is insufficient to confirm an actual fall, and serves as... Output the status basis for reversible pre-tightening action, resistance increase action, or low-level warning action.

[0084] Furthermore, when the pre-lockout requirement is met, but the number of valid return check channels is satisfied: When the edge control generates insufficient return data, or when the number of valid return channels meets the requirement but the number of channels meeting the return conditions is less than [a certain number]... If the data used to generate the pre-lockout requirement state contains window data anomaly markers and the heterogeneous return data does not support real fall characteristics, the edge controller generates a return failure state. Both the insufficient return data state and the return failure state are return abnormal states; the return abnormal state is accompanied by a sub-state identifier, which is used to distinguish between the insufficient return data state and the return failure state, and is output to S5 along with the return abnormal state.

[0085] Furthermore, if S3 outputs an exception handling flag related to insufficient main decision data, and S4 still contains at least... If there is one valid return check channel, the edge controller will prioritize constructing the pre-execution return check state based on the heterogeneous return check data. If the abnormal handling identifier output by S3 and the insufficient return check data state in S4 exist at the same time, they will be merged into a return check abnormal state, and two types of triggering reason information, namely insufficient main judgment data and insufficient return check data, will be added for S5 to distinguish and process.

[0086] It should also be noted that the edge controller will record the pre-execution return check status, the number of valid return check channels, the number of return check conditions met, the trigger type identifier, the actual fall confirmation status, the high-risk disturbance status, or the return check abnormal status as the pre-execution return check status information for this pre-lock event, and output the pre-execution return check status information to S5 along with the status, while retaining it for S6 to write the fall prevention event record.

[0087] Thus, S4 forms a reliable gate before execution between the pre-lockout demand state and the graded fall protection control action. By cross-confirming heterogeneous data, the pre-lockout demand state is diverted into the actual fall confirmation state, the high-risk disturbance state, or the return check anomaly state, so that the safety locking action only has a state basis when the actual fall is confirmed.

[0088] S5. Based on the fall margin status and the pre-execution check status, output the corresponding graded fall protection control actions; specifically including: Within the current data processing window, the edge controller receives the pre-execution check status information output by S4 and the fall margin status output by S3. The pre-execution check status information includes at least the actual fall confirmation status, high-risk disturbance status, check anomaly status, sub-status identifier of the check anomaly status, number of valid check channels, number of check conditions met, trigger type identifier, and corresponding trigger reason information.

[0089] Furthermore, the edge controller generates graded fall protection control actions based on the correspondence between the pre-execution check status information and the fall margin status. The graded fall protection control actions include at least one of the following: safety locking action, reversible pre-tensioning action, drag increase action, low-level warning action, and anomaly handling action.

[0090] It should be noted that the graded fall arrest control action is not directly triggered by the pre-lockout requirement state, but is determined jointly by the pre-execution check state output by S4 and the fall margin state output by S3. The pre-lockout requirement state is only used to initiate the pre-execution check process, and the safety locking action is only output when the pre-execution check state is a confirmed actual fall state.

[0091] Furthermore, when S4 generates a true fall confirmation state, regardless of the fall margin state output by S3, the edge controller prioritizes outputting a safety locking action to the fall arrestor unit, causing the fall arrestor unit to switch from a free rope extension state, a light tension state, or a resistance-increasing state to a safety locking state. The safety locking action includes stopping the fall arrestor from extending the rope further, locking the safety rope's load path, recording the locking trigger time, and generating a safety locking action identifier.

[0092] It should also be noted that the safety locking action is based on the actual confirmed state of the fall, and is not directly triggered by a single drop in height, a single abnormal posture, or a single sudden change in tension.

[0093] Furthermore, when S4 generates a high-risk disturbance state, the edge controller, in conjunction with the fall margin status output by S3, outputs at least one of the following actions: reversible pre-tightening, resistance-increasing, or low-level warning. The reversible pre-tightening action refers to controlling the fall arrestor to tighten to a preset length, or maintaining the safety rope tension within a preset light tension range; the resistance-increasing action refers to increasing the resistance of the fall arrestor's rope extension or limiting the rope extension speed; the low-level warning action includes outputting alert information to the worker's wearing end, the edge controller display end, or the on-site safety management platform.

[0094] Furthermore, when S4 generates a return inspection anomaly, the edge controller determines the anomaly handling method based on the sub-state identifier of the return inspection anomaly. When the sub-state identifier is insufficient return inspection data, the edge controller does not output a safety locking action, and outputs a reversible pre-tightening action, a low-level warning action, or an anomaly handling action according to the fall margin status, so that the fall arrestor unit remains in a recoverable state; when the sub-state identifier is a failed return inspection, the edge controller does not output a safety locking action, and outputs a low-level warning action, maintains the current fall arrestor unit state, or outputs an anomaly handling action.

[0095] Furthermore, the exception handling actions include generating an exception handling identifier, recording the reason for the exception during re-checking, generating an exception data source identifier, and requesting subsequent data processing windows to continue re-checking. It should be noted that the exception data source identifier is used for reference in the status judgment process of subsequent data processing windows; S5 does not directly adjust the data source weight or re-check weight.

[0096] Furthermore, when the pre-execution review is not initiated, or the pre-execution review status does not include the actual fall confirmation status, high-risk disturbance status, and abnormal review status, and S3 generates a fall margin concern status, the edge controller outputs a low-level warning action or maintains the current fall protection execution unit status; when S3 generates a normal fall margin status and there is no pre-execution review status requiring intervention, the edge controller does not output an active intervention action and maintains the fall protection execution unit in a free rope state or the current stable state.

[0097] Furthermore, the edge controller determines the final graded fall protection control actions according to the action mapping priority rule between the pre-execution check state and the fall margin state. Specifically, when the actual fall confirmation state is established, a safety locking action is output first; when the actual fall confirmation state is not established but the high-risk disturbance state is established, a reversible pre-tightening action, a drag-increasing action, or a low-level warning action is output; when the check abnormal state is established, an abnormal handling action, a reversible pre-tightening action, or a low-level warning action is output according to the sub-state identifier and the fall margin state; when there is no pre-execution check state that requires intervention and only the fall margin state that needs attention exists, a low-level warning action is output.

[0098] It should also be noted that when the edge controller outputs graded fall protection control actions, it generates action control instructions. These instructions include at least the action type, target fall protection execution unit, trigger status, trigger type identifier, execution hold condition, and release condition.

[0099] Furthermore, the execution hold conditions include at least one of the following: after S6 execution, the return check has not yet generated a valid action flag or a release condition has not been generated; the fall margin state in the subsequent data processing window has not been restored to the normal fall margin state; and the edge controller has not received a new state diversion result within the preset hold time. The release conditions include at least one of the following: the return check result after execution meets the preset recovery condition; the fall margin state is restored to the normal fall margin state; and the anomaly handling flag is released by subsequent valid data.

[0100] Through the above processing, S5 maps the fall margin state generated by S3 and the pre-execution check state generated by S4 to different levels of fall protection control actions. Among them, the safety locking action is only output under the actual fall confirmation state; the reversible pre-tightening action and the drag-increasing action are mainly used in high-risk disturbance states, and can also be used as conservative control actions when the check data is insufficient and the fall margin state has not recovered; the low-level warning action and the anomaly handling action are mainly used to pay attention to the fall margin state, the abnormal check state, or the situation of insufficient data.

[0101] S6. After the graded fall protection control actions are executed, fall protection status data is re-collected and a post-execution check is performed. A status identifier is generated based on the post-execution check results and written into the fall protection event log. Specifically, this includes: After the S5 outputs the graded fall protection control action, the edge controller uses the output time of the action control command as the reference time for the post-execution check start time, within the preset post-execution check time window. Internally, the fall protection status data was re-collected.

[0102] Specifically, the edge controller extracts data from the original data source of the S1 multi-source fall arrest status data and sets a preset post-execution check time window. The fall protection status data corresponds to the time range, and the same window feature extraction method as S1 is used to generate the post-execution review window feature values; when the preset post-execution review time window... With the data processing window period in S1 If misaligned, use the preset post-execution check time window. Feature extraction is performed as an independent re-inspection window.

[0103] Furthermore, the re-collected fall protection status data includes at least the fall protection execution unit status, the real-time rope extension length of the fall protection device, the rope extension speed of the fall protection device, the tension value of the safety rope, the height of the worker's connection point, the worker's posture data, the rope impact acceleration data, and the motion control command feedback information.

[0104] It should be noted that S4 is used to confirm the risk nature of the pre-lockout requirement state before the action is executed, and S6 is used to confirm whether the graded fall protection control action has achieved the corresponding target state after the action is executed. The post-execution review results are used to characterize the actual execution effect of the graded fall protection control action and serve as the basis for generating status identifiers, equipment configuration update identifiers, parameter offset update information, and fall protection event records.

[0105] Furthermore, the edge controller determines the target response value based on the action type of the graded fall protection control action, and determines the actual action response value based on the feature value of the post-execution feedback window, and calculates the action effect deviation:

[0106] in, This is the deviation in the effect of the movement. The actual response value of the action obtained after execution and re-check. This represents the target response value for the corresponding graded fall arrest control actions.

[0107] Furthermore, the target response value The target response value is determined based on the action type: When the graded fall arrest control action is a safety locking action, the target response value includes the target locking state and the target rope delivery speed; when the graded fall arrest control action is a reversible pre-tensioning action, the target response value includes the target rope delivery length or the preset light tension range; when the graded fall arrest control action is a drag-increasing action, the target response value includes the target rope delivery speed decrease or the target rope delivery resistance state; when the graded fall arrest control action is a low-level warning action or an anomaly handling action, the target response value is determined according to the corresponding instruction feedback confirmation conditions.

[0108] It should also be noted that for discrete state quantities such as the state of the fall arrestor actuator, the edge controller uses state consistency judgment; for continuous response quantities such as the real-time rope extension length of the fall arrestor, the rope extension speed of the fall arrestor, the tension value of the safety rope, and the downward trend of the height of the worker's connection point, the edge controller uses action effect deviation judgment.

[0109] Furthermore, when the deviation in the action effect... If the deviation does not exceed the preset action deviation threshold and the fall arrestor's status matches the action control command, the edge controller generates an action validity flag; when the action effect deviation is less than the preset action deviation threshold, the edge controller generates an action validity flag. If the preset action deviation threshold is exceeded, or if the status of the fall protection execution unit is inconsistent with the action control command, the edge controller generates an action failure flag.

[0110] Furthermore, when the graded fall protection control action is a safety locking action, if the fall protection execution unit reports a safety locking state, the fall protection device's rope extension speed drops below the preset locking speed threshold, and the downward trend of the worker's connection point height is suppressed, then an action validity indicator is generated; if the fall protection execution unit does not enter the safety locking state, or the fall protection device's rope extension speed still exceeds the preset locking speed threshold, then an action failure indicator or an abnormal handling indicator is generated.

[0111] Furthermore, when the graded fall arrest control action is a reversible pre-tightening action or a resistance-increasing action, if the real-time rope extension length of the fall arrester, the safety rope tension value, the decrease in the rope extension speed of the fall arrester, or the rope extension resistance state reaches the corresponding target response value, then an action validity indicator is generated; otherwise, an action failure indicator or an abnormal handling indicator is generated.

[0112] Furthermore, when the graded fall protection control action is a low-level warning action, the edge controller generates a warning output completion mark based on at least one of the following: the execution confirmation signal returned by the vibrator, sound and light module or display module of the operator's wearing end, the message delivery confirmation of the edge controller display end, and the data reception confirmation of the on-site safety management platform.

[0113] Furthermore, when the graded fall protection control action is an abnormal handling action, if the same data source is not marked as abnormal data again in the preset observation window, and the same type of return check abnormal state does not reappear, an abnormal handling completion mark is generated; if the same data source continues to be marked as abnormal data, or the same type of return check abnormal state reappears, an abnormality persistence mark is generated.

[0114] Furthermore, the edge controller performs a consistency comparison between the post-execution check results and the pre-execution check status information output by S4. When S4 outputs a true fall confirmation status and S5 outputs a safety locking action, if S6 does not generate a valid action flag, an action failure flag and an abnormal handling flag are generated, and the safety control status of the fall arrestor is maintained or upgraded. When S4 outputs a high-risk disturbance status and S5 outputs a reversible pre-tightening action or a drag-increasing action, if the subsequent fall margin status recovers to a normal fall margin status, a risk release flag is generated. If the fall margin status is still monitored, the current control action is maintained or a request is made for the subsequent data processing window to continue checking.

[0115] Furthermore, when the post-execution check detects that the safety locking action has been executed, the rope impact acceleration exceeds the preset impact threshold, the action failure indicator appears continuously, or the fall arrest execution unit status is inconsistent with the action control command, the edge controller generates an equipment configuration update indicator.

[0116] It should be noted that when the post-execution check detects that the rope impact acceleration exceeds the preset impact threshold, the edge controller determines whether to generate an equipment configuration update identifier based on the order of the impact start time and the action execution time, as well as the impact amplitude level. If the impact start time is earlier than the safety locking action execution time, or if the impact start time is later than the safety locking action execution time but the impact amplitude exceeds the preset lockable impact limit, then an equipment configuration update identifier is generated.

[0117] Furthermore, the equipment configuration update flag indicates that the compensation parameters, action response parameters, or fall arrestor status of the current fall arrest equipment need to be reconfigured. Until the equipment configuration update flag is removed, the edge controller suspends the use of the buffer extension compensation in the original equipment configuration table. Seatbelt extension compensation and fall arrester braking margin compensation As a valid configuration parameter, it is used in the subsequent data processing window to call the updated equipment configuration table to participate in the construction of the dynamic fall risk coefficient of S2.

[0118] Furthermore, the edge controller counts the cumulative number of abnormal events corresponding to the same work area, the same attachment point type, or the same fall arrest equipment within a preset statistical period:

[0119] in, This represents the cumulative number of abnormal events. The number of fall prevention incidents within a preset statistical period, For the first The indicator for whether a fall prevention incident meets the abnormal statistical conditions; when the first... When a fall prevention incident meets at least one of the following abnormal statistical conditions: confirmed fall status, high-risk disturbance status, non-compliance action indicator, or equipment configuration update indicator. ,otherwise .

[0120] It should also be noted that the number of fall incidents It refers to the number of pre-locked-out demand states generated within a preset statistical period. Each pre-locked-out demand state is counted as a fall prevention event as the complete process from generation, pre-execution check, output of graded fall prevention control actions, to post-execution check completion.

[0121] Furthermore, when the cumulative number of abnormal events... When the preset cumulative threshold is reached, the edge controller generates parameter bias update information. In subsequent work cycles or subsequent data processing windows, the parameter bias update information applies to at least one of the following: equipment configuration table parameters in S2, fall margin status threshold in S3, and return check confirmation condition parameters in S4. The specific update scope is determined based on the abnormal event type, the cumulative number of abnormal events, and the equipment configuration update identifier, and does not change the historical fall prevention event records that have already been written.

[0122] It should also be noted that the status identifiers include at least one of the following: action valid, action not met, risk cleared, warning output completed, anomaly handling completed, anomaly persisting, and anomaly handling. Equipment configuration update identifiers and parameter offset update information are generated based on the post-execution review results and the cumulative results of anomaly events, and are written into the fall protection event record along with the status identifiers. All of the above status identifiers, equipment configuration update identifiers, and parameter offset update information are automatically generated by the edge controller and do not rely on manual review or confirmation.

[0123] Furthermore, the edge controller writes the data and status generated from S1 to S6 into the fall protection event log. The fall protection event log includes at least the event number, work area identifier, currently valid anchor points, redundant re-check anchor point information, dynamic fall risk coefficient and its rate of change, slack margin, fall margin status, pre-lockout requirement status, trigger type identifier, and the corrected second coefficient threshold. Dynamic stability margin over-limit ratio The system includes: pre-execution status check information, graded fall protection control actions, action control commands, post-execution check results, status indicators, equipment configuration update indicators, and parameter offset update information.

[0124] Through the above processing, S6 completes the post-execution review of graded fall protection control actions, status identifier generation, equipment configuration update judgment, parameter bias update judgment, and fall protection event recording, thus forming a complete fall protection control closed loop from S1 to S6. Example 2 This embodiment provides a high-altitude operation fall prevention and control device, which is used to execute the high-altitude operation fall prevention and control method described in Embodiment 1. The device in this embodiment and the method in Embodiment 1 belong to the same inventive concept. The limitations in Embodiment 1 regarding multi-source fall prevention status data, pre-processed fall prevention status data, current effective anchor points, dynamic fall risk coefficient, fall margin status, pre-lockout requirement status, heterogeneous back-check data, pre-execution back-check status, graded fall prevention control actions, post-execution back-check, status identification, and fall prevention event records are all applicable to this embodiment.

[0125] The high-altitude operation fall prevention and control device includes a controller, which is equipped with a data preprocessing module, an effective anchor point determination module, a fall margin generation module, a pre-execution check module, a graded control module, and a post-execution check module. It should be noted that each functional module can be implemented by program instructions, firmware logic, or a data processing unit running in the controller. The module division is used to describe the logical division of labor for the controller in performing different data processing functions and does not limit the controller to forming mutually independent physical components.

[0126] Furthermore, the data preprocessing module is used to acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data. Specifically, the data preprocessing module unifies the time base of the multi-source fall protection status data, divides it into pre-calibrated data, quasi-static configuration data, and dynamic fall protection status data according to data attributes, and generates a window feature set for the dynamic fall protection status data; simultaneously, the data preprocessing module performs data quality judgment on the window feature set, adds window data anomaly markers according to the judgment results, and outputs the pre-calibrated data, quasi-static configuration data, window feature set, and corresponding window data anomaly markers as preprocessed fall protection status data.

[0127] Furthermore, the effective anchor point determination module receives the preprocessed fall protection status data output by the data preprocessing module. Based on candidate anchor point locations, candidate anchor point heights, hook locking status, safety rope tension, real-time rope extension length of the fall arrester, worker connection point height, and window data anomaly markers, it filters candidate anchor points for validity. Combining the consistency between the real-time rope extension length of the fall arrester and the spatial distance between the worker connection point and the candidate anchor point, it determines whether there is a currently effective anchor point or no effective anchor point. The effective anchor point determination module also determines the slack margin, anchor point height relationship, worker vertical movement status, and fall protection equipment configuration parameters based on the currently effective anchor points. Furthermore, based on the slack margin, anchor point height relationship, worker vertical movement status, and fall protection equipment configuration parameters, it determines the predicted potential fall displacement and effective load-bearing rope length, thereby constructing a dynamic fall risk coefficient and a dynamic fall risk coefficient change rate.

[0128] Furthermore, the fall margin generation module receives the dynamic fall risk coefficient, dynamic fall risk coefficient change rate, slack margin, current effective attachment point, and no effective attachment point status from the effective attachment point determination module, and calls the personnel posture data, window feature set, and window data anomaly flag from the preprocessed fall protection status data to generate a fall margin status. The fall margin generation module is also used to generate an IMU attitude status based on the personnel posture data, generate an instability trend status based on the IMU attitude status, and obtain a corrected second coefficient threshold based on the dynamic stability margin exceeding the limit ratio. The fall margin generation module generates a normal fall margin status or a fall margin of concern status based on the dynamic fall risk coefficient, dynamic fall risk coefficient change rate, slack margin, no effective attachment point status, IMU attitude status, instability trend status, and window data anomaly flag; when the fall margin status reaches a preset risk condition, a pre-lockout requirement status is generated, and a corresponding trigger type identifier is generated based on the trigger path that triggers the preset risk condition.

[0129] Furthermore, the pre-execution backcheck module is used to call heterogeneous backcheck data from the preprocessed fall protection status data when the fall margin generation module generates the pre-lockout requirement state, construct the pre-execution backcheck state, and generate one of the following states based on the pre-execution backcheck state: a true fall confirmation state, a high-risk disturbance state, and a backcheck anomaly state. Specifically, the pre-execution backcheck module selects a pre-execution backcheck channel or adjusts the backcheck weight of the heterogeneous backcheck data according to the trigger type identifier, extracts the backcheck features corresponding to the heterogeneous backcheck data within a preset backcheck time window, determines the number of valid backcheck channels based on the validity of the data channels participating in the pre-execution backcheck, and determines the number of backcheck conditions met based on whether each valid backcheck channel meets the corresponding backcheck conditions. The pre-execution backcheck module generates a true fall confirmation state, a high-risk disturbance state, or a backcheck anomaly state based on the number of valid backcheck channels and the number of backcheck conditions met; wherein, the backcheck anomaly state includes a backcheck data insufficiency state or a backcheck failure state, with an attached sub-state identifier.

[0130] Furthermore, the hierarchical control module receives the pre-execution review status output by the pre-execution review module and the fall margin status output by the fall margin generation module. Based on the correspondence between the pre-execution review status and the fall margin status, it outputs corresponding hierarchical fall protection control actions. Specifically, when the pre-execution review status includes a confirmed fall status, the hierarchical control module outputs a safety locking action; when the pre-execution review status includes a high-risk disturbance status, the hierarchical control module outputs a reversible pre-tightening action, a drag-increasing action, or a low-level warning action; when the pre-execution review status includes a review anomaly status, the hierarchical control module outputs an anomaly handling action, a reversible pre-tightening action, or a low-level warning action based on the sub-status identifier of the review anomaly status. When outputting the hierarchical fall protection control actions, the hierarchical control module generates action control instructions, which include action type, target fall protection execution unit, trigger status, trigger type identifier, execution hold condition, and release condition.

[0131] Furthermore, the post-execution review module is used to re-collect fall protection status data and perform a post-execution review after the execution of the graded fall protection control action. Based on the post-execution review result, a status identifier is generated and written into the fall protection event record. Specifically, the post-execution review module uses the output time of the action control command as the starting reference time for the post-execution review, re-collects fall protection status data within a preset post-execution review time window, and generates a post-execution review result based on the re-collected fall protection status data. The post-execution review module determines whether the graded fall protection control action has reached the corresponding target state based on the post-execution review result and generates the corresponding status identifier; it also determines whether to generate an equipment configuration update identifier or parameter bias update information based on the post-execution review result. The post-execution review module associates at least one of the graded fall protection control action, action control command, post-execution review result, status identifier, equipment configuration update identifier, and parameter bias update information with the current effective attachment point, dynamic fall risk coefficient, fall margin status, pre-lockout requirement status, trigger type identifier, and pre-execution review status information, and writes them into the fall protection event record.

[0132] It should also be noted that the data transfer between the data preprocessing module, effective attachment point determination module, fall margin generation module, pre-execution check module, hierarchical control module, and post-execution check module in this embodiment can be achieved through the controller's internal data cache, event queue, or task scheduling mechanism. The above description is only used to illustrate the execution order and data flow relationship of the controller's internal functional logic and does not limit the use of a specific hardware connection structure between the functional modules.

[0133] With the above device configuration, the controller can sequentially complete the preprocessing of multi-source fall protection status data, determination of the current effective attachment point, construction of dynamic fall risk coefficient, generation of fall margin status, generation of pre-lockdown requirement status, pre-execution check, output of graded fall protection control actions, post-execution check, and writing of fall protection event records, so that the high-altitude operation fall protection control device can execute the high-altitude operation fall protection control method described in Example 1.

[0134] Example 3 In this embodiment, five types of high-altitude work scenarios are selected: steel structure installation, scaffolding corner passage, edge work at floor openings, suspended platform transition work, and roof equipment area maintenance. The high-altitude work fall prevention and control method is then simulated and verified on-site. Each work scenario includes candidate attachment points, fall protection equipment worn by workers, fall arrestors, an IMU attitude acquisition unit, a rope length detection unit, a rope speed detection unit, a safety rope tension detection unit, a redundant hook force detection unit, and a rope impact acceleration detection unit. The edge controller runs the S1 to S6 processes described in Embodiment 1.

[0135] In the experiment, the existing solution was used as a control group. The existing solution mainly relies on the hook locking status, changes in personnel height, and sudden changes in single tension to trigger an alarm or lock. Using this invention as the test group, this invention performs the following steps according to Example 1: multi-source fall protection status data preprocessing, determination of current effective attachment points, construction of dynamic fall risk coefficients, generation of fall margin status, pre-execution heterogeneous back-check, graded fall protection control action output, and post-execution back-check.

[0136] In each scenario, real fall samples, non-fall disturbance samples, and data anomaly samples were set up. Real fall samples simulated the rapid descent of the worker's connection point, the increased speed of the safety rope exit, and the sudden increase in rope impact acceleration through a controlled fall test frame; non-fall disturbance samples included workers bending over to pick up tools, crossing low obstacles, briefly squatting, switching hooks, and turning to pull the safety rope; data anomaly samples included partial signal loss, instantaneous spikes in tension, inconsistencies between the rope length and spatial distance, and the state where the hook was locked but not under load.

[0137] For each test subject, the total number of test events, the number of correct actual fall confirmations, the number of missed actual fall detections, the number of times non-fall disturbances caused false locking, the number of times high-risk disturbances were correctly handled, the number of times abnormal re-inspections were correctly diverted, the average action response time, the number of times actions failed to meet standards, and the completeness rate of fall prevention event records were recorded.

[0138] During the test, the existing solution directly outputs an alarm or locking action when a single threshold is met; after generating the pre-lockdown requirement state, the present invention first constructs the pre-execution back-check state through heterogeneous back-check data, and then outputs different levels of graded fall protection control actions according to the actual fall confirmation state, high-risk disturbance state, or back-check abnormal state, and re-collects fall protection status data after the action is executed to form a fall protection event record.

[0139]

[0140] As can be seen from the data in the table, in the three comparison scenarios of the steel beam edge area, scaffolding corner area, and floor opening, the existing solution correctly confirmed the actual fall 14 times, 12 times, and 15 times, respectively, and missed the actual fall 4 times in each scenario. This indicates that the existing solution is greatly affected by single height changes, single hook status, or single tension change in the identification of actual falls, and cannot reliably distinguish between "a downward trend has been formed" and "still under controllable disturbance".

[0141] In comparison, the present invention achieved only one missed fall detection in each of the five test scenarios, with 17, 16, 18, 15, and 16 correct fall confirmations, respectively. This demonstrates that by determining the current effective anchor point, constructing a dynamic fall risk coefficient, and assessing fall margin status, a more stable risk identification foundation can be established under different anchor point heights, rope lengths, and personnel postures. Further comparison of the number of false lockouts due to non-fall disturbances shows that existing solutions resulted in 12, 10, and 14 false lockouts, respectively, while the present invention achieved only 3, 2, 3, 2, and 2 false lockouts in the five scenarios. The false alarm or false intervention rate was also reduced from approximately 20% to 23% in existing solutions to approximately 4.6% to 5.7%.

[0142] This difference indicates that the present invention does not simply improve alarm sensitivity, but rather further diverts the pre-locking requirement state into a real fall confirmation state, a high-risk disturbance state, and a return inspection abnormal state by performing a pre-heterogeneous back check, so that disturbances such as normal squatting, stepping, hook switching, and short-term pulling are no longer directly mapped as safety locking actions.

[0143] Looking at the number of correct responses to high-risk disturbances, the existing solutions only achieved 21, 19, and 23 responses in three scenarios, while the present invention achieved 50, 47, 53, 44, and 49 responses in five scenarios. This demonstrates that the reversible pre-tightening action, drag-increasing action, and low-level early warning action can effectively handle the intermediate state of "not yet being able to confirm a real fall but having a risk evolution trend."

[0144] In terms of the number of correct routing times for return inspection anomalies, this invention is significantly higher than existing solutions, indicating that a traceable anomaly handling chain is formed between window data anomaly marking, heterogeneous return inspection data, and post-execution return inspection.

[0145] Regarding the average action response time, the present invention achieves approximately 143.8ms to 158.2ms, lower than the existing solution's approximately 305.8ms to 338.4ms. This is because the edge controller of the present invention does not wait for confirmation from the remote platform, but completes risk state construction, pre-execution backcheck, and action output locally. In terms of fall protection event record completeness, the present invention maintains a rate above 96.9%, while the existing solution achieves approximately 79.8% to 84.1%. This indicates that the present invention, through the associated writing of status identifiers, equipment configuration update identifiers, parameter bias update information, and fall protection event records, ensures that each pre-lockout requirement state, pre-execution backcheck, graded fall protection control action, and post-execution backcheck forms a closed-loop record.

[0146] It can be seen that the advantage of this invention over the prior art is not just the increase in the number of sensors, but the establishment of a closed-loop state of "risk prediction - pre-execution heterogeneous confirmation - graded action - post-execution review - event recording". This allows real falls, non-fall disturbances and data anomalies to be distinguished and processed in the control chain, thereby demonstrating a lower false locking rate, higher real fall confirmation capability, better anomaly diversion capability and more complete post-event traceability capability.

[0147] The technical features of this invention not described can be implemented by or using existing technology, and will not be repeated here. Of course, the above description is not a limitation of this invention, and this invention is not limited to the examples above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of this invention should also be within the protection scope of this invention.

Claims

1. A method for preventing falls during high-altitude operations, characterized in that, Includes the following steps: S1. Acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data. S2. Determine the current effective attachment point based on the preprocessed fall protection status data, and construct a dynamic fall risk coefficient based on the current effective attachment point; S3. Generate a fall margin state based on the dynamic fall risk coefficient and the pre-processed fall protection status data, and generate a pre-lockdown requirement state when the fall margin state reaches the preset risk condition. S4. When the pre-lock-up requirement state is generated, heterogeneous back-check data is called from the pre-processed fall protection state data to construct the pre-execution back-check state, and one of the following is generated based on the pre-execution back-check state: real fall confirmation state, high-risk disturbance state, and back-check abnormal state. S5. Based on the fall margin status and the pre-execution check status, output the corresponding graded fall protection control action; S6. After the graded fall protection control action is executed, fall protection status data is collected again and a post-execution check is performed. A status identifier is generated based on the post-execution check result and the status identifier is written into the fall protection event record.

2. The high-altitude operation fall prevention and control method according to claim 1, characterized in that, Step S1 includes: Acquire multi-source fall protection status data corresponding to workers, fall protection equipment, candidate attachment points, and work areas; The time reference of the multi-source fall protection status data is unified, and the data is divided into continuous data processing windows according to a preset sampling period; Based on the data attributes, the multi-source fall protection status data is divided into pre-calibration data, quasi-static configuration data, and dynamic fall protection status data. The pre-calibration data and quasi-static configuration data serve as spatial reference data in the current effective attachment point determination process, while the dynamic fall protection status data is used to generate the corresponding window feature set. Perform data quality assessment on the window feature set, and add window data anomaly markers based on the assessment results; The precalibrated data, quasi-static configuration data, window feature set, and corresponding window data anomaly markers are output as the preprocessed fall protection status data.

3. The high-altitude operation fall prevention and control method according to claim 2, characterized in that, Step S2 includes: Extract candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, real-time rope extension length of the fall arrester, worker connection point height, and window data anomaly markers from the preprocessed fall arrest status data; Based on the candidate attachment point location, candidate attachment point height, hook locking status, safety rope tension value, and window data anomaly markers, the candidate attachment points are screened for validity. The current valid attachment point or no valid attachment point status is determined by combining the real-time rope extension length of the fall arrester with the spatial distance between the worker's connection point and the candidate attachment point. Based on the current effective anchor points, determine the slack margin between the worker's connection point and the current effective anchor points, the relationship between the anchor point heights, the worker's vertical movement status, and the configuration parameters of the fall protection equipment; Based on the slack margin, the relationship between the hanging point heights, the vertical movement state of the personnel, and the configuration parameters of the fall protection equipment, the predicted potential fall displacement and the effective load-bearing rope length are determined, and a dynamic fall risk coefficient is constructed based on the predicted potential fall displacement and the effective load-bearing rope length. The dynamic fall risk coefficient of the current data processing window is compared with the dynamic fall risk coefficient of the previous data processing window to obtain the rate of change of the dynamic fall risk coefficient. The dynamic fall risk coefficient and the rate of change of the dynamic fall risk coefficient are then output to subsequent steps.

4. The high-altitude operation fall prevention and control method according to claim 3, characterized in that, Step S3 includes: The personnel posture data is retrieved from the preprocessed fall protection status data, and the IMU posture state is generated based on the personnel posture data. The IMU posture state includes the torso tilt angle, torso tilt angular velocity, vertical acceleration rate of change, and dynamic stability margin. An instability trend state is generated based on at least one of the following: trunk tilt angle, trunk tilt angular velocity, vertical acceleration rate of change, and dynamic stability margin. The deviation of the predicted center of mass position from the foot support domain boundary is determined based on the personnel posture data, and a dynamic stability margin over-limit ratio is generated based on the deviation. The second coefficient threshold is corrected based on the dynamic stability margin over-limit ratio to obtain the corrected second coefficient threshold. The IMU attitude state, instability trend state, dynamic stability margin overshoot ratio, and the corrected second coefficient threshold are used to generate the fall margin state.

5. The high-altitude operation fall prevention and control method according to claim 4, characterized in that, Step S3 also includes: Based on the dynamic fall risk coefficient, the rate of change of the dynamic fall risk coefficient, the slack margin, the current effective attachment point, the state without an effective attachment point, the IMU attitude state, the instability trend state, and the window data anomaly marker, a normal fall margin state or a fall margin state of concern is generated. When at least one of the dynamic fall risk coefficient, dynamic fall risk coefficient change rate, slack margin, no effective anchor point state, instability trend state and dynamic stability margin exceeding limit ratio meets the preset risk conditions, a pre-lock-up requirement state is generated. A sliding time window voting confirmation is performed on the data processing windows that meet the preset risk conditions. When the proportion of the number of data processing windows that meet the preset risk conditions reaches a preset proportion threshold, the fall margin state is confirmed to have met the preset risk conditions. When generating the pre-locked-out requirement state, a corresponding trigger type identifier is generated according to the trigger path that triggers the preset risk condition, and the pre-locked-out requirement state and the trigger type identifier are output to the pre-execution backcheck.

6. The high-altitude operation fall prevention and control method according to claim 5, characterized in that, In step S4, when the pre-lockout requirement state is generated, heterogeneous back-check data is retrieved from the preprocessed fall protection state data to construct the pre-execution back-check state, including: When the pre-lockout requirement state is generated, the pre-execution back-check process is started, and heterogeneous back-check data is called from the pre-processed fall protection status data. The heterogeneous backtesting data includes at least two of the following: the fall arrester rope release speed, the worker connection point height descent rate, the safety rope tension abrupt change characteristics, the redundant hook force abrupt change characteristics, the rope impact acceleration characteristics, and the fall arrester execution unit status feedback. The pre-execution back-check channel is selected based on the trigger type identifier, or the back-check weight of the heterogeneous back-check data is adjusted. The back-check weight is determined by adjusting the preset threshold of the corresponding back-check condition or adjusting the contribution coefficient of the corresponding pre-execution back-check channel. Extract the return inspection features corresponding to the heterogeneous return inspection data within the preset return inspection time window, and construct the pre-execution return inspection state based on the return inspection features; For high-frequency data, the back-check feature includes at least one of window peak value, window maximum change rate, and duration of mutation feature; for low-frequency data or event-driven data, the back-check feature includes the most recent valid state value or event occurrence timestamp within the preset back-check time window.

7. The high-altitude operation fall prevention and control method according to claim 6, characterized in that, Step S4 generates one of the following states based on the pre-execution re-check status: actual fall confirmation status, high-risk disturbance status, and re-check anomaly status: The validity of the data channels involved in the pre-execution review is assessed to determine the number of valid review channels. The number of channels that meet the corresponding return inspection conditions is determined based on whether each valid return inspection channel meets the return inspection conditions. Based on whether the number of valid return check channels has reached the minimum number of valid return check channels, and the relationship between the number of return check conditions met and the minimum number of conditions met for actual fall confirmation and the minimum number of conditions met for high-risk disturbance state, a real fall confirmation state, a high-risk disturbance state, or a return check abnormal state is generated. Specifically, when the pre-execution re-check status meets the actual fall confirmation condition, the actual fall confirmation status is generated; when the pre-execution re-check status meets the high-risk disturbance condition but does not meet the actual fall confirmation condition, the high-risk disturbance status is generated; when the number of effective re-check channels is insufficient, or when the pre-execution re-check status does not meet the actual fall confirmation condition and the high-risk disturbance condition, the re-check abnormal status is generated. The abnormal return check status includes insufficient return check data status or return check failure status, and is accompanied by a sub-status identifier to distinguish between the insufficient return check data status and the return check failure status.

8. The high-altitude operation fall prevention and control method according to claim 7, characterized in that, Step S5 includes: Receive the pre-execution check status and the fall margin status, and determine the graded fall protection control action based on the correspondence between the pre-execution check status and the fall margin status; When the pre-execution check status includes a confirmed actual fall status, a safety locking action is output. When the pre-execution check state includes a high-risk disturbance state, a reversible pre-tightening action, a resistance increase action, or a low-level early warning action is output. When the pre-execution check status includes a check abnormal status, an abnormal handling action, a reversible pre-tightening action, or a low-level warning action is output according to the sub-status identifier of the check abnormal status. When the pre-execution review status does not include the actual fall confirmation status, high-risk disturbance status, and review abnormal status, and the fall margin status is the fall margin of concern status, a low-level warning action is output or the current fall prevention execution unit status is maintained. When outputting the graded fall protection control action, an action control instruction is generated. The action control instruction includes the action type, target fall protection execution unit, trigger status, trigger type identifier, execution hold condition, and release condition.

9. The high-altitude operation fall prevention and control method according to claim 8, characterized in that, Step S6 includes: The output time of the action control command is used as the reference time for the start of the post-execution review. The fall protection status data is re-acquired within the preset post-execution review time window, and the post-execution review result is generated based on the re-acquired fall protection status data. Based on the post-execution check results, determine whether the graded fall protection control action has reached the corresponding target state, and generate the corresponding status identifier; Based on the post-execution check results, determine whether to generate an equipment configuration update identifier or parameter offset update information; At least one of the graded fall protection control actions, action control commands, post-execution check results, status identifiers, equipment configuration update identifiers, and parameter offset update information is associated with the current effective attachment point, dynamic fall risk coefficient, fall margin status, pre-lockout requirement status, trigger type identifier, and pre-execution check status information and written into the fall protection event record.

10. A high-altitude operation fall prevention and control device, characterized in that, The controller includes a data preprocessing module, an effective attachment point determination module, a fall margin generation module, a pre-execution check module, a graded control module, and a post-execution check module. The data preprocessing module is used to acquire multi-source fall protection status data and preprocess the multi-source fall protection status data to generate preprocessed fall protection status data. The effective attachment point determination module is used to determine the current effective attachment point based on the preprocessed fall protection status data, and to construct a dynamic fall risk coefficient based on the current effective attachment point. The fall margin generation module is used to generate a fall margin state based on the dynamic fall risk coefficient and the pre-processed fall protection status data, and to generate a pre-lockdown requirement state when the fall margin state reaches a preset risk condition. The pre-execution back-check module is used to call heterogeneous back-check data from the preprocessed fall protection status data when the pre-lock-up requirement state is generated, construct the pre-execution back-check state, and generate one of the following states based on the pre-execution back-check state: real fall confirmation state, high-risk disturbance state, and back-check abnormal state. The graded control module is used to output corresponding graded fall protection control actions based on the fall margin status and the pre-execution check status. The post-execution check module is used to re-collect fall protection status data and perform post-execution check after the graded fall protection control action is executed, generate a status identifier based on the post-execution check result, and write the status identifier into the fall protection event record.

Images