Working well platform personnel falling hidden danger early warning method, system and equipment

Abstract

The invention discloses a working well platform personnel falling hidden danger early warning method, system and equipment, and relates to the technical field of falling risk monitoring. The method comprises the following steps: acquiring and preprocessing personnel-edge dynamic data, constraint effectiveness data and protection environment and process data, and respectively calculating an edge exposure index, a constraint effectiveness index and a protection environment and process index; dynamic weighted mapping is carried out based on the indexes to obtain a continuous risk score, a scene triggering threshold is combined to obtain a falling risk score, and the scene triggering threshold is deduced based on human kinematics and a falling protector braking mechanism; according to the falling risk score, graded early warning and linkage disposal are executed, prompt / alarm / emergency stop is triggered, and an event log is recorded. The method is used for solving the problems that the existing falling risk assessment is difficult to quantify the multi-source risk cumulative effect, and the response to typical high-risk scenes such as untied and hung, falling protector danger triggering and protection deficiency is insufficient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fall risk monitoring, and more particularly, to a well platform personnel fall hazard early warning method, system and device. BACKGROUND

[0002] In open edge operation scenarios such as work wells and wellhead platforms, workers usually perform operations such as up-and-down transfer, maintenance and monitoring near narrow platforms, bridges or well edges. Such operations have characteristics such as narrow operation surface, complex structure, many environmental constraints, and great difficulty in monitoring. Once a worker slips, falls or uses a fall prevention device improperly, a high fall accident can easily occur. Currently, physical protection facilities such as guardrails, bridges, fall arrestors and safety belt anchor points are mainly used in engineering to reduce the risk of falling, and management measures such as operation ticket system, monitoring system and on-site inspection are used in conjunction to reduce the risk of falling. However, the above measures are mainly based on "pre-installation and post-protection", and there is a lack of fine and quantitative co-monitoring and early warning means for the dynamic relationship between workers and well edges during operation, the real-time effectiveness of fall prevention constraints and the compliance of the protection installation process.

[0003] For high-altitude operation safety risks, for example, a high-altitude operation fall prevention safety monitoring method and system disclosed in the invention patent with the announcement number CN118628320B, by collecting physiological sign data and operation behavior data of high-altitude operation personnel, usage state data of fall prevention devices and environmental data, human risk indicators, equipment risk indicators and environmental risk indicators are determined respectively; when the risk value of any risk indicator is greater than the corresponding first preset threshold value, an alarm prompt is triggered immediately and a first operation guide is generated and pushed to the worker; when the risk values of all risk indicators do not exceed the first preset threshold value, the fall risk value is determined according to the three types of risk indicators, and compared with the second preset threshold value; when the fall risk value is greater than the second preset threshold value, an alarm prompt is triggered and a second operation guide is generated and pushed to the worker. This method to some extent realizes the quantitative evaluation and hierarchical disposal of high-altitude fall risk, and improves the safety management level of high-altitude operation.

[0004] However, the risk discrimination logic in the prior art mainly relies on the comparison of various risk indicators with fixed thresholds or the comparison of the simple weighted sum with a single falling risk threshold, which has the following deficiencies and problems: first, when multiple risk indicators are at a medium or high level but have not individually exceeded the respective preset thresholds, it is difficult to reflect the cumulative effect of the risk, and early warning may lag or even be missed; second, for typical high-risk scenarios such as "not tied but entering the near edge area of the well", "self-recovery type fall arrestor has triggered or is close to the braking limit", "guardrail is not locked or the bridge is not in place and the personnel is close to the well edge", the existing method usually uses the same weighting and threshold determination logic as other general risk factors, and the key signals are easily diluted in the weighted average, which cannot reflect their due high priority response; third, the weights of the risk indicators are mainly fixed by experience, and there is no mechanism for adaptive optimization based on historical event data, and the special working conditions of the well platform, such as small opening, large depth, concentrated channel and limited operation space, are not fully considered, so that the existing method is difficult to model and real-time fusion evaluation of "personnel-edge dynamic exposure", "constraint effectiveness change" and "protection environment and installation process state" in the well platform scenario. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a well platform personnel falling hazard early warning method, which quantifies personnel-edge dynamics, constraint effectiveness and protection environment and process information into three types of risk indexes, and obtains a falling risk score by using weighted mapping combined with scenario trigger thresholds, solving the problem that the existing falling risk evaluation cannot quantize the cumulative effect of multiple sources of risk and is insufficient in response to typical high-risk scenarios such as untied, dangerous trigger of fall arrestor and lack of protection.

[0006] To achieve the above object, the present application provides the following technical scheme: A well platform personnel falling hazard early warning method, comprising the following steps: collecting and preprocessing personnel-edge dynamic data, constraint effectiveness data and protection environment and process data to obtain a first data set; calculating edge exposure index, constraint effectiveness index and protection environment and process index based on the first data set to obtain a second data set; performing dynamic weighted mapping based on the second data set to obtain a continuous risk score, and combining scenario trigger thresholds to obtain a falling risk score, wherein the scenario trigger thresholds are derived based on human kinematics and fall arrestor braking mechanism; performing graded early warning and linked disposal according to the falling risk score, triggering prompt / alarm / emergency stop and recording event log.

[0007] In a preferred embodiment, the personnel-edge dynamic data is collected by deploying ultra-wideband positioning base stations around the wellhead perimeter and wearing positioning tags by the operating personnel, including the closest distance and approaching speed of the personnel to the well edge; the restraint effectiveness data is collected by setting a harness detection switch at the safety belt anchor point and using a self-recovery fall arrestor, including the harness state and the triggering sign of out-of-belt overspeed or free fall phenomenon; the protective environment and process data is collected by setting in-place locking switches at the guardrails and trestles and recording the completion of installation steps by scanning codes or near field communication, including the guardrail locking state, the trestle in-place state and the installation process completion or violation state.

[0008] In a preferred embodiment, the edge exposure index calculation formula is as follows:

[0009] wherein, , a and b are weights and , is a structure protection correction coefficient; when the guardrail is locked and the trestle is in place, and , otherwise, ; is an approaching speed reference, d is the closest distance of the personnel to the well edge, v is the approaching speed in the direction of the well edge, is an alert distance reference; The restraint effectiveness index calculation formula is as follows:

[0010] wherein, is a harness state, when correctly harnessed, when not harnessed; ; is a dangerous triggering sign of the self-recovery fall arrestor, when detecting the occurrence of overspeed or free fall phenomenon, otherwise, ; The protective environment and process index calculation formula is as follows:

[0011] wherein, is a guardrail locking state, when locked in place, when not locked; ; is a trestle in-place state, when in place, when dislocated; ; is an installation process violation sign, when any step is not completed or violated, otherwise, .

[0012] In a preferred embodiment, the continuous risk score is calculated according to the following formula:

[0013] wherein, is a non-negative weight coefficient.

[0014] In a preferred embodiment, the calculation of the continuous risk score further comprises determining the optimal weight coefficient, and the specific steps are as follows: determining the optimization objective function of the weight as:

[0015] wherein, is the cost coefficient corresponding to the false negative and false positive respectively, is an indicator function, is a regularization coefficient, is the continuous risk score of the i-th sample, is the label of whether the sample should trigger an early warning, and N is the number of historical samples; under the constraints of , and , solving the that makes the minimum, and using the for the calculation of the continuous risk score.

[0016] In a preferred embodiment, the fall risk score is calculated according to the following formula:

[0017] wherein, is a scene trigger threshold.

[0018] In a preferred embodiment, the scene trigger threshold is derived based on human kinematics and the braking mechanism of the fall arrestor, and the calculation formula is as follows:

[0019] wherein, is a near-edge threshold, is a preset risk increase coefficient, and satisfies .

[0020] In a preferred embodiment, the hierarchical early warning is performed according to the fall risk score, specifically: the fall risk score is divided into four continuous intervals according to the field calibration , , and; when the score falls into , output a prompt; falls into When the falling risk score falls into the third risk level, a sound-light alarm is started in the work area and the guardian is informed; the falling risk score falls into the second risk level When the falling risk score falls into the third risk level, a sound-light alarm is started in the work area and the guardian is informed; the falling risk score falls into the second risk level When the falling risk score falls into the third risk level, a sound-light alarm is started in the work area and the guardian is informed; the falling risk score falls into the second risk level .

[0021] The present application provides a kind of well platform personnel falling hazard early warning system, comprising: data acquisition module, personnel-edge dynamic data, constraint effectiveness data and protection environment and process data are acquired and preprocessed, and first data set is obtained;Index calculation module, edge exposure index, constraint effectiveness index and protection environment and process index are calculated based on first data set, and second data set is obtained;Risk mapping module, continuous risk score is obtained by dynamic weighted mapping based on second data set, and falling risk score is obtained in combination with scene trigger threshold, the scene trigger threshold is derived based on human kinematics and anti-falling device braking mechanism;Risk warning module, according to falling risk score, execute hierarchical early warning and linkage disposal, trigger prompt / alarm / emergency stop and record event log.

[0022] A kind of well platform personnel falling hazard early warning equipment, comprising memory and processor: the memory is used to store program;The processor is used to execute the program, realizes each step of the well platform personnel falling hazard early warning method.

[0023] The technical effect and advantage of a kind of well platform personnel falling hazard early warning method of the present application are as follows: The present application realizes continuous quantitative evaluation and hierarchical early warning linkage disposal to personnel falling hazard by uniformly collecting and preprocessing personnel-edge dynamic data, constraint effectiveness data and protection environment and process data in the process of well platform operation, constructing edge exposure index, constraint effectiveness index and protection environment and process index three types of risk indicators, and forming falling risk score by dynamic weighted mapping in combination with scene trigger threshold, compared with the mode of relying on static threshold and human experience judgment, can more accurately, timely identify falling risk, and provide data support for subsequent accident traceability and safety management through event log record. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 A kind of well platform personnel falling hazard early warning method process schematic diagram provided for the embodiment of the present application; Figure 2 A kind of well platform personnel falling hazard early warning system composition block diagram provided for the embodiment of the present application; Figure 3 The structural block diagram of the exemplary electronic device capable of being used to realize the embodiment of the present disclosure is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1, Figure 1 This invention provides a method for early warning of personnel fall hazards on well platforms, comprising the following steps: S1 collects and preprocesses personnel-edge dynamic data, constraint validity data, and protection environment and process data to obtain the first dataset.

[0027] In this embodiment, S1 is specifically as follows: The personnel-edge dynamic data is collected by deploying ultra-wideband positioning base stations around the wellhead and having operators wear positioning tags. This includes the closest distance between the person and the wellhead edge, and the speed at which they approach. Specifically: First, a unified spatial reference coordinate system is established for the well platform, with the wellhead center as the origin and the wellhead plane as the horizontal plane. A three-axis coordinate system is set to describe the positional relationship between the workers and the wellhead. Several ultra-wideband (UWB) positioning base stations are fixedly installed around the wellhead. These UWB base stations are wireless transceivers operating in the UWB band, capable of transmitting or receiving positioning signals with nanosecond-level pulse widths. The positioning tags worn by the workers are UWB wireless terminals that communicate with the positioning base stations. By periodically sending or responding to positioning signals, the tags measure the time difference of arrival or round-trip time with at least three positioning base stations. The positioning server or edge gateway then calculates the three-dimensional coordinates of the positioning tag in the spatial reference coordinate system, specifically at the sampling time. Calculate the coordinates of the workers' positions The wellbore is designed with a series of discrete reference points. During calibration, the system calculates the Euclidean distance between the operator and each reference point at each sampling time, and takes the minimum value as the closest distance between the operator and the well edge. The formula for calculating the closest distance is as follows: , in, Sampling time The corresponding nearest distance, Here are the coordinates of the worker at that moment. Let be the coordinates of the j-th wellbore reference point. To reflect the dynamic process of personnel approaching the wellbore, the system calculates the approach velocity based on the nearest distance change and time interval between adjacent sampling times. The formula for calculating the approach velocity is as follows: , wherein, is the sampling time is the approach velocity along the well edge direction, is the time interval between two adjacent acquisitions. In the above manner, the positioning base station and the positioning tag are used to continuously acquire and filter and smooth the positioning data, to eliminate obvious jump points and packet loss points, and to interpolate and align the position and velocity according to the time stamp, so as to form personnel-edge dynamic data including the closest distance and the approach velocity of the personnel and the edge.

[0028] The constraint effectiveness data is acquired by setting a lanyard detection switch at the safety belt anchor point and using a self-recovery fall arrestor, and includes a lanyard state and trigger flags of out-of-belt overspeed or free fall phenomena, and specifically: A lanyard detection switch is arranged at the safety belt anchor point, and the lanyard detection switch can be a mechanical travel switch, a magnetic reed switch or a Hall switch, which is used to detect whether the safety belt connecting piece is inserted and locked in place; when the connecting piece is inserted and reaches a predetermined travel, the switch is switched from an open state to a closed state, and the system marks the lanyard state as lanyarded accordingly, and vice versa. The self-recovery fall arrestor is a fall protection device with a built-in reel, and the reel applies a recovery force to the safety rope through a spring mechanism, and the safety rope automatically extends and retracts when the worker moves. The self-recovery fall arrestor is internally provided with a centrifugal brake mechanism or a speed detection mechanism, which triggers the brake when the safety rope out-of-belt speed exceeds a preset overspeed threshold, and simultaneously drives a contact or a Hall element to act, forming an out-of-belt overspeed trigger flag. In addition, the self-recovery fall arrestor can also detect the continuous abnormal change of the linear acceleration of the safety rope or the angular acceleration of the reel to identify the free fall phenomenon and output the corresponding danger trigger signal. The system encodes the open and closed states of the lanyard detection switch as a binary lanyard state, encodes the danger trigger signal of the self-recovery fall arrestor as a trigger flag of the occurrence of overspeed or free fall phenomenon, and records them into the data buffer according to a unified time stamp, time-aligns them with the personnel-edge dynamic data, and forms the constraint effectiveness data.

[0029] The protection environment and process data are acquired by setting an in-place locking switch at the guardrail and the temporary bridge and recording the installation step completion condition through code scanning or near field communication, and include the guardrail locking state, the temporary bridge in-place state and the installation process completion or violation state, and specifically: The connecting position between the guardrail post and the cross bar arranged around the wellhead is provided with a position locking switch, the lap joint or hinged position of the bridge at both ends with the support structure is also provided with a position locking switch, the position locking switch can be a mechanical limit switch, a reed switch or a Hall switch with a magnet, when the guardrail post is inserted to a predetermined depth and the locking pin is inserted in place, the corresponding switch changes from an open state to a closed state, and the system marks the guardrail locking state as locked in place accordingly; when the bridge is placed to a predetermined position and pressed to the support seat, the corresponding switch changes from open to closed, and the system marks the in-place state of the bridge as in place accordingly, if any switch is not closed, it is marked as not locked or out of position respectively. In terms of installation process recording, in order to standardize the installation sequence of the support, a two-dimensional code or a near field communication tag is arranged at the position of each key process, the installation personnel uses a handheld terminal or a mobile terminal to scan the two-dimensional code after completing a process, or approaches the near field communication tag to trigger the read-write operation, the terminal uploads the process number, operator identification and completion time to the edge computing unit, the edge computing unit compares the actual record according to the preset process sequence and time logic, if all the required processes have been completed and the sequence has not been skipped or omitted, the installation process state is marked as completed, otherwise it is marked as an exception. Through the above-mentioned mode, the protection environment and process data are coded and written into the data buffer in the form of guardrail locking state, bridge in-place state and installation process completion or exception state.

[0030] To improve the reliability of subsequent risk assessment, the above three types of original collected data are also subjected to unified preprocessing operations, specifically: first, the coordinate data from the ultra-wideband positioning tag and the state data from the on-off signal are synchronized and aligned according to the time stamp, and for the case of uneven sampling time interval, linear interpolation or zero-order hold method is used to convert to a unified sampling period; second, median filtering or low-pass filtering is used to suppress short-time noise for distance and speed data, and the front and rear valid data are used for interpolation for short-time packet loss interval; third, the on-off signal is subjected to anti-jitter processing, and only when the state remains unchanged for more than a set anti-jitter time is the state change confirmed; finally, the processed personnel-edge dynamic data, constraint validity data and protection environment and process data are organized in time sequence into a unified structure of data records to form a first data set, each record of the first data set contains at least a time stamp, the nearest distance, the approaching speed, the hitch state, the danger trigger flag, the guardrail locking state, the bridge in-place state and the installation process state corresponding to the time stamp.

[0031] The step is achieved by arranging an ultra-wideband positioning base station around the wellhead and wearing a positioning tag by the operating personnel, cooperating with the hanging detection switch at the safety belt anchor point, the self-recovery anti-falling device, and the in-place locking switch on the guardrail and the convenient bridge, and combining the code scanning or near field communication mode to record the installation process, realizing the unified collection and preprocessing of the personnel-edge position relationship, the constraint effectiveness, and the protection environment and process state, so that the originally dispersed sensing signals are aligned on the same time axis and the noise and jitter are eliminated, obtaining a structured first data set, providing complete and reliable basic data for subsequent calculation of the edge exposure index, the constraint effectiveness index, and the protection environment and process index, thereby improving the real-time and accuracy of the falling hazard early warning.

[0032] S2, calculating the edge exposure index, the constraint effectiveness index, and the protection environment and process index based on the first data set, obtaining a second data set.

[0033] In the embodiment, S2 is specifically as follows: At each sampling time, the nearest distance between the person and the well edge, the approaching speed, the hanging state, the self-recovery anti-falling device dangerous trigger flag, the guardrail locking state, the convenient bridge in-place state, and the installation process state are read from the first data set, the above-mentioned data is substituted into the pre-set index calculation formula, the edge exposure index, the constraint effectiveness index, and the protection environment and process index at the sampling time are obtained, and are written into the second data set in time sequence.

[0034] The edge exposure index is used to describe the spatial proximity and approaching trend between the operating personnel and the well edge. In the embodiment, the calculation formula of the edge exposure index is: , wherein, is the normalization result of the distance component, which is used to reflect the proportion of the nearest distance between the person and the well edge relative to the warning distance, and the calculation formula is: , wherein, d is the nearest distance between the person and the well edge at the current sampling time, is the pre-set warning distance reference; the is the normalization result of the approaching speed component, which is used to reflect the proportion of the approaching speed in the well edge direction relative to the speed reference, and the calculation formula is: , wherein, v is the approaching speed in the well edge direction at the current sampling time, is the approaching speed reference; coefficients a and b are the weights of the distance component and the speed component respectively, which are used to adjust the relative contribution of the two to the edge exposure index, and ; is the structure protection correction factor, used to reflect the reduction effect of physical protection measures such as guardrails and temporary bridges on the actual risk, and is taken as and , preferably a specific value is determined by field test calibration to reduce the edge exposure index under perfect protection conditions; otherwise, , the actual exposure degree is not reduced. Through the above definition, the edge exposure index EEI varies between 0 and 1, and increases rapidly when personnel approach an uncontrolled well along, and is inhibited when complete guardrails and temporary bridges are set.

[0035] The constraint effectiveness index is used to describe the influence of the safety belt hanging state and the self-recovery type fall arrestor working state on the personnel protection capability, and in the embodiment, the calculation formula of the constraint effectiveness index is: , wherein, is the hanging state, and the hanging detection switch from the safety belt anchor point, and is recorded as when it is detected that the safety belt is correctly connected to the specified anchor point and is in the locked state, indicating that it has been hung; and is recorded as when it is not connected or connected abnormally, indicating that it is not hung or the hanging is invalid; is the dangerous trigger flag of the self-recovery type fall arrestor, and the overspeed or free fall detection mechanism inside the self-recovery type fall arrestor, and is recorded as when it is detected that the safety rope out-of-belt speed exceeds the preset overspeed threshold or a sustained free fall phenomenon occurs, indicating that there is a falling danger event or precursor at present, and is recorded as otherwise. Through the above structure, the constraint effectiveness index gives the highest priority to the response of the two high-risk states of "not hung" and "fall arrestor dangerous trigger".

[0036] The protection environment and process index is used to describe the in-situ state of physical protection such as guardrails and temporary bridges and whether the installation process is standardized, and in the embodiment, the calculation formula of the protection environment and process index is: , wherein, is the guardrail locking state, and the in-situ locking switch from the connection between the guardrail stand and the crossbar, and is recorded as when all specified locking points are in the closed state, indicating that the guardrail is locked in place, and is recorded as otherwise; is the temporary bridge in-situ state, and the in-situ locking switch from the support end of the temporary bridge, and is recorded as when the temporary bridge is laid in place and well supported, indicating that the temporary bridge is in place, and is recorded as otherwise, indicating that the temporary bridge is out of place or missing; To install the installation process violation flag, the installation step information from the step S1 scanning the two-dimensional code or near field communication record, and when the system determines that all key procedures are completed in the predetermined order, it is recorded as When it is found that there is any step that is not completed, the order is wrong, or the timeout, etc. is violated, it is recorded as Through the above definition, the guard environment and process index SEPI unifies the guardrail missing, the bridge dislocation, and the installation process violation into a normalized index between 0 and 1, and adopts the "maximum value" method to highlight the shortest board factor.

[0037] After the calculation of the above three indexes is completed, the system writes the edge exposure index, the constraint effectiveness index, and the guard environment and process index together with the corresponding time stamp into the second data set at each sampling time, so that the second data set forms a three-dimensional index vector sequence driven by time sequence.

[0038] This step converts the original distance, speed, switch value and other multi-source heterogeneous data into three risk indexes with clear physical meaning and unified value range through normalization, weighting and maximum value operation, realizes the multi-dimensional quantitative description of the personnel proximity to the well edge risk, the constraint protection effectiveness and the compliance of the protection environment and the operation process, retains the key risk information, reduces the data dimension, provides a structured and interpretable input for subsequent continuous risk score acquisition based on weighted mapping and fall hazard early warning combined with scene trigger threshold, and improves the robustness and adjustability of the overall risk assessment.

[0039] S3, based on the second data set, a continuous risk score is obtained by dynamic weighted mapping, and a fall risk score is obtained by combining a scene trigger threshold.

[0040] It should be noted that, unlike the traditional scheme which only triggers an alarm or calculates a fall risk value by comparing whether each risk index exceeds one or two fixed thresholds, the traditional scheme is prone to dangerous accumulation without triggering an early warning when multiple indexes are high but have not individually exceeded the preset threshold. There is no essential difference between the response to the key scenes such as "unhanging", "overspeed trigger of the fall protector", and "guardrail missing" and the ordinary risk state. The present application continuously weights the three types of risk indexes into a continuous risk score between 0 and 1, and further introduces hard trigger thresholds for typical fall scenes such as unhanging, overspeed braking, and protection missing, and defines the final fall risk score as the maximum value of the continuous risk score and the scene trigger item, thereby avoiding the defect that the key signals are diluted by weighted average. While maintaining the smoothness and adjustability of the overall risk trend, the extreme dangerous situation is given priority and more intense response.

[0041] In this embodiment, S3 is specifically as follows: In the calculation process of the continuous risk score, the system reads the corresponding edge exposure index, constraint effectiveness index and protection environment and process index in the second data set at each sampling time, linearly combines the three according to the weights, and obtains the continuous risk score at this time. The calculation formula of the continuous risk score is as follows:

[0042] wherein, is the continuous risk score, EEI, RII and SEPI are respectively the edge exposure index, the constraint effectiveness index and the protection environment and process index defined in step S2, is a non-negative weight coefficient corresponding to the three indexes one by one, used to reflect the relative importance of each risk index in the overall risk assessment. In order to ensure the normalization characteristics of the continuous risk score, the three satisfy , so that always falls within the interval of 0 to 1.

[0043] In order to avoid the setting of the weight completely depending on experience and possibly invalidating with the change of the scene, the calculation of the continuous risk score in this embodiment also includes automatically determining the optimal weight coefficient based on historical event data. Specifically: The system accumulates historical event logs through long-term operation, records the edge exposure index, the constraint effectiveness index and the protection environment and process index corresponding to each historical sample at the occurrence time, and at the same time, the safety management personnel or the post-event analysis result marks whether the sample should trigger an early warning, to constitute a training sample set. On this basis, the optimization objective function of the weight is defined, which is used to measure the fitting quality of the continuous risk score to the historical samples and the safety preference under the given weight. The form of the objective function is: , wherein, is the label of whether the sample should trigger an early warning, taking the value of 1 when it should trigger an early warning, and taking the value of 0 when it should not trigger an early warning, and N is the number of historical samples; is an indicator function, taking the value of 1 when the condition in the parentheses is true, and taking the value of 0 otherwise; is the continuous risk score of the i-th sample, wherein , , is the value of the three indexes corresponding to the sample; and are respectively the penalty coefficients corresponding to the false negative and the false positive, satisfying , used to reflect that the penalty for the false negative (should alarm but not high score) is greater than the penalty for the false positive (should not alarm but the score is too high); is a non-negative regularization coefficient, which is used to limit the excessive bias of the weight coefficient, avoid the extreme case that one weight approaches 1 and other weights approach 0, and thus maintain the stability of the risk assessment. Through this design, a set of weights can be automatically found in the training process, so that the scores of samples that should have been warned in history are as close to 1 as possible, and the scores of samples that should not have been warned are as close to 0 as possible, while the smoothness of the weight vector itself is taken into account.

[0044] Under the constraints of , and , the system uses analytical solution, gradient descent or other optimization algorithms to find the that minimizes the objective function , and solidifies this set of optimal weights as the calculation parameters of the continuous risk score in the current working condition, which is used in to weight the three indices of each sampling time to obtain the real-time updated continuous risk score.

[0045] After obtaining the continuous risk score, this embodiment does not directly use it as the final falling risk assessment result, but introduces a scene trigger threshold item derived based on human kinematics and the braking mechanism of the fall arrester, to overwrite the continuous risk score. Specifically, the calculation formula of the falling risk score is: , wherein, is the final falling risk score, is a scene trigger threshold, which is used to raise the risk score to a preset level when a typical high-risk scene occurs, even if the continuous risk score has not reached a high value at that time, the final risk level can be forcibly improved, so as to realize the priority response to extreme risk scenarios. The scene trigger threshold item is derived based on human kinematics and the braking mechanism of the fall arrester, and comprehensively considers three typical dangerous scenarios, i.e., unfastening and approaching the well edge, fall arrester overspeed triggering, and guardrail or bridge failure approaching the well edge, and its calculation formula is: , wherein, is the closest distance between the human and the well edge at the current sampling time, is a near-edge threshold, which is used to define the dangerous range of approaching the well edge, and is generally selected to be slightly larger than the guardrail or safety distance; is a preset risk improvement coefficient corresponding to the three types of scenes, and satisfies , so as to reflect that the criticality of the actual triggering of the self-recovery fall arrester is higher than that of simply unfastening or guardrail / bridge failure approaching the well edge. The scene trigger threshold item The maximum value is taken, that is, as long as any typical high-risk scenario occurs, the final falling risk score will be raised to no less than the corresponding preset level.

[0046] This step first fuses the multi-source risk indexes into an adjustable continuous score based on the continuous risk score and supplemented by the scene trigger threshold, and then overwrites the continuous score with the highest priority using the typical high-risk scenarios corresponding to human kinematics and the braking mechanism of the fall arrester, thereby realizing rapid and robust response to high-risk working conditions such as "not wearing a harness near the edge", "fall arrester overspeed braking" and "missing protection near the well edge", effectively overcoming the problems of early warning delay and insensitivity to extreme scenarios caused by single fixed threshold discrimination in existing solutions, thereby providing an accurate, interpretable and engineering implementable basis for subsequent graded warning and linkage disposal based on the falling risk score.

[0047] S4, performing graded warning and linkage disposal according to the falling risk score, triggering prompt / alert / emergency stop and recording event log.

[0048] In this embodiment, S4 is specifically as follows: To make the warning grading both have good engineering operability and be consistent with the aforementioned continuous scoring system, this embodiment divides the value range of the falling risk score into four continuous and non-overlapping sub-intervals, specifically , , , and , wherein A, B and C are grading thresholds obtained by field calibration according to the work well platform operation scene, and satisfy , preferably, A is a prompt threshold corresponding to a low risk reminder, B is a warning threshold corresponding to a medium risk requiring intervention, and C is an emergency threshold corresponding to a serious risk with a high falling danger. In one preferred test of this embodiment, for the typical operation scene of a certain work well platform, the threshold A is set to 0.4, the threshold B is set to 0.7, and the threshold C is set to 0.9, that is, when the falling risk score is less than 0.4, only a prompt is made, when the falling risk score is between 0.4 and 0.7, a second-level warning is executed, when the falling risk score is between 0.7 and 0.9, a third-level warning is executed, and when the falling risk score is not less than 0.9, a fourth-level warning is executed. The above values are a preferred example, which can be adjusted according to the field test results under different working conditions.

[0049] During the online operation of the system, when the falling risk score R calculated at a certain time falls into the interval When the system determines the current warning level to be Level 1, indicating a low-risk state, it will only output text or icon-based prompts through on-site display terminals or wearable terminals, such as reminding workers to watch their step and maintain a proper working posture, without triggering any mandatory control actions; when the fall risk score R falls into the range... When the system determines the warning level to be Level 2, it activates the audible and visual alarms deployed at and around the wellhead to alert the workers and monitoring personnel in a perceptible but not excessively disturbing manner. It also sends a warning notification to the designated monitoring terminal via wireless communication, requesting the monitoring personnel to confirm the site status or take necessary intervention measures. When the fall risk score R falls into the specified range... When the system determines the warning level to be Level 3, it considers the fall risk to have reached a high level. It maintains continuous operation of the audible and visual alarms and, through the control interface with related equipment on the well platform (such as hoisting equipment and wellhead operating mechanisms), implements flow restriction or speed reduction control of related equipment in the work area. This includes reducing equipment operating speed and restricting new personnel from entering the edge area to reduce the possibility of further escalation of the risk. When the fall risk score R falls into the range... When the system determines the warning level to be Level 4, it considers the current situation to be in an emergency and dangerous state. It immediately executes an emergency stop or power cut-off command through the linkage interface with the equipment control system, so that the relevant equipment can quickly stop running or cut off the power source. At the same time, it sends an automatic distress call to the pre-configured emergency contact person and management terminal, and continuously issues high-priority audible and visual alarms on site to prompt the site to quickly enter the emergency response process.

[0050] To avoid frequent fluctuations in warning levels due to transient noise or individual sampling anomalies, this embodiment introduces a judgment duration parameter T in the aforementioned grading determination process to determine the duration of the fall risk score. That is, when a warning level changes, it does not rely directly on a single sampling result, but requires that the fall risk score R continuously falls within a certain grading interval for a period not less than the preset judgment duration T before confirming the triggering or escalation of the corresponding warning level. T can be configured according to the sampling period and the permissible response delay on-site, for example, ranging from 0.5 seconds to several seconds. Specifically, when the calculated fall risk score first crosses a certain threshold, the system starts timing. Only when the score remains within the interval corresponding to the new level for a period of T and does not return to the lower level interval will the corresponding linkage command be executed. Similarly, a similar duration confirmation mechanism can be used when the warning level falls from high to low to prevent repeated switching near the critical value. Through the above anti-jitter logic, false triggers caused by sensor noise, short-term motion fluctuations, etc., can be significantly reduced, improving the stability and reliability of the warning response.

[0051] When the linkage treatment is performed, the system also logs each early warning related event, and the event log at least includes the start time and end time of the event, the corresponding falling risk score interval, the triggered or released early warning level, the specific values of the three indexes at the time, whether the scene trigger threshold item is effective, and the execution result of the linkage treatment, and the identity of the worker, the equipment running state, and the guardian confirmation and intervention operation can also be recorded if necessary. The event log is stored in time sequence in the edge device or the background server, which is used for subsequent accident tracing, risk statistical analysis and weight optimization training, and provides real sample data for the optimization objective function of the continuous risk score weight in step S3.

[0052] Through the above hierarchical early warning and linkage treatment mechanism, this step realizes a progressive control strategy from slight reminder to forced stop under the premise of ensuring rapid response to typical falling risk scenarios, and combines the anti-shake design of the judgment duration parameter T and the closed-loop recording of the event log, which not only reduces the false positive rate, but also enhances the explainability and traceability of the early warning result, thereby further improving the reliability and practical value of the personnel falling hazard early warning method for shaft platform in actual engineering application.

[0053] Embodiment 2, Figure 2 A personnel falling hazard early warning system for a shaft platform is given, which comprises: A data acquisition module is configured to acquire and preprocess personnel-edge dynamic data, constraint effectiveness data, and protection environment and process data to obtain a first data set. An index calculation module is configured to calculate the edge exposure index, the constraint effectiveness index, and the protection environment and process index based on the first data set to obtain a second data set. A risk mapping module is configured to perform dynamic weighted mapping based on the second data set to obtain a continuous risk score, and obtain a falling risk score in combination with a scene trigger threshold derived based on human kinematics and fall protector braking mechanism. A risk early warning module is configured to perform hierarchical early warning and linkage treatment according to the falling risk score, trigger prompt / alarm / emergency stop, and record event log.

[0054] Embodiment 3, A personnel falling hazard early warning device for a shaft platform, as shown in Figure 3 comprises a memory and a processor: the memory is configured to store a program; and the processor is configured to execute the program to implement any of the embodiments of embodiment 1.

[0055] The above formulas are dimensionless values, and the formulas are derived from a large amount of data to obtain a formula closest to the real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0056] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product.

[0057] Those of ordinary skill in the art can realize that the modules and algorithm steps of the examples described in connection with the embodiments disclosed herein can be realized by electronic hardware, or a combination of computer software and electronic hardware. Whether the functions are performed by hardware or software depends on the specific application and design constraints of the technical solution. Those of ordinary skill in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0058] In addition, each functional module in each embodiment of the present application can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0059] The above description is merely preferred embodiments of the present application, but the protection scope of the present application is not limited thereto, and any modification or substitution within the technical scope disclosed by the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0060] Finally, the above description is only the preferred embodiments of the present application, and is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for early warning of personnel falling hazards on a shaft platform, characterized in that, The method comprises the following steps: collect and pre-process personnel-edge dynamic data, constraint effectiveness data, and protection environment and process data to obtain a first data set; calculate edge exposure index, constraint effectiveness index, and protection environment and process index based on the first data set to obtain a second data set; perform dynamic weighted mapping based on the second data set to obtain a continuous risk score, and obtain a falling risk score in combination with a scenario trigger threshold derived based on human kinematics and anti-falling device braking mechanism; perform hierarchical early warning and linkage treatment according to the falling risk score, trigger prompt / alarm / emergency stop, and record event log.

2. The well platform personnel falling hazard early warning method according to claim 1, wherein the personnel-edge dynamic data is collected by arranging an ultra-wideband positioning base station around the well mouth and wearing a positioning tag by the operating personnel, and includes the closest distance and approaching speed of the person to the well edge; the constraint effectiveness data is collected by setting a harness detection switch at the safety belt anchor point and using a self-recovery anti-falling device, and includes the harness state and trigger signs of out-of-belt overspeed or free-fall phenomenon; the protection environment and process data is collected by setting an in-place locking switch at the guardrail and convenient bridge, and recording the installation step completion condition by scanning the code or near field communication, and includes the guardrail locking state, the convenient bridge in-place state, and the installation process completion or violation state. The edge exposure index calculation formula is as follows:

3. The method according to claim 2, wherein, The constraint effectiveness index calculation formula is as follows: wherein, , , a and b are weights and , is a structure protection correction factor; taken as and , otherwise ; is a closing speed reference, d is the closest distance of the person to the well edge, v is the closing speed in the direction of the well edge, is a warning distance reference; The protection environment and process index calculation formula is as follows: wherein, is the hitched state, correct hitching , not hitching ; is the dangerous trigger sign of the self-recovery type fall arrestor, when detecting the out-of-band overspeed or free-fall phenomenon , otherwise ; The continuous risk score calculation formula is as follows: Wherein, Guardrail locking state, locked in place , not locked ; Convenient bridge in place state, in place , dislocation ; Installation process exception flag, any step is not completed or exception exists , otherwise .

4. The method according to claim 3, wherein, The continuous risk score calculation further includes determining the optimal weight coefficient, and the specific steps are as follows: wherein are non-negative weight coefficients.

5. The method according to claim 4, wherein, The optimization objective function of the weight is as follows: The falling risk score calculation formula is as follows: wherein, are cost coefficients corresponding to false negatives and false positives, respectively, is an indicator function, is a regularization coefficient, is a continuous risk score for the i-th sample, is a label indicating whether the sample should trigger an alert, and N is the number of historical samples. Under the constraints that , and , find the that minimizes and use for the calculation of the continuous risk score.

6. The method according to claim 5, wherein, The scenario trigger threshold is derived based on human kinematics and anti-falling device braking mechanism, and the calculation formula is as follows: wherein, is a scene trigger threshold.

7. The method according to claim 6, wherein, The hierarchical early warning according to the falling risk score is as follows: wherein, is a near edge threshold, is a preset risk increasing coefficient, and satisfies .

8. The method of claim 1, wherein the method further comprises: including: The fall risk score is divided into four consecutive intervals according to the field calibration , , and ; When the score falls into output a prompt; fall into The audio-visual alarm in the starting area is activated and the guardian is informed. fall into time the work area flow is limited or the associated equipment is slowed down and a sound and light alarm is activated; fall into emergency stop or power off and automatically call for help; wherein .

9. A system for using a personnel fall hazard early warning method for a shaft platform according to any one of claims 1 to 8, characterized in that, a data collection module for collecting and pre-processing personnel-edge dynamic data, constraint effectiveness data, and protection environment and process data to obtain a first data set; an index calculation module for calculating edge exposure index, constraint effectiveness index, and protection environment and process index based on the first data set to obtain a second data set; a risk mapping module for performing dynamic weighted mapping based on the second data set to obtain a continuous risk score, and obtaining a falling risk score in combination with a scenario trigger threshold derived based on human kinematics and anti-falling device braking mechanism; a risk early warning module for performing hierarchical early warning and linkage treatment according to the falling risk score, triggering prompt / alarm / emergency stop, and recording event log. including a memory and a processor:

10. A shaft platform personnel falling hazard early warning device, characterized by, the memory is used to store programs; the processor is used to execute the programs to realize the steps of the well platform personnel falling hazard early warning method according to any one of claims 1-8. ​

Images