A real-time monitoring system and method for the state of a power transmission line closure

Abstract

The application relates to the technical field of power transmission line state monitoring, and discloses a power transmission line netting state real-time monitoring system and method, wherein a power transmission line netting state real-time monitoring method comprises the following steps: acquiring multi-source data, establishing a netting virtual state chain and a netting section, and calculating the credibility of each data source; estimating the equivalent tension of each netting section and forming a traction constraint abnormality mark; calculating the estimated sag and the estimated transverse position of each netting section, calculating a netting coverage integrity index based on the estimated sag and the estimated transverse position; obtaining a corrected netting state, calculating a comprehensive risk score; and outputting a netting state grade; the application can quantitatively express the key states that affect the protection effect, such as insufficient netting coverage, increased sag and transverse deviation, so that the operation personnel can judge whether the netting area meets the protection requirements of the crossed objects.

Description

Technical Field

[0001] This application relates to the field of transmission line condition monitoring technology, and more specifically, it relates to a real-time monitoring system and method for the network closure status of transmission lines. Background Technology

[0002] During the construction, renovation, and conductor / ground wire replacement of power transmission lines, it is often necessary to conduct netting operations across roads, railways, rivers, existing lines, and other facilities to prevent conductors, traction ropes, tools, or construction materials from falling and intruding into the protected area. Netting operations rely on netting ropes, netting sheets, pulleys, traction devices, and anchoring devices to form a temporary protective surface. Its protective effect depends not only on whether the equipment is operating normally, but also on the coverage area, sag, lateral offset, and stress constraint stability of the netting ropes or netting sheets during deployment.

[0003] Currently, with the application of intelligent integrated machines for netting operations, self-locking pulleys, netting pulleys, and remote control devices, it is possible to acquire information such as the machine's travel distance, travel speed, wheel pressure, battery power, communication status, traction load, pulley anchoring or release status, and on-site video footage. This information helps to understand the local operating status of the equipment; however, existing monitoring methods mostly remain at the level of equipment status display or single-item alarms, focusing on the integrated machine, pulleys, and other actuators, making it difficult to directly reflect the spatial state of the netting ropes or netting sheets themselves.

[0004] In actual scenarios involving crossing safety nets, the netting area spans a large area with numerous traction and anchoring points. The netting ropes or net panels are affected by factors such as their own weight, wind load, humidity, changes in traction force, changes in pulley status, and communication delays. Even if the equipment indicates normal operation, the safety net itself may still experience issues such as insufficient coverage in certain areas, boundary shifts, increased sagging, abnormal stress, or failure to effectively shield the object being crossed. Because it is typically inconvenient to continuously deploy sensors on the safety netting ropes or net panels, existing technologies struggle to directly capture changes in their spatial morphology. While on-site video monitoring can assist in observing the safety netting status, it is susceptible to limitations in high-altitude, long-distance environments due to strong light, backlight, rain, fog, obstructions, image shakiness, complex backgrounds, and shooting angles. This makes it difficult to stably and continuously identify the edges of the netting ropes, the boundaries of the net panels, and their relative positions to the object being crossed. Relying primarily on manual observation or single video judgments can easily lead to delayed, misjudged, or missed assessments.

[0005] Therefore, the core problem with existing transmission line grid closure monitoring technology is that, under the condition that the grid body lacks continuous sensors and video observation is unstable, there is a lack of unified state correlation and reliability judgment among equipment operation data, pulley status, environmental data and local visual information. It is difficult to determine in real time whether the grid coverage is complete and whether there is any risk in the grid status, which affects the operator's judgment on handling measures such as continuing traction, suspending operation or manual verification. Summary of the Invention

[0006] The purpose of this application is to address the problem that existing power transmission line netting operations mainly rely on equipment status displays or manual video observation, which makes it difficult to make real-time comprehensive judgments on the integrity of netting coverage, sagging risk, lateral offset risk, and the reliability of monitoring results when the netting ropes or netting bodies lack continuous sensors and video observation is easily affected by the on-site environment. Therefore, this application provides a real-time monitoring system and method for power transmission line netting status.

[0007] This application provides a method for real-time monitoring of the grid closure status of transmission lines, including the following steps: From the start of the network closure operation, multi-source data is acquired according to the preset sampling period. Based on the multi-source data, a virtual state chain and network closure segment are established, and the credibility of each data source is calculated. The traction force of the integrated machine is estimated based on multi-source data. The equivalent tension of each sealing segment is estimated based on the traction force of the integrated machine and the virtual state chain of the sealing network, and anomaly markers for traction constraints are formed. Based on the netting segments, multi-source data, and equivalent tension, the estimated sagging and estimated lateral position of each netting segment are calculated, and the netting coverage integrity index is calculated based on the estimated sagging and estimated lateral position. Based on the credibility of each data source, the inversion state composed of the estimated droop and the estimated lateral position is visually corrected to obtain the corrected netting state. The mean value of the credibility of each data source is calculated to obtain the average credibility of the multi-source data. Based on the corrected netting state, the average credibility of the multi-source data, the traction constraint anomaly marker, and the netting coverage integrity index, a comprehensive risk score is calculated. The network closure status level is output based on the comprehensive risk score and the average credibility of multi-source data.

[0008] Furthermore, methods for establishing virtual state chains and blocking segments for network blocking include: Based on the numbering and binding relationships of the integrated machine, pulley, traction point, anchor point and fixed connection point in the multi-source data, a virtual state chain for net enclosure is established with the integrated machine, pulley, traction point, anchor point and fixed connection point as nodes and the boundaries of the net rope, net sheet or traction boundary as connecting edges. The locked all-in-one machine, the anchored trolley, and the fixed connection point are recorded as constraint nodes, and the corresponding connection edges are updated when unhooking or release occurs; Network segments are divided based on constraint nodes.

[0009] Furthermore, methods for calculating the credibility of each data source include: Multi-source data is filtered for validity based on communication status, communication delay, and power consumption. For the multi-source data that has been filtered for validity, the credibility of each data source is calculated, including the credibility of non-visual data sources and visual credibility. For non-visual data sources, a predicted value is generated based on the closed virtual state chain. The credibility of the non-visual data source is calculated based on the normalized squared residual between the current measurement value and the predicted value, as well as the data time delay of the non-visual data source from the current sampling time. For video data, visual credibility is calculated based on image clarity, occlusion ratio, image shake intensity, and lighting anomaly degree obtained from the video recognition module.

[0010] Furthermore, methods for estimating the equivalent tension of each closure segment include: Estimate the traction force of the integrated machine based on the traction load sensor or motor current; For any network segment, the set of traction points bound to the corresponding network segment is determined according to the network virtual state chain, and the traction force of the integrated machine transmitted through the trolley is reduced according to the preset force transmission node friction reduction coefficient to obtain the effective traction force of each traction point. Determine the angle between the effective traction force direction of the traction point and the projection direction of the sealing section onto the horizontal plane, and multiply the effective traction force of each traction point by the cosine value of the corresponding angle to obtain the horizontal projection component of the traction point. The equivalent tension of each sealing segment is obtained by summing the horizontal projection components whose cosine value of the included angle is greater than zero.

[0011] Furthermore, methods for generating traction constraint anomaly markers include: Calculate the product of the preset friction coefficient or preset equivalent friction parameter and the wheel pressure of the integrated machine, and then subtract the traction force of the integrated machine to obtain the friction margin of the integrated machine. When the friction margin is less than the preset friction margin threshold, it is determined that the corresponding integrated machine has the risk of slippage or insufficient anchoring. When the equivalent tension of the sealing section is less than the preset minimum effective tension, the sealing section is marked as a suspected abnormal stress section; When the increase in the traction force of the integrated machine corresponding to the blocked section is greater than the preset threshold for the increase in the traction force of the integrated machine in adjacent sampling times, and the increase in the travel distance of the integrated machine is less than the preset threshold for the increase in travel distance in adjacent sampling times, the blocked section will be marked as a suspected stuck section. When it is determined that the corresponding integrated machine has a risk of slippage or insufficient anchoring, or is marked as a suspected jamming section or a suspected abnormal force section, a traction constraint abnormality mark is formed.

[0012] Furthermore, the methods for calculating the estimated sag and estimated lateral position of the enclosure segment include: The horizontal distance and azimuth of the netting section are calculated based on the coordinates of the constraint nodes at both ends of the netting section, and the wind speed component perpendicular to the direction of the netting section is determined based on the wind speed, wind direction and azimuth of the netting section. Based on the estimated sag and horizontal distance of the previous moment, the preset parabolic approximation conditions are determined. When the preset parabolic approximation conditions are met, the parabolic approximation model is adopted, and the estimated sag of the netting section is calculated based on the wind speed component, horizontal distance and equivalent tension. When the preset parabolic approximation conditions are not met, the catenary model is adopted, and the estimated sag of the netting section is calculated based on the wind speed component, horizontal distance and equivalent tension. Project the constraint nodes at both ends of the enclosure segment onto the plane of the preset target enclosure area, and use the average of the lateral coordinates of the two projection points as the estimated lateral position.

[0013] Furthermore, methods for obtaining the network coverage integrity index include: The estimated sag and estimated lateral position of each netting segment are mapped to the plane of the preset target netting area to generate the boundary curve of each netting segment. The gap between the boundary curves of adjacent netting segments is filled in according to the preset boundary filling gap threshold to obtain the real-time estimated coverage area. The effective coverage area is obtained by subtracting the area of ​​the void area located inside the real-time estimated coverage area and the area corresponding to the uncovered gaps that have not been filled in the boundary from the overlapping area between the real-time estimated coverage area and the target closed area. Divide the effective coverage area by the target netting area to obtain the netting coverage integrity index.

[0014] Furthermore, methods for obtaining the corrected network closure status include: Arrange the estimated sag and estimated lateral position of each netting segment in the order of the netting segments to form the inversion state before visual correction; When the visual credibility of each data source is greater than or equal to the preset visual credibility threshold, coordinate registration is performed based on the position of the conductor obtained by video recognition. The visual observation value is composed of the registered net rope edge, net sheet edge or pulley position, and the visual observation noise matrix is ​​adjusted according to the visual credibility. The correction gain is calculated based on the adjusted visual observation noise matrix. The inversion state before visual correction is corrected according to the correction gain and the visual observation value to obtain the corrected netting state. When the visual credibility is less than the preset visual credibility threshold, the inversion state before visual correction is taken as the corrected network closure state.

[0015] Furthermore, methods for calculating the comprehensive risk score include: Based on the corrected netting status, the estimated sag and estimated lateral position of each netting segment are obtained; The sag risk score is calculated based on the estimated sag amount and the preset maximum allowable sag amount. The lateral offset risk score is calculated based on the estimated lateral position. The net coverage integrity risk score is calculated based on the net coverage integrity index. The traction constraint risk item is determined based on the traction constraint anomaly marker. The comprehensive risk score is obtained by multiplying the network coverage integrity risk score, the maximum value of the sagging risk score of each network segment, the maximum value of the lateral offset risk score of each network segment, and the result after subtracting the average confidence of multi-source data from one, along with the traction constraint risk item, by the corresponding preset risk weights and summing them.

[0016] Furthermore, methods for outputting the network closure status level include: Calculate the minimum credibility of the key data source based on the credibility of each data source; When the average credibility of multi-source data is less than the preset threshold for insufficient monitoring credibility, or when the minimum credibility of the key data source is less than the preset threshold for the credibility of the key data source, the output network closure status level is insufficient monitoring credibility. In the absence of output monitoring credibility, when the comprehensive risk score is greater than or equal to the preset abnormal state risk threshold, the output network closure status level is abnormal. When the comprehensive risk score is greater than or equal to the preset risk threshold for the state of concern, but less than the preset risk threshold for the state of abnormality, or when the average credibility of multi-source data is less than the preset credibility threshold for the state of normality, the output network closure status level is network closure status of concern. When the comprehensive risk score is less than the preset risk threshold for the state of concern, and the average credibility of the multi-source data is greater than or equal to the preset credibility threshold for the state of normal, the output network closure status level is normal.

[0017] This application provides a real-time monitoring system for the grid closure status of transmission lines, used to execute the aforementioned real-time monitoring method for the grid closure status of transmission lines, including: The multi-source data module is used to acquire multi-source data according to a preset sampling period from the start of the net sealing operation, establish a virtual state chain and net sealing segment based on the multi-source data, and calculate the credibility of each data source. The anomaly marking module is used to estimate the traction force of the integrated machine based on multi-source data, estimate the equivalent tension of each sealing segment based on the traction force of the integrated machine and the virtual state chain of the sealing network, and form traction constraint anomaly markings. The coverage evaluation module is used to calculate the estimated sagging and estimated lateral position of each netting segment based on the netting segment, multi-source data and equivalent tension, and to calculate the netting coverage integrity index based on the estimated sagging and estimated lateral position. The risk scoring module is used to visually correct the inversion state composed of the estimated droop and the estimated lateral position based on the credibility of each data source, to obtain the corrected netting state, calculate the mean of the credibility of each data source to obtain the average credibility of the multi-source data, and calculate the comprehensive risk score based on the corrected netting state, the average credibility of the multi-source data, the traction constraint anomaly marker and the netting coverage integrity index. The network closure level module is used to output the network closure status level based on the comprehensive risk score and the average credibility of multi-source data.

[0018] The beneficial effects of this application are as follows: This application acquires multi-source data such as integrated machine, trolley, environmental status and local vision at the beginning of the net sealing operation, and establishes a virtual state chain and net sealing segment based on the multi-source data. This expands the net sealing monitoring object from the single equipment operation status to the overall connection relationship and segmented status of the net rope or net sheet. It can perform real-time quantitative monitoring of the net sealing status under the condition that the sensors are not continuously deployed on the net body, thereby improving the completeness of the net sealing operation status judgment.

[0019] This application estimates the traction force of the integrated machine based on multi-source data and estimates the equivalent tension of each netting segment by combining the virtual state chain of the netting. At the same time, it forms an abnormal traction constraint marker, so that problems such as slippage at the traction end, insufficient anchoring, jamming, or abnormal force transmission can be included in the netting status judgment process. This avoids ignoring the actual force abnormalities of the netting segment based solely on whether the equipment is communicating or moving normally, and improves the pertinence of risk identification.

[0020] This application calculates the estimated sag and estimated lateral position of each netting segment based on the netting segment, multi-source data, and equivalent tension, and further obtains the netting coverage integrity index. This allows key states affecting the protection effect, such as insufficient netting coverage, increased sag, and lateral offset, to be quantified, making it easier for operators to determine whether the netting area meets the protection requirements for the object being crossed.

[0021] This application introduces the credibility of various data sources and performs visual correction on the inversion state based on the credibility. When the video is clear and stable, visual data is used to improve the accuracy of the network closure status judgment. When the video is affected by occlusion, shaking, lighting or rain and fog, the impact of unreliable visual data on the results is reduced, thereby improving the reliability and robustness of the network closure status monitoring results in complex field environments.

[0022] This application calculates a comprehensive risk score based on the corrected grid closure status, the average reliability of multi-source data, and the grid closure coverage integrity index, and outputs the grid closure status level by combining traction constraint anomaly markers. This allows the grid closure operation results to not only reflect whether there is a risk of coverage, sagging, or offset, but also to simultaneously reflect the reliability of the current monitoring data. This provides a clear basis for suspending traction, adjusting the grid closure posture, verifying the pulley anchoring status, or conducting manual verification, thereby improving the safety, reliability, and intelligence level of transmission line crossing grid closure operations. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for real-time monitoring of the network closure status of a transmission line according to this application; Figure 2 This is an example diagram illustrating the calculation of the comprehensive risk score in this application; Figure 3 This is a module example diagram of a real-time monitoring system for the network closure status of a transmission line according to this application. Detailed Implementation

[0024] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0025] A real-time monitoring system and method for the network closure status of transmission lines, including the following embodiments: Example 1: A method for real-time monitoring of the status of power transmission line netting is disclosed. This method integrates intelligent netting operation equipment, pulleys, environmental sensors, and local visual data without relying on the continuous deployment of sensors on the netting ropes or netting panels. It performs real-time inversion of the spatial status of the netting ropes or netting panels and outputs netting coverage integrity index, sagging risk score, lateral offset risk score, and average reliability of multi-source data. This allows operators to make real-time judgments on netting coverage integrity, sagging risk, lateral offset risk, and monitoring reliability.

[0026] The method of this embodiment is applied to the process of power transmission line crossing netting operation; at least one intelligent integrated netting operation machine is configured at the operation site, hereinafter referred to as the integrated machine. The integrated machine travels along the conductor or Dyneema rope and pulls the netting rope or netting sheet. The Dyneema rope is an ultra-high molecular weight polyethylene fiber traction rope; at least one self-locking pulley, netting pulley or fixed connection point is configured at the operation site to participate in the deployment, anchoring, release or unhooking of the netting rope or netting sheet. The self-locking pulley and netting pulley are collectively referred to as pulleys; the operation area crosses roads, railways, rivers, existing lines or other protected objects; the integrated machine or remote control terminal can obtain at least some information from the integrated machine's travel distance, travel speed, wheel pressure, motor current or traction load, communication status, pulley status, ambient wind speed, ambient humidity and video images; before the operation begins, the coordinate relationship between the target netting area, conductor or Dyneema rope path, anchor point, traction point and the object being crossed has been established in a unified coordinate system according to the construction plan.

[0027] The unified coordinate system is obtained from the coordinate data of the construction plan and the site reference points before the operation begins. The site reference points with fixed relationships with the protection boundary of the object to be crossed, the path of the conductor or Dyneema rope, and the anchor points in the construction plan are selected as the coordinate registration references. The three-dimensional coordinates of the site reference points, anchor points, traction points, initial positions of the pulleys, and boundary points of the object to be crossed are obtained by at least one of the following methods: total station, satellite positioning, rangefinder, or manual measurement. The above three-dimensional coordinates are then transformed to the same engineering coordinate system. The unified coordinate system uses the horizontal reference plane determined by the construction plan as the horizontal plane, the vertical direction as the height direction, the projection direction of the planned traction path as the longitudinal direction, and the direction perpendicular to the longitudinal direction in the horizontal plane as the lateral direction. The positions obtained by converting the travel distance of the video module and the integrated machine, the position of the pulley, and the position of the fixed connection point are all expressed through this unified coordinate system. When the local coordinates or image coordinates of the equipment are inconsistent with the unified coordinate system, the coordinate transformation relationship is obtained based on the site reference points or geometric reference points before participating in the netting state inversion.

[0028] In an optional embodiment, when all three-dimensional coordinates cannot be directly obtained on site, a local measurement coordinate system is first established using the obtained coordinates of the site reference points, anchor points, or boundary points of the crossed objects. The local measurement coordinate system is a temporary coordinate expression determined by the distance, azimuth, and elevation difference between the site reference points, used to record the initial position of the pulley, the position of the traction point, and the position of the fixed connection point that have not been directly converted to the unified coordinate system. After the construction plan coordinate data or supplementary measurement data are obtained, the points in the local measurement coordinate system are converted to the unified coordinate system as a whole, and the conversion relationship between the local measurement coordinate system and the unified coordinate system is saved in the basic work information.

[0029] The monitoring system implementing the method of this embodiment includes a multi-source data module, an anomaly marking module, a coverage evaluation module, a risk scoring module, and a network closure level module. The multi-source data module includes a multi-source data acquisition unit, a time synchronization unit, and a data credibility calculation unit. The network closure level module includes a human-machine interaction and alarm unit. The multi-source data acquisition unit receives multi-source data uploaded from an all-in-one machine, a trolley, environmental sensors, and a video module. The time synchronization unit performs timestamp correction and validity screening on data uploaded from different devices. The data credibility calculation unit calculates the credibility of each data source. The human-machine interaction and alarm unit displays the network closure status on a remote control, monitor, or backend terminal, and issues prompts, alarms, or interlock suggestions when an anomaly occurs.

[0030] The time synchronization unit performs validity screening before data enters the processing flow. When the communication status of any field device is interrupted, the communication delay of the field device exceeds the preset allowable delay constant, or the power of the field device is lower than the preset minimum operating power threshold, the current data of the field device is marked as invalid data, and the data is not included in the subsequent network closure status inversion and multi-source data average reliability calculation until the field device is restored to validity. The preset allowable delay constant is determined based on the preset sampling period, communication link delay statistics, and status inversion time synchronization requirements. The preset minimum operating power threshold is determined based on the power supply requirements of the field device to maintain data upload, self-locking, release, and alarm response.

[0031] A method for real-time monitoring of the grid closure status of transmission lines, such as Figure 1 As shown, it includes the following steps: Step 100: Starting from the start of the network closure operation, acquire multi-source data according to the preset sampling period, establish a network closure virtual state chain and network closure segment based on the multi-source data, and calculate the credibility of each data source.

[0032] During monitoring, multi-source data is collected and processed cyclically according to a preset sampling cycle. This multi-source data includes integrated machine status information, trolley status information, environmental status information, local visual information, and basic operational information. The preset sampling cycle is determined before operation based on the data refresh capability of on-site equipment, communication link delay statistics, and alarm response requirements for network closure status, and remains consistent throughout the same network closure operation. At each sampling moment, integrated machine status information, trolley status information, environmental status information, local visual information, and basic operational information are acquired. Integrated machine status information includes the integrated machine number, the machine's travel distance, travel speed, wheel pressure, motor current or traction load, self-locking status, disengagement status, communication delay, and battery level. Trolley status information includes the trolley number, anchoring status, release status, disengagement status, communication status, battery level, and the binding relationship between the trolley and its corresponding network closure segment. To ensure consistency... The description of the validity screening targets includes field equipment such as integrated machines and pulleys, each with corresponding communication latency, power consumption, and communication status; environmental status information includes wind speed, wind direction, temperature, humidity, and rainfall or fog indicators; local visual information includes the positions of conductors, net rope edges, net sheet edges, pulley positions, obstacle positions, image clarity, occlusion ratio, image shake intensity, and degree of lighting anomalies identified in the video image; basic operational information includes the target netting area, the plane where the preset target netting area is located, the span, the planned traction path, the planned traction speed, the planned slope, the planned netting boundary, the binding relationship between the planned netting boundary and the netting segment, the preset maximum allowable sagging, the preset maximum allowable lateral offset, the preset boundary discrete spacing, the preset curve sampling spacing, the preset dry unit length mass of the boundary rope, and the unit length equivalent mass of the netting rope or net sheet boundary rope.

[0033] The preset maximum allowable sag and preset maximum allowable lateral offset are determined before operation based on the protective clearance of the object being crossed, the boundary of the target enclosure area, and construction safety control requirements. The preset dry unit length mass of the boundary rope is determined based on the boundary rope model, factory data, and pre-operation weighing or sampling weighing results. When both factory data and weighing results are available, the pre-operation weighing results of the same batch of boundary ropes are used as the verification basis. When pre-operation weighing cannot be performed, the factory data is used as the source of the preset dry unit length mass of the boundary rope, and the source is written into the operation basic information. The unit length equivalent mass is the equivalent value of the preset dry unit length mass of the boundary rope after correction for current humidity and rainfall or rain / fog conditions. The correction logic is determined by the boundary rope material moisture absorption calibration results and rainfall or rain / fog condition indicators. When moisture absorption calibration is not performed on site, the historical moisture absorption calibration record of the same material and the same structure of the boundary rope is used as the correction basis. When the historical moisture absorption calibration record is unavailable, the unit length equivalent mass is taken as the preset dry unit length mass of the boundary rope, and the humidity correction is marked as unavailable. The correction source is written into the operation closed-loop record.

[0034] At the start of the netting operation, a virtual state chain for netting is established based on the numbering and binding relationships between the integrated machine, pulley, and fixed connection points. This virtual state chain is a topological data structure describing the connection relationships between the integrated machine, pulley, traction point, anchor point, fixed connection point, and the netting rope or netting sheet. It uses the integrated machine, pulley, traction point, fixed connection point, and anchor point as nodes, and the boundaries of the netting rope, netting sheet, or traction boundary as connecting edges. When the integrated machine is in a locked state (self-locking) or the pulley is in an anchored state (anchored), the corresponding node is recorded as a constraint node. When a disengaged or released state occurs, the connecting edges of the corresponding node are updated to ensure that the virtual state chain for netting remains consistent with the actual connection relationships on site.

[0035] The virtual state chain of the closed network is adopted because the connection relationship of the closed network changes with the self-locking, unhooking and release actions; expressing nodes and connecting edges in a topological way can synchronously update the constraint nodes and closed network segments after the connection relationship changes, so that the subsequent tension transmission, coverage boundary and risk location use the same connection relationship.

[0036] In an optional embodiment, the node numbers, connection edge numbers, and netting segment numbers of the netting virtual state chain are generated according to the same numbering rule before the operation. The node numbers are used to identify the integrated machine, pulley, traction point, fixed connection point, and anchor point; the connection edge numbers are used to identify the net rope, net sheet boundary, or traction boundary; and the netting segment numbers are used to identify the calculation unit between adjacent constraint nodes. When rope replacement, pulley replacement, temporary adjustment of connection point, release, or unhooking occurs on site, the historical state record corresponding to the original number is retained, and a new number is generated for the newly formed node, connection edge, or netting segment, so that the historical connection relationship in the operation closed-loop record can be distinguished from the current netting virtual state chain.

[0037] Based on the planned traction path of the construction plan or the results of on-site calibration, a spatial path for the conductor or Dyneema rope is established in advance. When establishing the spatial path, the path control points given in the construction plan or the path sampling points obtained by on-site calibration along the conductor or Dyneema rope are uniformly converted into three-dimensional coordinates under the aforementioned unified coordinate system, and each path sampling point is saved in correspondence with the travel distance of the integrated machine. The spatial path takes the travel distance of the integrated machine as input and the three-dimensional coordinates under the unified coordinate system as output. When a set of path sampling points arranged in the path order is obtained by calibration along the planned traction path, each path sampling point contains the corresponding travel distance and three-dimensional coordinates. For any travel distance of the integrated machine falling between two adjacent path sampling points, the travel distance is used first. Subtract the walking distance corresponding to the previous path sampling point, and then divide by the difference between the walking distance corresponding to the next path sampling point and the walking distance corresponding to the previous path sampling point to obtain the interpolation ratio of the walking distance in the current path segment; then subtract the three-dimensional coordinates of the previous path sampling point from the three-dimensional coordinates of the next path sampling point to obtain the three-dimensional coordinate increment of the current path segment; multiply the interpolation ratio by the three-dimensional coordinate increment, and then add it to the three-dimensional coordinates of the previous path sampling point to obtain the estimated position of the integrated machine in the unified coordinate system; when the path control points given by the construction plan cannot cover the walking interval that needs to be inverted, the walking interval is supplemented by the path sampling points obtained by on-site calibration, and the source of the supplementation is written into the basic work information.

[0038] The walking distance of the integrated machine is used as the spatial path input because this data is continuously output by the integrated machine and directly corresponds to the traction deployment progress; by interpolating the path sampling points to convert the three-dimensional coordinates, the walking data can be converted into the position of the constraint node participating in the inversion.

[0039] The estimated positions of each integrated machine, the positions of the anchored pulleys, and the positions of the fixed connection points are used together as constraint nodes for the netting state inversion. According to the connection edges in the virtual state chain of the netting, the boundaries of the net ropes, net sheets, or traction boundaries between adjacent constraint nodes are divided into netting segments, forming a set of netting segments composed of several netting segments. The position of the anchored pulley is determined by converting the coordinates of the anchor point to which the pulley is bound, the initial position of the pulley calibrated before operation, or the visual positioning results to a unified coordinate system. The position of the fixed connection point is determined by converting the coordinates of the connection point in the construction plan and the measurement results of the on-site benchmark point to a unified coordinate system. The numbers and three-dimensional coordinates of the constraint nodes at both ends of each netting segment are recorded for subsequent calculation of the geometric parameters of the netting segment.

[0040] The reason for using the netting segment as the calculation unit is that the netting rope or netting sheet has a clear geometric boundary and force transmission path between adjacent constraint nodes; the sag, lateral position and risk source of each netting segment can be obtained by calculating by segment.

[0041] The binding relationship between the planned perimeter of the net enclosure and the net enclosure segment is determined before operation based on the planned perimeter number, net enclosure segment number, traction deployment sequence, and connection point number in the construction plan. This binding relationship records the planned boundary curve corresponding to each net enclosure segment and the path sequence interval of the net enclosure segment on the planned boundary curve. If the construction plan does not directly provide the correspondence between the net enclosure segment number and the planned perimeter number, the planned initial projection points of the constraint nodes at both ends of the net enclosure segment are matched with the planned perimeter control points. The planned initial projection points are the points obtained after projecting the constraint nodes at both ends of the net enclosure segment to the plane where the preset target net enclosure area is located according to the planned position in the construction plan. The planned boundary curve interval corresponding to the planned initial projection points at both ends of the net enclosure segment is selected according to the traction deployment sequence to form the binding relationship between the planned perimeter of the net enclosure and the segment, and the matching source is written into the basic operation information.

[0042] This binding relationship ensures that the real-time netting segment corresponds to the planned netting boundary in the construction plan, thus using the same target boundary when calculating the lateral offset risk characteristics and netting coverage integrity index.

[0043] After establishing the virtual state chain for network closure, the credibility of each data source is calculated for the multi-source data that has been filtered for validity, including the credibility of non-visual data sources and visual data sources, in order to avoid making judgments based on normal data when the video is unclear, communication is delayed, or the sensor is abnormal.

[0044] For any non-visual data source, including data from the all-in-one machine's movement, wheel pressure, motor current or traction load, trolley status, and environmental sensing, a priori calculation is first performed on the data source using the closed-loop virtual state chain to obtain a predicted value. When the data source has a historical sequence, the predicted value is taken as the state value of the data source at the previous valid moment. When the data source does not have a valid historical sequence but has adjacent similar devices, the predicted value is taken as the value obtained by inferring the state of the adjacent devices according to the connection relationship. Specifically, based on the nodes, connection edges, and binding relationships in the closed-loop virtual state chain, the similar devices adjacent to the device corresponding to the data source are determined. When the corresponding adjacent similar device can be uniquely determined based on the connection edges and binding relationships, and the adjacent similar device has a valid state value of the same type as the data source after validity screening at the current sampling moment, the valid state value is taken as the predicted value of the data source. When the corresponding adjacent similar device cannot be uniquely determined, or the corresponding adjacent similar device does not have a valid state value of the same type as the data source at the current sampling moment, the data source is treated as... The previous measured value is used as the initial predicted value, and the sampling time is marked as the data source initialization time. When the data source has neither a historical sequence nor the effective status of adjacent similar equipment, the current measured value of the data source is used as the initial predicted value, and the sampling time is marked as the data source initialization time. The reliability of the initialization time is not used to determine equipment abnormality, but to form a subsequent prediction benchmark. The error vector is obtained by subtracting the predicted value vector from the current measured value vector of the data source. The normalized residual square of the data source is obtained by multiplying the transpose of the error vector, the inverse matrix of the preset error covariance matrix of the data source, and the error vector in sequence. The preset error covariance matrix is ​​obtained by statistically analyzing the measurement error of the data source during the calibration phase and is saved according to the data source type. When the calibration samples are insufficient to form a complete matrix, a diagonal matrix is ​​established with the measurement resolution of the data source, the equipment accuracy level, and the static sampling error before operation, and the matrix is ​​made invertible by adding a preset data source matrix regularization term. The preset data source matrix regularization term is determined according to the numerical resolution of the data source.

[0045] Normalized residual squares are used because the dimensions and error distributions of different data sources are inconsistent. After normalization by pre-setting the error covariance matrix, the deviation of the integrated machine's walking, wheel pressure, motor current or traction load, trolley status and environmental sensor data can be compared under the same confidence level.

[0046] The reliability of the non-visual data source is calculated as the product of two natural exponential factors. The exponent of the first natural exponential factor is the negative of the squared normalized residual of the data source divided by two. The exponent of the second natural exponential factor is the negative of the data time delay of the data source from the current moment divided by a preset allowable delay constant. According to this calculation method, the reliability of the non-visual data source decreases as the squared normalized residual or the data time delay increases. The reliability value is greater than zero and does not exceed one. The data time delay from the current moment is the difference between the current sampling moment of the system and the sampling timestamp carried by the data source data packet. The time synchronization unit first corrects the sampling timestamp according to the system clock, device clock synchronization record, or data packet reception moment. Then, the difference between the corrected sampling timestamp and the current sampling moment is used to obtain the data time delay of the data source from the current moment. When there are multiple valid data packets of the same data source in the current sampling period, the valid data packet with the corrected sampling timestamp closest to the current sampling moment is used in the calculation. When the sampling timestamp is missing, the data packet reception moment is used as the sampling timestamp of the data source, and the data source is marked as a timestamp substitute data source.

[0047] For example, the credibility of a non-visual data source can be represented by the following formula: ,and ;in, Indicates the non-visual data source number. Indicates the current sampling time sequence number. Indicates the first The non-visual data source in the first The reliability of non-visual data sources at each sampling time. This represents the normalized squared residual of the non-visual data source. This represents the current measurement vector of the non-visual data source. This represents the vector of predicted values ​​from the non-visual data source. This represents the preset error covariance matrix of the non-visual data source. This indicates the time delay of the non-visual data source from the current moment. This indicates the preset allowable delay constant. This represents the natural exponential function.

[0048] For video data, the video recognition module first determines the recognition region in each frame of video image. The recognition region is jointly determined by the detection boxes of wires, rope edges, mesh edges, pulleys or obstacles, segmentation masks, and the projected area of ​​the planned netting boundary. When there are multiple recognition targets in the same frame, the union of the regions corresponding to each recognition target is taken as the recognition region of that frame. The positions of wires, rope edges, mesh edges, pulleys, and obstacles obtained by video recognition are obtained by the video recognition module within the recognition region through target detection, semantic segmentation, or edge extraction, and are bound according to the target category marked before the operation and the nodes and connecting edges in the netting virtual state chain. Recognition results that cannot be bound to the netting virtual state chain are only used as screen prompts and do not participate in 3D state correction. The recognition region constraint is used because there are backgrounds, objects being crossed, and construction equipment in the netting image. Limiting the detection range first can make the positions of wires, rope edges, mesh edges, and pulleys correspond to objects in the netting virtual state chain, avoiding irrelevant image targets from entering the state correction.

[0049] The target detection, semantic segmentation, or edge extraction algorithms used in the video recognition module are trained on samples before the operation. During training, video image samples covering the lighting conditions, rain or fog conditions, occlusion conditions, and image shaking conditions of the operation site are collected. Detection boxes and segmentation masks are labeled for wires, rope edges, net edges, pulleys, and obstacles in the samples according to the target categories determined before the operation, forming a labeled training sample set. Training samples and validation samples are then separated from this set. The training samples are then input into the video recognition module to obtain the module's recognition results for the positions of wires, rope edges, net edges, pulleys, and obstacles. The deviation between the recognition results and the corresponding labels forms the training deviation, and the model parameters of the video recognition module are adjusted in reverse according to the training deviation, so that the training deviation decreases with the number of iterations. After each round of training, validation samples are used... The sample verification checks the discrepancy between the recognition results and the annotations. When the verification discrepancy no longer decreases with the number of iterations, training is stopped and the model parameters obtained from this training are saved as the basis for the video recognition module to identify the positions of wires, rope edges, mesh edges, pulley positions, and obstacle positions during the operation. When the target category, video module installation angle, or operation scene changes, the corresponding labeled samples are added and the training is retrained. The source of the training samples and the version of the model parameters are written into the basic information of the operation. Pre-operation sample training is used because the image features of wires, rope edges, mesh edges, pulleys, and obstacles in the recognition area change with the on-site lighting, rain, fog, and occlusion conditions. Labeling samples according to the on-site conditions and iteratively adjusting the model parameters can make the recognition results of the video recognition module correspond to the nodes and connecting edges in the virtual state chain of the net enclosure.

[0050] Image sharpness, occlusion ratio, image shake intensity, and lighting anomaly level are all normalized parameters obtained in real time by the video recognition module, with values ​​ranging from zero to one. The normalization calibrator is calibrated by the video recognition module before operation. Image sharpness is calculated by combining the grayscale texture response, edge gradient response, Laplacian focus evaluation value, and camera focus state of the recognition area. Before operation, the video recognition module collects statistical results of sharp and out-of-focus frames to form a sharpness normalization interval. During operation, the above-mentioned sharpness evaluation value of the current frame is converted to this normalization interval, and the conversion result is truncated to obtain the image sharpness. When the camera focus state is unavailable, the grayscale texture response, edge gradient response, and Laplacian focus evaluation value are used for conversion, and the missing focus state is written into the visual data record. Image sharpness, occlusion ratio, image shake intensity, and lighting anomaly level are selected as inputs because these four factors correspond to focus state, visible area, camera stability, and exposure conditions, respectively, and directly determine whether the visual observation value can reflect the edge of the rope, the edge of the net, and the position of the trolley.

[0051] The occlusion ratio is obtained by the ratio of the occlusion mask area to the recognition area area within the recognition area; the occlusion mask is determined by the obstacle segmentation results, the foreign object occlusion detection results, and the continuity gap of the target edge within the recognition area; when the recognition area area is zero, the video data is marked as unusable and does not participate in the visual credibility calculation; the image shake intensity is calculated by the feature point displacement, global translation, rotation, and motion blur between consecutive frames; in adjacent frames, guide wire reference points, trolley reference points, or background stable points are matched, and the target motion components generated by the movement of the integrated machine and the normal traction of the sealing net rope or net sheet are removed. Then, the remaining global motion components are normalized according to the statistical results of the static stable frames and shake frames calibrated before the operation to obtain the image shake intensity.

[0052] The degree of lighting anomaly is calculated by combining the proportion of overexposed pixels, underexposed pixels, and strong reflective mask within the identification area, as well as the deviation of the brightness histogram from the normal exposure statistical range before the operation. Before the operation, the video recognition module obtains the statistical results of normal exposure frames, overexposed frames, underexposed frames, and strong reflective frames to form a lighting anomaly normalization interval. During the operation, the exposure anomaly evaluation of the current frame is converted to this normalization interval, and the conversion result is truncated to obtain the degree of lighting anomaly. When there are rain or fog indicators in the environmental status information, the edge gradient attenuation caused by rain and fog scattering is included in the image sharpness calculation, and the detection results of raindrop reflection, fog scattering, and local strong reflection are included in the lighting anomaly degree calculation.

[0053] In an optional embodiment, when the visual calibration sample before the operation is unavailable, historical visual calibration records of the same model video module, the same installation angle, and the same operation scenario are used as the normalization basis, and this source is written into the operation basic information; when the historical visual calibration records are also unavailable, the video data is only used for auxiliary prompts on the display end and does not participate in the main calculations for three-dimensional state correction and net coverage integrity judgment; firstly, the result after subtracting the occlusion ratio, the result after subtracting the image shaking intensity, and the result after subtracting the degree of lighting abnormality are multiplied by the image clarity to obtain the initial value of visual credibility; then, the initial value of visual credibility is truncated to obtain the visual credibility. The truncating process includes: if the initial value of visual credibility is less than zero, the visual credibility is zero; if the initial value of visual credibility is greater than one, the visual credibility is one; otherwise, the visual credibility is the initial value of visual credibility itself.

[0054] Step 200: Estimate the traction force of the integrated machine based on multi-source data, estimate the equivalent tension of each sealing segment based on the traction force of the integrated machine and the virtual state chain of the sealing network, and form a traction constraint anomaly marker.

[0055] The traction force of the integrated machine is estimated based on the traction load sensor or motor current model, and the equivalent horizontal tension of each sealing segment is further estimated. The traction force of the integrated machine refers to the output force of the integrated machine along the traction direction when a certain integrated machine pulls the sealing net rope or net sheet and is used for the calculation of the sealing state. The measured traction load output by the traction load sensor and the estimated value of the traction force of the integrated machine calculated by the motor current model are all converted into the traction force of the integrated machine and then used in subsequent calculations.

[0056] In an optional embodiment, if a certain integrated machine has a traction load sensor, the measured traction load of the integrated machine is directly used as the traction force of the integrated machine for calculating the netting state; if a certain integrated machine does not have a traction load sensor, the traction force of the integrated machine is estimated based on the motor current; specifically, the motor current is first subtracted from the preset no-load current to obtain the effective current component of the corresponding effective output torque; then, the effective current component is multiplied by the preset motor torque constant to obtain the equivalent output torque converted to the output end of the drive wheel; finally, the equivalent output torque is divided by the preset equivalent radius of the drive wheel to obtain... The estimated traction force of the integrated machine is obtained from the current, and this estimated traction force is used as the traction force of the integrated machine for the calculation of the net sealing state; the preset no-load current is the current corresponding to the motor maintaining its own rotation under no-load conditions, which is determined by the integrated machine loading test or the on-site calibration results before operation; the preset motor torque constant is the equivalent torque constant converted to the output end of the drive wheel, and its value has taken into account the motor shaft torque constant, reduction ratio, preset transmission efficiency and wheel-rope contact efficiency, which is determined by the integrated machine factory data, loading test or on-site calibration results before operation under constant speed, planned traction speed and planned slope conditions.

[0057] The planned traction speed is the traction speed of the integrated machine set in the construction plan; the planned gradient is the gradient of the planned traction path along the walking direction within the netting operation area. Both are determined by the construction plan and the path calibration results before operation and are written into the basic operation information. The preset transmission efficiency is determined based on the factory data and loading test of the integrated machine's transmission mechanism, and the wheel-rope contact efficiency is determined based on the drive wheel material, working rope diameter, wheel pressure, and on-site traction calibration results. Both are saved with the integrated machine model and remain consistent in the same netting operation. The preset equivalent radius of the drive wheel is the effective radius of the drive wheel under the working rope diameter. The effective radius is determined by the drive wheel's factory data, the working rope diameter, and the on-site calibration results before operation, and is saved along with the integrated machine model. The integrated machine traction force estimated by the motor current is used as the estimated value of the integrated machine traction force within the calibrated working condition range for the same net sealing operation. When the integrated machine's walking speed or line gradient exceeds the calibrated working condition range and is in a dynamic working condition of acceleration / deceleration, climbing, or descending, the integrated machine traction force estimated by the motor current is marked as the dynamic working condition estimate value, and when the integrated machine has a traction load sensor, the measured traction load is preferentially used as the integrated machine traction force.

[0058] In an optional embodiment, when the same integrated device simultaneously has traction load sensor and motor current data, the measured traction load is used as the main traction force data of the integrated device, and the estimated traction force value of the integrated device obtained from the motor current model is used as the consistency verification data. If the absolute value of the difference between the measured traction load and the estimated traction force value of the integrated device exceeds the preset traction force consistency threshold, the traction data source of the integrated device is marked as an abnormal traction force consistency data source, the measured traction load is retained to participate in the equivalent tension calculation, and the credibility of the traction data source of the integrated device is adjusted according to the credibility calibrator of the data credibility calculation unit. The preset traction force consistency threshold is determined based on the measurement resolution of the traction load sensor, the calibration error of the motor current model, and the allowable traction force fluctuation range under the planned traction conditions.

[0059] The method of prioritizing traction load sensors and supplementing with motor current models is adopted because the measured traction load directly corresponds to the output force at the traction end, while the motor current can still reflect the output torque even without traction load sensors; this combination enables integrated machines with different configurations to form a unified traction force.

[0060] For any given netting segment, the set of traction points bound to that segment is determined according to the virtual state chain of the netting, and the equivalent horizontal tension of the segment is estimated from this set of traction points, hereinafter referred to as equivalent tension. Equivalent tension is the engineering equivalent estimate of the horizontal tension component of the flexible cable used in the sag calculation, not the precise tension within the netting segment. For each traction point, the effective traction force is first determined. The effective traction force of a traction point refers to the force value obtained after the traction point transmits the force to the target netting segment and is reduced along the force transmission path. When the traction point is directly provided with traction by an integrated machine, the effective traction force of the traction point is taken as the traction force of the integrated machine. When the traction force of the integrated machine is transmitted to the netting segment along the virtual state chain through the force transmission nodes, it is successively reduced according to the preset friction reduction coefficient of each force transmission node to obtain the effective traction force of the traction point transmitted to the netting segment. A force transmission node refers to a node located at the traction point... The pulley or connection point between the lead point and the sealing section, which participates in the transmission of traction force of the integrated machine, has a preset friction reduction coefficient for the force transmission node. This coefficient is determined by the node type, rope sheave angle, contact structure, and on-site traction calibration results of the corresponding force transmission node, and is saved along with the force transmission node during the same sealing operation. If the traction force of the integrated machine passes through multiple force transmission nodes sequentially along the force transmission path, the traction force of the integrated machine before entering the first force transmission node is taken as the initial force value. Each time the force transmission node is passed, the force value entering the node is multiplied by the preset friction reduction coefficient of the corresponding force transmission node to obtain the reduced force value. The reduced force value is used as the force value leaving the node and entering the next segment of the force transmission path. After all force transmission nodes have been reduced, the resulting force value is used as the effective traction force of the traction point. When there are no force transmission nodes in the force transmission path, the effective traction force of the traction point is equal to the traction force of the integrated machine.

[0061] For example, the effective traction force and equivalent tension at the traction point can be expressed by the following formula: , ;in, Indicates the network segment number, Indicates the traction point number. Indicates the current sampling time sequence number. Indicates the first The traction point towards the first The effective traction force transmitted by each sealed section of the traction point Indicates the first The traction force of the integrated machine corresponding to each traction point Indicates the first The traction point to the first The set of force transmission nodes that participate in force transmission between each network segment. Indicates the force transmission node number, Indicates the first Preset friction reduction coefficient for each force transmission node Indicates the relationship with the first The set of traction points that are bound to each netting segment and participate in the equivalent tension calculation. Indicates the first The effective traction force direction of the first traction point is related to the first... The angle between the tension direction of each section of the netting and the horizontal plane. Indicates the first The equivalent tension of each blocked segment. This indicates that the maximum value among the comparison objects within the parentheses is taken. Represents the cosine function. Indicates the angle Specific directional projection coefficients.

[0062] The friction reduction at the force transmission nodes is adopted because the traction force of the integrated machine will be lost after passing through the pulley or connection point; by reducing the force node by node according to the force transmission path, the equivalent tension of the target sealing section can correspond to the on-site connection structure.

[0063] When determining the effective traction force direction of a traction point, first obtain the sequence of connecting edges transmitting force from the traction point to the target netting segment in the netting virtual state chain, then take the last connecting edge before the force transmission path enters the target netting segment as the direction determination edge; when there are no intermediate force transmission nodes, the direction determination edge is the connecting edge between the traction point and the constraint node at the access end of the netting segment; subtract the coordinates of the two nodes in the unified coordinate system of the direction determination edge to obtain the direction of the direction determination edge in three-dimensional space, and project this direction onto the horizontal plane as the effective traction force direction of the traction point; the direction of the netting segment is determined by the direction from the starting constraint node to the ending constraint node of the netting segment, and the starting constraint node and the ending constraint node are determined according to the sequence of connecting edges in the netting virtual state chain; the coordinates of the two constraint nodes are... Subtract the values ​​from the target values ​​and project them onto the horizontal plane to obtain the direction of the sealing section. Then, determine the angle between the effective traction force direction of the traction point and the direction of the sealing section on the horizontal plane, and multiply the effective traction force of the traction point by the cosine of the angle to obtain the horizontal projection component of the traction point along the horizontal projection direction of the sealing section. When the cosine of the angle is greater than zero, it indicates that the effective traction force of the traction point acts along the tension direction of the sealing section, and the horizontal projection component of the traction point is included in the equivalent tension. When the cosine of the angle is less than or equal to zero, it indicates that the effective traction force of the traction point does not act along the tension direction of the sealing section, and the horizontal projection component of the traction point is not included in the equivalent tension. Sum the horizontal projection components of all traction points that are included in the equivalent tension for the sealing section to obtain the equivalent tension of the sealing section.

[0064] In an optional embodiment, when the same network segment is synchronously pulled from both ends by two traction points, a local tensioning direction is determined for each of the two traction points. The local tensioning direction is the horizontal projection direction from the constraint node at the access end of the traction point to the interior of the target network segment. It is only used to determine whether the traction point has a tensioning effect on the target network segment and does not change the numbering order of the constraint nodes at the beginning and end of the network segment. For each traction point, the effective traction force direction of the traction point is compared with its local tensioning direction. The horizontal projection component acting along the local tensioning direction is included in the equivalent tension, while the horizontal projection component not acting along the local tensioning direction is not included in the equivalent tension.

[0065] The horizontal projection component is used in the equivalent tension calculation because the sag calculation uses the horizontal tension component of the flexible cable; only the component acting in the tensioning direction along the sealing section is included, which avoids the reverse or lateral components being used as tension force.

[0066] When the equivalent tension of a certain meshing segment is greater than or equal to the preset minimum effective tension, the equivalent tension is used in step 300 to calculate the estimated sag, estimated lateral position, and lateral offset risk characteristics. When the equivalent tension of a certain meshing segment is less than the preset minimum effective tension, the meshing segment is marked as a suspected abnormal stress segment, and the preset minimum effective tension is used instead of the equivalent tension in the calculation of the estimated sag, estimated lateral position, and lateral offset risk characteristics of the meshing segment to avoid the calculation denominator approaching zero and causing the results to diverge. At the same time, the results of the estimated sag, estimated lateral position, and lateral offset risk characteristics of the meshing segment are only used as trend judgments. The estimated tension is not used as a basis for precise calculations; the minimum effective tension is determined based on the numerical stability requirements of the lower limit of horizontal tension and sag required to maintain the tension of the netting rope or the boundary rope of the netting; since the internal tension distribution of the netting section changes after passing through the force transmission node, anchor point and the boundary of the netting, the equivalent tension is the engineering equivalent estimate after being reduced and transmitted along the virtual state chain of the netting through the force transmission node; when the netting section is marked as a suspected abnormal stress section or a suspected stuck section, its equivalent tension estimate is not used as a basis for precise calculation of the estimated sag, estimated lateral position and lateral offset risk characteristics, but is only used for stress trend judgment and abnormal indication.

[0067] Simultaneously, the friction margin between the integrated machine and the conductor or Dyneema rope is calculated using wheel pressure. For any integrated machine, the wheel pressure is multiplied by the preset friction coefficient or the preset equivalent friction parameter obtained through on-site calibration, and then the traction force of the integrated machine is subtracted to obtain the friction margin of the integrated machine. When the friction margin of an integrated machine is less than the preset friction margin threshold, it is determined that the corresponding integrated machine has a risk of slippage or insufficient anchoring, and the corresponding integrated machine node is marked in the virtual state chain of the netting. The preset friction coefficient and preset equivalent friction parameter are determined by the surface condition of the conductor or Dyneema rope, the wheel surface material, and the on-site traction calibration results. During on-site traction calibration, the friction margin is calculated when the integrated machine brake is released and the drive wheel is... With the conductor or Dyneema rope in static frictional contact, the traction output torque is gradually increased. The maximum traction force and corresponding wheel pressure before the drive wheel begins to slip relative to each other are recorded. The maximum traction force is divided by the corresponding wheel pressure to obtain the preset equivalent friction parameter for this working condition, and is saved according to the surface condition of the conductor or Dyneema rope. The preset friction margin threshold is determined based on the margin requirement for the integrated machine to maintain effective adhesion in locked, traction, and braking states. The friction margin is calculated using wheel pressure and the integrated machine's traction force because wheel pressure characterizes the normal constraint of the drive wheel on the conductor or Dyneema rope, while the integrated machine's traction force characterizes the traction requirement. Comparing the two can determine whether the adhesion at the traction end meets the current traction state.

[0068] Furthermore, the force state of the sealed section is judged by combining the travel distance and travel speed of the integrated machine; when the increase of the traction force of the integrated machine corresponding to a certain sealed section exceeds the preset traction force increment threshold in adjacent sampling times, while the increase of the travel distance of the integrated machine in adjacent sampling times is less than the preset travel increment threshold, the sealed section is marked as a suspected jammed section; the preset traction force increment threshold is determined based on the measurement resolution, calibration error, and allowable traction force variation range of the integrated machine under the planned traction conditions obtained from the traction load sensor or motor current model; the preset travel increment threshold is based on the preset sampling period, planned traction speed, and... The resolution of the integrated machine's walking distance measurement is determined; when the change in wheel pressure of the integrated machine corresponding to a certain sealing section within the preset speed fluctuation identification window does not exceed the preset wheel pressure change threshold, and the walking speed of the integrated machine exhibits repeated changes exceeding the preset speed fluctuation judgment threshold within the same window, the sealing section is also marked as a suspected jamming section; the preset speed fluctuation identification window is determined based on the preset sampling period and the integrated machine's control response time, the preset wheel pressure change threshold is determined based on the wheel pressure sensor measurement resolution and static sampling error, and the preset speed fluctuation judgment threshold is determined based on the planned traction speed, walking speed measurement resolution, and the integrated machine's speed regulation control error.

[0069] If the traction force increases but the travel distance does not increase, or if the travel speed changes back and forth but the wheel pressure remains unchanged, these conditions correspond to the working conditions where the traction output cannot be converted into the displacement of the sealing section. Therefore, this condition is used to identify suspected stuck sections.

[0070] In step 200, nodes of the integrated machine with a risk of slippage or insufficient anchoring, suspected jamming sections, and suspected abnormal force sections are marked. The purpose is to identify the sources of anomalies caused by the attachment of the traction end, anchoring constraints, or force transmission of the sealing section to the inversion of the sealing state. The above markings together form traction constraint anomaly markings. Traction constraint anomaly markings serve as inputs for the comprehensive risk scoring in step 400, alarm location in step 500, anomaly handling, and operation closed-loop recording. Among them, nodes of the integrated machine with a risk of slippage or insufficient anchoring are used to locate integrated machines that need to be reviewed, suspected jamming sections are used to locate sealing sections that may have mechanical obstruction, and suspected abnormal force sections are used to locate sealing sections where the equivalent tension cannot be directly used as the basis for estimating sagging, estimating lateral position, and calculating lateral offset risk characteristics.

[0071] Step 300: Based on the netting segments, multi-source data, and equivalent tension, calculate the estimated sag and estimated lateral position of each netting segment, and calculate the netting coverage integrity index based on the estimated sag and estimated lateral position.

[0072] Based on the constraint node positions at both ends of the enclosure segment, the equivalent mass per unit length of the enclosure rope or the boundary rope of the mesh, the equivalent tension, and the wind speed, the estimated sag, estimated lateral position, and wind-induced additional lateral displacement of the enclosure segment are calculated. The lateral offset risk characteristics are obtained based on the estimated lateral position, the planned lateral position, and the wind-induced additional lateral displacement. The estimated lateral position is the current lateral reference coordinate of the enclosure segment in the plane of the preset target enclosure area. The planned lateral position is the lateral reference coordinate of the planned enclosure boundary specified in the construction plan at the corresponding longitudinal position of the enclosure segment. Both the estimated lateral position and the planned lateral position are determined after being mapped to the plane of the preset target enclosure area in a unified coordinate system, and a unified longitudinal coordinate and lateral coordinate caliber is adopted.

[0073] The inputs are the constrained nodes at both ends, the equivalent mass per unit length, the equivalent tension, and the wind speed, because these data determine the span, self-weight, tension state, and lateral environmental effects of the enclosure section, and are the direct factors affecting the calculation of sagging and lateral offset.

[0074] Before the operation, the plane where the pre-set target enclosure area is located is determined by the protection boundary of the object to be crossed, the projection direction of the conductor or Dyneema rope path, and the unified coordinate system in the construction plan. When mapping the three-dimensional coordinates to the plane where the pre-set target enclosure area is located, the projection is along the normal of the plane, and the longitudinal coordinates, transverse coordinates, and topological relationship of the coverage boundary are retained after the projection. The longitudinal coordinates are the coordinates along the planned traction path direction, and the transverse coordinates are the coordinates perpendicular to the planned traction path direction.

[0075] First, using the coordinates of the constraint nodes at both ends of any netting segment in step 100, calculate the horizontal distance between the constraint nodes at both ends of the netting segment. The horizontal distance is the projected length of the line connecting the constraint nodes at both ends on the horizontal plane. The estimated sag of the netting segment is defined as the vertical distance between the midpoint of the netting segment and the line connecting the constraint nodes at both ends. When there is a height difference between the constraint nodes at both ends, the sag is still measured relative to the line in the vertical direction and converted to the reference plane where the object being crossed is located when assessing the protective clearance for the object being crossed. When the netting boundary is a non-collinear broken line, an oblique crossing, or composed of multiple netting segments, calculate the horizontal distance of each netting segment segment by segment according to the connecting edge in the virtual state chain of the netting, and then project and sum the horizontal distances of each segment along the traction deployment direction to obtain the total horizontal projection length of the current netting deployment path. This total horizontal projection length is used to check the progress of the netting deployment and is not required to be strictly equal to the straight-line span of the object being crossed.

[0076] The equivalent mass per unit length is the equivalent value of the dry unit length mass of the preset boundary rope after correction for current humidity and rainfall or fog. The preset wind load correction coefficient and preset wind-induced offset weight are the calibration values ​​corresponding to the current temperature and humidity. The preset wind load correction coefficient is the equivalent calibration coefficient of wind-induced additional sag, which is calibrated by the correspondence between the field-measured sag and wind speed components, horizontal distance, and equivalent tension, and has a compensation dimension that makes the dimensions of wind load and field correction terms length. During calibration, the effects of air density, drag coefficient, equivalent windward diameter of the boundary rope, and wind-received projected area on additional sag have been converted into a unified equivalent calibration coefficient. The preset wind-induced offset weight is the equivalent calibration coefficient of wind-induced additional lateral displacement, which is calibrated by the correspondence between the field-measured lateral displacement and wind speed components, equivalent tension, and horizontal distance, and has a compensation dimension that makes the dimensions of wind-induced additional lateral displacement length length.

[0077] The aforementioned preset calibration coefficients are grouped and saved as discrete calibration values ​​according to temperature and humidity. The calibration samples cover the expected temperature range, humidity range, wind speed component range, and equivalent tension range at the work site. When the current temperature or humidity is between adjacent calibration conditions, linear interpolation is performed between the calibration values ​​of adjacent calibration conditions according to temperature and humidity to obtain the parameter value corresponding to the current condition, so as to maintain the consistency of parameter source under the same condition. When the current temperature, humidity, wind speed component, or equivalent tension exceeds the coverage range of the calibration samples, the parameter value of the most recent calibration condition is used in the calculation, and the relevant calculation of the netting section is marked as a condition exceeding the calibration range. The results under this condition are only used as a trend indication and are not used as the main basis for judging the integrity of the netting coverage.

[0078] In an optional embodiment, when any of the wind speed data, wind direction data, or the azimuth of the netted section is unavailable, the wind speed component perpendicular to the direction of the netted section is marked as unavailable. The wind load and on-site correction items are replaced by a preset wind-induced sagging amount, and the wind-induced additional lateral displacement is replaced by a preset wind-induced lateral displacement amount. The preset wind-induced sagging amount and the preset wind-induced lateral displacement amount are determined before the operation based on the allowable wind speed range required by the construction plan, the preset maximum allowable sagging amount, the preset maximum allowable lateral displacement amount, and historical qualified operation records, and are written into the operation basic information. When the above replacement amounts are enabled, the corresponding netted section is marked as a wind data replacement condition, and the influence of the environmental data source reliability is retained in the multi-source data average reliability.

[0079] In an optional embodiment, for a net enclosure segment that meets the preset parabolic approximation conditions, a parabolic approximation model is used to estimate the sag. The preset parabolic approximation conditions refer to the ratio of the sag of the net enclosure segment to the horizontal distance being within the applicable range of the preset parabolic approximation. The applicable range of the preset parabolic approximation is obtained before operation or during the model calibration stage. Calibration samples corresponding to the boundary rope material, span, equivalent tension range, and on-site wind speed range of this operation are selected, and the midpoint sag of the same sample is calculated using both the parabolic approximation model and the catenary model, respectively, and verified against the on-site measured sag or manually confirmed sag. The ratio between the sag and the horizontal distance is used as the basis for range division, and the continuous proportional range in which the deviation of the parabolic approximation model from the catenary model and the measured results does not exceed the preset model deviation allowable value is saved as the applicable range of the preset parabolic approximation. The preset model deviation allowable value is determined based on the preset maximum allowable sag, the protective clearance of the crossed object, and the construction safety control requirements, and is written into the basic operation information.

[0080] The parabolic approximation model is adopted because, within the pre-defined approximation range, this model can obtain the midpoint sag using horizontal distance, equivalent tension, and gravity per unit length, and the calculation process is suitable for real-time execution according to the sampling period.

[0081] When determining the pre-defined parabolic approximation conditions, the estimated sag from the previous valid time step is used first. If the estimated sag from the previous valid time step is unavailable or the initial calculation is performed, the equivalent mass per unit length at the current sampling time, the pre-defined gravitational acceleration, the horizontal distance of the netting segment, the equivalent tension of the netting segment, wind load, and on-site correction terms are used to first generate an initial sag using the parabolic approximation model. This initial sag is only used for the initial determination of the pre-defined parabolic approximation conditions. If the ratio of the initial sag to the horizontal distance is within the applicable range of the pre-defined parabolic approximation, the estimated sag for the netting segment is calculated using the parabolic approximation model. If the ratio of the initial sag to the horizontal distance is not within the applicable range of the pre-defined parabolic approximation, the estimated sag for the netting segment is calculated using the parabolic approximation model. When the approximation of the linear approximation is applicable, the catenary model is used to calculate the estimated sag of the enclosure segment. This model is derived from the mechanical approximation of the flexible cable under uniformly distributed load. When calculating the estimated sag of the enclosure segment under the parabolic approximation model, the equivalent mass per unit length, the preset gravitational acceleration, and the square of the horizontal distance of the enclosure segment are multiplied together, and then divided by eight times the equivalent tension of the enclosure segment to obtain the sag caused by its own weight. Then, wind load and on-site correction terms are added to obtain the estimated sag of the enclosure segment. The preset gravitational acceleration adopts the engineering calculation parameters or the unified verification value of the project corresponding to the work location. The product of the equivalent mass per unit length and the preset gravitational acceleration is used as the gravity per unit length.

[0082] For example, when using a parabolic approximation model, the estimated droop can be expressed by the following formula: , ;in, Indicates the network segment number, Indicates the current sampling time sequence number. Indicates the first The first blocked segment in the The estimated droop at each sampling time point is obtained using a parabolic approximation model. Indicates the first The equivalent mass per unit length of each blocked segment. Indicates the preset gravitational acceleration. Indicates the first The horizontal distance between the nodes at both ends of each network segment is constrained. Indicates the first The equivalent tension of each blocked segment. This indicates the preset minimum effective tension. Indicates the first The tension of each netting segment involved in the sag calculation. Indicates the first Wind load and on-site corrections for each section of the network enclosure. This indicates that the maximum value of the comparison objects within the parentheses is taken.

[0083] The wind load and field correction are calculated using a preset calibration function based on the wind speed component perpendicular to the direction of the netting section, the horizontal distance between the constraint nodes at both ends of the netting section, and the equivalent tension. The wind speed component perpendicular to the direction of the netting section is determined by the wind speed, wind direction, and azimuth of the netting section. Specifically, the azimuth of the netting section is first subtracted from the wind direction, then the sine of the difference is taken, and finally the wind speed is multiplied by the sine. The azimuth of the netting section is the angle of the projection direction of the line connecting the constraint nodes at both ends of the netting section onto the horizontal plane relative to the selected azimuth reference in the unified coordinate system horizontal plane. It is calculated by the arctangent of the coordinate difference of the projections of the constraint nodes at both ends of the netting section onto the horizontal plane in step 100. Before participating in the calculation of the wind speed component, the wind direction is uniformly converted to the same azimuth reference as the azimuth of the netting section, so that the wind direction and the azimuth of the netting section can be directly subtracted under the same azimuth reference.

[0084] The preset calibration function is an equivalent calculation function for wind-induced additional sag. During the calibration stage, it is obtained by fitting the correspondence between the measured sag and the wind speed component, horizontal distance, and equivalent tension. When calculating the value of the preset calibration function, the square of the wind speed component perpendicular to the direction of the netting section is first taken, then multiplied by the square of the horizontal distance of the netting section and the preset wind load correction coefficient, and then divided by the equivalent tension of the netting section to obtain the wind load and field correction terms. Since the preset wind load correction coefficient has the corresponding compensation dimension, the calculation results of the wind load and field correction terms are length quantities, which are consistent with the dimension of the sag caused by self-weight, and the two can be directly added. The wind load and field correction terms are engineering equivalent correction terms for wind-induced additional sag, used to equivalently compensate for the additional sag caused by lateral wind load on the basis of self-weight sag, rather than being obtained by strict mechanical inversion.

[0085] In an optional embodiment, for a net enclosure segment that does not meet the preset parabolic approximation conditions, a calibrated catenary model is used instead of the parabolic approximation model. The catenary model is an engineering approximation model of the net enclosure segment's shape, which approximates the shape of the net enclosure segment under its own weight as a catenary, but is not used as a strict mechanical inversion. First, the equivalent tension of the net enclosure segment is divided by the weight per unit length to obtain the catenary characteristic parameters. Then, half of the horizontal distance of the net enclosure segment is divided by the catenary characteristic parameters to obtain the input of the hyperbolic cosine function. The hyperbolic cosine value of this input is taken and subtracted by one, and then multiplied by the catenary characteristic parameters to obtain the result of the self-weight action. The estimated sag of the catenary under its own weight is calculated by adding the wind load and field correction term calculated as described above. When there is a lateral wind load, the actual catenary section is subjected to the combined distributed load of gravity load and wind load, and its shape is located in the inclined stress plane determined by the direction of the combined load. In this embodiment, the midpoint sag under the combined load is estimated by superimposing the catenary sag under its own weight with the wind load and field correction term. The wind load and field correction term are used to equivalently compensate for the effect of the wind load component in the inclined stress plane on the midpoint sag.

[0086] For example, when using the catenary model, the estimated sag can be expressed by the following formula: , ;in, Indicates the first The first blocked segment in the Catenary characteristic parameters at each sampling time, Indicates the first The first blocked segment in the The estimated sag at each sampling time point is obtained using the catenary model. Represents the hyperbolic cosine function. , , , and The meaning is the same as in the parabola approximation model; when the parabola approximation model is selected in step 300, the first... Estimated sagging of each blocked segment Pick When the catenary model is selected in step 300, the first Estimated sagging of each blocked segment Pick .

[0087] The reason for using the catenary model instead of the parabolic approximation model is that when the sag ratio of the netting section is not within the applicable range of the preset parabolic approximation, the flexible cable shape needs to retain the characteristic parameters of the catenary in the calculation so that the estimated sag corresponds to the geometric relationship of the flexible cable under this working condition.

[0088] Based on the location of the constraint nodes, the planned net enclosure boundary, and the influence of wind direction, the lateral offset risk characteristics of any net enclosure segment are calculated. The planned traction path direction is the longitudinal direction extending along the planned traction path projection within the plane where the preset target net enclosure area is located. First, the constraint nodes at both ends of the net enclosure segment are projected onto the plane where the preset target net enclosure area is located, and the longitudinal and lateral coordinates of the projection points at both ends are obtained. The average position of the longitudinal coordinates of the projection points at both ends is taken as the corresponding longitudinal position of the net enclosure segment, and the average position of the lateral coordinates of the projection points at both ends is taken as the estimated lateral position before visual correction. After step 400 completes visual correction, the estimated lateral position is taken as the estimated lateral position of the net enclosure segment in the corrected net enclosure state.

[0089] When obtaining the planned lateral position, first convert the control points of the planned net enclosure boundary in the construction plan to the plane where the preset target net enclosure area is located, and generate the planned boundary curve according to the planned net enclosure boundary sequence; the preset boundary discrete spacing is determined before the operation based on the spacing of the planned net enclosure boundary control points, the resolution of on-site coordinate measurement, and the accuracy requirements for calculating the area of ​​the target net enclosure area, and remains consistent in the same net enclosure operation; when the planned boundary curve contains arcs, splines, or polylines plus arc structures, convert them into ordered planned boundary points according to the preset boundary discrete spacing; based on the binding relationship between the planned net enclosure boundary and the net enclosure segment, select the planned boundary curve corresponding to the net enclosure segment, and then on the planned boundary curve The system searches online for adjacent planned boundary points that cover the longitudinal position of the network segment. For planned boundary segments between adjacent planned boundary points, the system interpolates the lateral coordinates of the two planned boundary points according to the proportion of the longitudinal position of the network segment within the planned boundary segment to obtain the planned lateral position of the network segment. When the longitudinal coordinates of two planned boundary points in the planned boundary segment are the same and cannot be interpolated by longitudinal coordinates, the system searches along the path of the planned boundary curve for the planned boundary point that is closest to the plane projection point corresponding to the longitudinal position of the network segment. The system uses the lateral coordinates of the planned boundary point as the planned lateral position and writes this acquisition method into the basic operation information.

[0090] The dimensionless geometric offset term is obtained by dividing the absolute value of the difference between the estimated lateral position and the planned lateral position by the preset maximum allowable lateral offset. Then, the wind-induced additional lateral displacement is calculated. Specifically, the square of the wind speed component perpendicular to the direction of the netting section is taken, multiplied by the horizontal distance of the netting section and the preset wind-induced offset weight, and then divided by the equivalent tension of the netting section. Since the preset wind-induced offset weight has a corresponding compensation dimension, the calculated result of the wind-induced additional lateral displacement is a length quantity, representing the equivalent lateral displacement caused by wind load. Finally, the dimensionless geometric offset term is obtained by dividing the wind-induced additional lateral displacement by the preset maximum allowable lateral offset. The wind-induced offset ratio is used to obtain the lateral offset risk characteristic of the enclosure segment by adding the geometric offset term to the wind-induced offset ratio. The lateral offset risk characteristic is dimensionless. The wind-induced additional lateral displacement is obtained by fitting the correspondence between the measured lateral offset and the wind speed component, equivalent tension, and horizontal distance during the calibration stage. The preset wind-induced offset weight is determined by this correspondence. The geometric offset term and the wind-induced offset ratio are used at the same time because the actual lateral risk of the enclosure segment comes from both the deviation of the constraint node from the planned enclosure boundary and the lateral wind load. The sum of the two forms the same dimensionless lateral offset risk characteristic.

[0091] After obtaining the estimated sag and estimated lateral position, the estimated sag and estimated lateral position of each netting segment are mapped to the plane of the preset target netting area to generate a real-time estimated coverage area. Specifically, the positions of the constraint nodes at both ends of each netting segment in the set of netting segments are used as the skeleton. For each netting segment, the boundary curve of the netting segment is a planar curve used for coverage area calculation, which is derived from the projection of the three-dimensional estimated line of the netting segment onto the plane of the preset target netting area. The endpoints of the boundary curve are the two projection points obtained after projecting the constraint nodes at both ends of the netting segment along the normal of the plane of the preset target netting area. The endpoints of the boundary curve are the start and end positions of the boundary curve of the netting segment, and the boundary curve of the netting segment passes through its two corresponding endpoints. The vertical offset of the midpoint is the displacement of the midpoint of the three-dimensional estimated line of the netting segment relative to the line connecting the constraint nodes at both ends in the vertical direction. Its value is taken as the estimated sag of the netting segment, and its direction is taken as the direction of gravity.

[0092] When generating the boundary curve of the sealing segment, firstly, curve sampling points are set between the three-dimensional constraint nodes corresponding to the endpoints of the two boundary curves according to the direction of the sealing segment and the preset curve sampling interval. The preset curve sampling interval is determined before the operation based on the horizontal distance of the sealing segment, the preset boundary discrete interval, and the accuracy requirements of the three-dimensional estimated linear shape expression, and remains consistent in the same sealing operation. For each curve sampling point, the path length from the starting constraint node along the direction of the sealing segment is recorded, and the path length is divided by the total path length along the direction of the sealing segment between the constraint nodes at both ends of the sealing segment to obtain the segment position ratio of the curve sampling point. Then, according to the parabolic approximation model or catenary model selected in step 300, the vertical displacement of each curve sampling point relative to the line connecting the constraint nodes at both ends is determined. The vertical displacement at the constraint nodes at both ends is zero, the vertical displacement at the midpoint is the vertical offset of the midpoint, and the vertical displacement of the remaining curve sampling points is obtained by the selected model according to the segment position ratio.

[0093] Next, the sampling points of each curve are laterally located according to the estimated lateral position. Lateral positioning refers to determining the lateral coordinates of each curve sampling point in the lateral direction perpendicular to the longitudinal direction based on the estimated lateral position of the network segment. The process is as follows: First, according to the segmented position ratio of the curve sampling point, linear interpolation is performed between the lateral coordinates of the starting boundary curve endpoint and the lateral coordinates of the ending boundary curve endpoint to obtain the reference lateral coordinates of the curve sampling point; then, the average of the lateral coordinates of the two boundary curve endpoints is subtracted from the estimated lateral position of the network segment to obtain the lateral correction amount of the midpoint of the network segment; finally, the result of multiplying the segmented position ratio of the curve sampling point by one and subtracting the segmented position ratio, and then multiplying by four, is obtained to obtain the curve sampling point. The normalized distribution factor is set to zero at the two constraint nodes and one at the midpoint. Finally, the midpoint lateral correction is multiplied by the normalized distribution factor of the curve sampling point to obtain the lateral correction amount of the curve sampling point. This lateral correction amount is then added to the reference lateral coordinate of the curve sampling point to obtain the lateral coordinate of the curve sampling point. According to this lateral positioning method, the lateral coordinates of each curve sampling point change continuously between the lateral coordinates of the endpoints of the boundary curves at both ends, and the lateral coordinate of the curve sampling point at the midpoint is equal to the estimated lateral position of the net segment. Thus, the three-dimensional estimated line shape formed by the curve sampling points is consistent with the lateral coordinates of the constraint nodes at both ends and consistent with the estimated lateral position of the net segment at the midpoint.

[0094] For example, the segmented position ratio, vertical displacement, and lateral coordinate of the curve sampling points can be represented by the following formula: , , , , When using a parabolic approximation model, the vertical displacement of the curve sampling points can be expressed as: When using the catenary model, the vertical displacement of the curve sampling points can be expressed as: ; in, Indicates the curve sampling point number, Indicates the first The curve sampling point along the first curve sampling point The path length from the starting constraint node in each blocked segment direction. Indicates the first The total path length along the direction of the net segment between the constraint nodes at both ends of each net segment. Indicates the first The proportion of the segmented positions of each curve sampling point Indicates the first Normalized distribution factor for each curve sampling point Indicates the first The lateral coordinates of the endpoints of the boundary curve at the beginning of each enclosure segment. Indicates the first The horizontal coordinates of the endpoints of the boundary curve of each blocked segment. Indicates the first The reference lateral coordinates of each curve sampling point Indicates the first The estimated lateral position of each blocked segment Indicates the first Lateral correction amount at the midpoint of each blocked segment. Indicates the first The horizontal coordinates of each curve sampling point Indicates the first The vertical displacement of each curve sampling point relative to the line connecting the two constraint nodes. Indicates the first Estimated sagging of each blocked section Indicates the first The horizontal distance between the nodes at both ends of each network segment is constrained. Indicates the first Characteristic parameters of the catenary of each sealed section.

[0095] After determining the longitudinal position, lateral coordinates, and vertical displacement of the line connecting the constraint nodes at both ends of each curve sampling point along the direction of the enclosure segment in the above manner, the three-dimensional estimated shape of the enclosure segment is obtained. Then, the three-dimensional estimated shape is projected along the normal direction of the plane where the preset target enclosure area is located to obtain the boundary curve of the enclosure segment used for coverage area calculation. When using the parabolic approximation model, the three-dimensional estimated shape passes through the three-dimensional constraint nodes corresponding to the endpoints of the two boundary curves, and the midpoint position is constrained by the vertical offset of the midpoint. When using the catenary model, the three-dimensional estimated shape passes through the three-dimensional constraint nodes corresponding to the endpoints of the two boundary curves, and the midpoint position is determined by the estimated sag of the midpoint obtained from the catenary model. The boundary curves of each enclosure segment are superimposed, and the gap between the boundaries of adjacent enclosure segments is processed according to the preset boundary completion gap threshold. The preset boundary completion gap threshold is determined based on the effective overlap width of the enclosure mesh and the protection boundary requirements of the crossed object.

[0096] Adjacent network segment endpoints refer to the boundary curve endpoints of two network segments that are connected in the virtual state chain, each located on one side of the connection position. Therefore, adjacent network segment endpoints belong to the boundary curve endpoints of the corresponding network segments and are not generated as independent endpoints. When two adjacent network segments share the same constraint node and the constraint node is neither unhooked nor released, the endpoints of the two adjacent network segments correspond to the projection points of the same constraint node. When on-site positioning or visual correction causes a coordinate difference between the two projection points, this coordinate difference forms the gap between the endpoints of adjacent network segments. When the gap between the endpoints of adjacent network segments is less than or equal to the preset boundary completion gap... When the gap threshold is reached, the endpoints of adjacent network segments are connected in sequence according to the connecting edges in the network virtual state chain to complete the boundary. The area formed by the boundary completion is marked as the estimated completion area. When the gap between the endpoints of adjacent network segments is greater than the preset boundary completion gap threshold, or when there is no connection between adjacent network segments in the network virtual state chain due to decoupling, release, or disconnection of connecting edges, no boundary completion is performed, and the gap is marked as an uncovered gap. The boundary curves of each network segment and the estimated completion area after the above processing are connected to form the real-time estimated coverage area, so that the real-time estimated coverage area constitutes an area on the same plane as the target network area.

[0097] The real-time estimated coverage area is generated using curve sampling points because the net boundary is no longer equivalent to the line connecting the two ends after sagging and lateral offset; projection of the curve sampling points can preserve the curve shape of the net segment boundary, so that the net coverage integrity index is calculated based on the current estimated boundary.

[0098] The netting coverage integrity index is defined as the ratio of the effective coverage area of ​​the estimated coverage area within the target netting area to the area of ​​the target netting area. The area of ​​the target netting area is obtained before the operation. First, the protection boundary of the object to be crossed, the control points of the planned netting boundary, and the on-site reference points determined by the construction plan are transformed to a unified coordinate system. Then, the normal projection along the plane where the preset target netting area is located is used as the plane boundary points, and the target netting area is enclosed according to the boundary sequence given in the construction plan. When the boundary contains arcs, splines, or polylines plus arcs, the curve segments are converted into ordered boundary points according to the preset boundary discrete spacing. The area of ​​the target netting area is the area of ​​the plane area enclosed by the above ordered boundary points, and is saved in the basic operation information. If the boundary point coordinates need to be corrected during on-site verification, the target netting area is re-enclosed with the corrected plane boundary points. The target netting area area is updated while retaining the source of correction. When calculating the netting coverage integrity index, the overlapping area between the real-time estimated coverage area and the target netting area is first taken. Then, the area of ​​the void area located inside the real-time estimated coverage area and not surrounded by any netting segment boundary curve, as well as the area corresponding to the uncovered gap, are subtracted from the overlapping area to obtain the effective coverage area. Finally, the effective coverage area is divided by the target netting area area to obtain the netting coverage integrity index. The estimated completion area formed by boundary completion is separately identified when included in the effective coverage area to distinguish it from the area directly covered by the netting segment boundary curve. When the real-time estimated coverage area exceeds the boundary of the target netting area, the part exceeding the target netting area is marked as a dangerous expansion area and separately indicated. Dangerous expansion areas are not included in the netting coverage integrity index.

[0099] In an optional embodiment, the areas of the real-time estimated coverage area and the target netting area are obtained through planar geometric Boolean operations. The boundary curves of the netting segment, the estimated completion area boundary, the target netting area boundary, and the dangerous expansion area boundary involved in the operation are all first converted into ordered planar boundary points and sorted in the same direction. If the boundary of the real-time estimated coverage area intersects itself, the boundary is divided into multiple non-intersecting sub-regions using the intersection point as the dividing point, and then the overlap relationship between each sub-region and the target netting area is calculated. An internal void is a planar area located within the target netting area, enclosed by the outer boundary of the real-time estimated coverage area, and not covered by the boundary curve of the netting segment or the estimated completion area. This internal void is not included in the effective coverage area.

[0100] By deducting internal voids, uncovered gaps, and excluding dangerous expansion areas, the netting coverage integrity index can reflect only the portion of the netting area effectively covered by the netting.

[0101] Step 400: Visually correct the inversion state composed of the estimated droop and estimated lateral position based on the credibility of each data source to obtain the corrected netting state. Calculate the mean of the credibility of each data source to obtain the multi-source data average credibility. Calculate the comprehensive risk score based on the corrected netting state, the multi-source data average credibility, the traction constraint anomaly marker, and the netting coverage integrity index, as detailed below. Figure 2 As shown.

[0102] The netting state is represented as a state vector, which alternates between the estimated droop and estimated lateral position of each netting segment in the order of the netting segments. The dimension of the state vector is determined by the number of netting segments and is synchronously reconstructed after the virtual state chain of the netting is updated. The current state vector obtained from the multi-source data inversion in steps 200 to 300 is denoted as the inversion state before visual correction. The covariance matrix of the inversion state represents the uncertainty of the inversion state, hereinafter referred to as the state covariance matrix. During the initial sampling, the state covariance matrix is ​​established based on the estimation error of the netting segment droop, the estimation error of the lateral position, and the calibration error of each data source. During continuous sampling, time updates are performed based on the incremental travel distance of the integrated machine, changes in equivalent tension, changes in ambient wind speed, and a preset process noise matrix. The preset process noise matrix is ​​obtained by statistically analyzing the state change errors of adjacent sampling times in historical netting operations. When historical records are unavailable, the state change range allowed by the construction plan, sensor measurement resolution, and preset sampling period are used. State vectors and state covariance matrices are used because the sag and lateral position of different netting sections need to be updated within the same correction framework. The state covariance matrix provides uncertainty weights for the fusion of visual observations and the inverted state before visual correction.

[0103] In an optional embodiment, when the virtual state chain of the net enclosure is reconstructed due to decoupling, release, or the addition of constraint nodes, the estimated sag, estimated lateral position, and corresponding elements of the state covariance matrix at the previous effective time are retained for net enclosure segments whose numbers remain unchanged; for newly generated net enclosure segments, the estimated sag and estimated lateral position are initialized based on the positions of the constraint nodes at both ends, equivalent tension, equivalent mass per unit length, and environmental state information, and the corresponding elements of the state covariance matrix are initialized based on the calibration error; for net enclosure segments that have been disconnected, their state is removed from the current state vector and retained in the operation closed-loop record.

[0104] In an optional embodiment, when the visual confidence level obtained in step 100 is greater than or equal to a preset visual confidence level threshold, the visual image is coordinate registered with the conductor position obtained from video recognition, and the registered local rope edge, mesh edge, or trolley position constitutes the current visual observation value to correct the inversion state; the video module completes the calibration of camera intrinsic and extrinsic parameters and installation attitude before operation, and uses the conductor, trolley, and anchor point with known three-dimensional coordinates in a unified coordinate system as geometric reference points to provide depth scale constraints; the camera intrinsic parameters are determined by a calibration plate, a reference target with known geometric dimensions, or the video module's factory calibration file, and include focal length, principal point, and distortion parameters; the camera extrinsic parameter installation attitude is solved by registering the image coordinates of the geometric reference point and the three-dimensional coordinates of the unified coordinate system, and includes the rotation and translation relationships from the camera coordinate system to the unified coordinate system; the positions of the rope edge, mesh edge, or trolley identified in the image are converted from the calibration parameters and the geometric reference points to the current visual observation value. Visual observations in a unified coordinate system participate in correction only along the inversion plane in the direction determined by geometric reference point constraints. When camera intrinsic and extrinsic parameters, installation attitude calibration parameters, or geometric reference point constraints are unavailable, video data does not participate in the above three-dimensional state correction and is only used as a two-dimensional image offset prompt for operator reference. First, the visual observation noise matrix is ​​adjusted according to visual credibility. During adjustment, the maximum value between visual credibility and the preset zero-prevention constant is taken as the divisor, and the preset basic observation noise matrix is ​​divided by this divisor to obtain the adjusted visual observation noise matrix. The preset zero-prevention constant is determined based on the numerical resolution of the visual credibility calculation result and is used to prevent division by zero. The preset basic observation noise matrix is ​​obtained by statistically analyzing the recognition errors of the net rope edge, net sheet edge, and pulley position during the calibration phase by the visual recognition module. In this way, when visual credibility decreases, the adjusted visual observation noise matrix increases, and the influence of visual observation on the correction result decreases accordingly.

[0105] In an optional embodiment, when two or more video modules are configured and different video modules have completed the calibration of camera intrinsic and extrinsic parameters and installation attitude, the net rope edge, net sheet edge or trolley position identified by each video module are converted into candidate visual observations in a unified coordinate system. For the same net segment and the same observation type, the fusion result of the candidate visual observations is determined according to the visual credibility and visual observation noise matrix of each video module, and the fused visual observations are written into the current visual observations. When any candidate visual observation cannot be bound to the net virtual state chain, the candidate visual observation does not participate in the fusion.

[0106] The visual confidence level is used to adjust the visual observation noise matrix because video quality directly affects the usability of visual observations. Mapping the visual confidence level to the visual observation noise allows for image correction when the visual quality meets a threshold, and preserves non-visual inversion results when the visual quality does not meet the threshold.

[0107] The preset visual credibility threshold is determined based on the credibility distribution of boundary recognition results in the video recognition calibration samples that can be used for network closure status correction. When the visual credibility is less than the preset visual credibility threshold, the video data is only used as an auxiliary prompt and is not used as the main basis for judging the integrity of network closure coverage.

[0108] The correction gain is calculated based on the adjusted visual observation noise matrix. The calculation involves multiplying the preset observation matrix, the covariance matrix of the inverted state before visual correction, and the transpose of the preset observation matrix sequentially, then adding the adjusted visual observation noise matrix to obtain an intermediate matrix. This intermediate matrix is ​​then inverted, and the covariance matrix of the inverted state before visual correction, the transpose of the preset observation matrix, and the inverse of the intermediate matrix are multiplied sequentially to obtain the correction gain at the current time. When the intermediate matrix cannot be directly inverted, a regularization term of the preset correction matrix is ​​added diagonally to the intermediate matrix before inverting it. The regularization term of the preset correction matrix is ​​determined based on the numerical resolution of the visual observation noise matrix. The preset observation matrix is ​​used to map the network closure state vector to the visual observation space. The system establishes a correspondence between the estimated droop and estimated lateral position of each net segment in the state vector and the visual observation. When establishing the preset observation matrix, the net segment, observation type, and coordinate direction corresponding to each visual observation value are first determined, and then the correspondence is written into the preset observation matrix so that the same visual observation value only affects its corresponding estimated droop, estimated lateral position, or a combination of the two. When a visual observation value cannot be bound to a net segment, the visual observation value is not included in the correction gain calculation. The role of the correction gain is to allocate the weights of the inverted state before visual correction and the visual observation value according to the state covariance matrix and the visual observation noise matrix, so that the corrected net state at the current time retains both the physical inversion result and the visual positioning result.

[0109] The network closure state is corrected according to the correction gain. First, the preset observation matrix is ​​multiplied by the inverted state before visual correction to obtain the predicted visual observation value. Then, the predicted visual observation value is subtracted from the current visual observation value to obtain the visual observation residual. The correction gain is multiplied by the visual observation residual and added to the inverted state before visual correction to obtain the corrected network closure state at the current time. After the correction is completed, the state covariance matrix is ​​updated synchronously. During the update, the correction gain is multiplied by the preset observation matrix, and then the product is subtracted from the identity matrix. Finally, the resulting matrix is ​​multiplied by the covariance matrix of the inverted state before visual correction to obtain the corrected state covariance matrix.

[0110] Extract the corrected estimated sag and estimated lateral position of each netting segment from the corrected netting state, and update the lateral offset risk characteristics, real-time estimated coverage area, and netting coverage integrity index according to step 300, as input for the risk score in this step.

[0111] In an optional embodiment, when the visual confidence level is less than a preset visual confidence level threshold, the above-mentioned visual correction is not performed. Instead, the inversion state before visual correction is directly taken as the current network closure state, and the estimated droop, lateral offset risk characteristics, and network closure coverage integrity index obtained from the inversion of the all-in-one machine, trolley, and environmental data are maintained. At the same time, since the visual confidence level is lower than the preset visual confidence level threshold at this time, its participation in the multi-source data average confidence level calculation in this step will reduce the final confidence level output value.

[0112] After determining the current network closure status, based on the estimated sagging amount, preset maximum allowable sagging amount, lateral offset risk characteristics, and network closure coverage integrity index obtained in step 300 and determined in this step, calculate the sagging risk score and lateral offset risk score for each network closure segment, and calculate the network closure coverage integrity risk score and comprehensive risk score. The risk quantities are converted into dimensionless scores because the original dimensions of sagging amount, lateral offset, coverage integrity, and confidence are different; after being unified into risk scores, they can be used in the same comprehensive risk scoring process.

[0113] For any blocked network segment, when calculating the sagging risk score, first subtract the preset maximum allowable sagging from the estimated sagging amount of the blocked network segment, and then divide by the preset maximum allowable sagging to obtain the sagging excess ratio. If the sagging excess ratio is less than zero, the sagging risk score for the blocked network segment is zero. If the sagging excess ratio is greater than or equal to zero, the sagging risk score for the blocked network segment is the sagging excess ratio. Therefore, when the estimated sagging amount does not exceed the preset maximum allowable sagging amount, the sagging risk score for the blocked network segment is zero.

[0114] For any network segment, when calculating the lateral offset risk score, first subtract one from the lateral offset risk characteristic of the network segment to obtain the lateral offset exceeding the limit. If the lateral offset exceeding the limit is less than zero, the lateral offset risk score of the network segment is zero. If the lateral offset exceeding the limit is greater than or equal to zero, the lateral offset risk score of the network segment is the lateral offset exceeding the limit. Therefore, when the lateral offset risk characteristic does not exceed one, the lateral offset risk score of the network segment is zero.

[0115] When calculating the network coverage integrity risk score, the network coverage integrity index is subtracted from one to obtain the network coverage integrity risk score.

[0116] The traction constraint risk term is calculated based on the traction constraint anomaly markers obtained in step 200. When there are no traction constraint anomaly markers at the current sampling time, the traction constraint risk term is zero. When there are any traction constraint anomaly markers at the current sampling time, the traction constraint risk term is one. The traction constraint risk term is used to represent the impact of abnormal force transmission at the traction end attachment, anchoring constraint, or netting section on the netting state inversion, and participates in the comprehensive risk score and key item constraints.

[0117] Further calculate the comprehensive risk score; first, obtain the risk score of network coverage integrity, the maximum value of the sagging risk score of all network segments, the maximum value of the lateral offset risk score of all network segments, the result after subtracting the average credibility of multi-source data, and the traction constraint risk item. The sagging risk score and the lateral offset risk score are taken as the maximum value for each network segment to reflect the extreme risk of a single network segment without being masked by the average of the other network segments; then, multiply the above five items by their corresponding preset risk weights and sum them to obtain the comprehensive risk weighted value; the sum of the five preset risk weights is one; the preset risk weights correspond to the network coverage integrity risk score, sagging risk score, lateral offset risk score, credibility risk item, and traction constraint risk item, respectively. Their value logic is determined based on the protection priority of the crossed object, historical operation risk records, and on-site safety control strategies, and is verified through historical qualified operation records; each preset risk weight is greater than or equal to zero; when the protection priority of the crossed object, on-site safety control strategies, or data source configuration changes, the preset risk weights are re-determined before the operation, and the determination basis is written into the operation basic information.

[0118] For example, the overall risk weighted value can be expressed by the following formula: ,and ; Indicates the first The comprehensive risk-weighted value at each sampling time. This indicates the risk score for the integrity of network coverage during the shutdown. This represents the maximum value among all the risk scores for sagging in the blocked network segment. Indicates the first Each blocked segment has a risk score for sagging, among which... Indicates the network segment number, Indicates the first Lateral offset risk assessment for each blocked section. This represents the maximum value in the lateral offset risk score for all blocked sections. Indicates credibility risk item, This indicates the average confidence level of multi-source data. Indicates the risk item of traction constraint. , , , and These represent the preset risk weights corresponding to the network coverage integrity risk score, sagging risk score, lateral offset risk score, credibility risk item, and traction constraint risk item, respectively. Indicates the number of all blocked network segments. Take the maximum value from the corresponding range.

[0119] In an optional embodiment, the weighted value of the comprehensive risk is directly truncated; when the weighted value of the comprehensive risk is less than zero, the comprehensive risk score is zero; when the weighted value of the comprehensive risk is greater than one, the comprehensive risk score is one; and in other cases, the comprehensive risk score is the weighted value of the comprehensive risk.

[0120] In another optional embodiment, to avoid key individual risks being masked by weighted summation and averaging, preset individual risk thresholds are set for the network coverage integrity risk score, the maximum value among all network segment sagging risk scores, the maximum value among all network segment lateral offset risk scores, and the credibility risk item, respectively, and anomaly marker constraints are set for the traction constraint risk item; when the number of occurrences of the same traction constraint anomaly marker at consecutive sampling times reaches the preset anomaly confirmation count, the initial comprehensive risk score is the maximum value between the comprehensive risk weighted value and the preset anomaly state risk threshold; when there is no aforementioned continuous occurrence, but any of the above four risk items with set individual risk thresholds exceeds its corresponding preset individual risk threshold, or the traction constraint risk item is equal to one, the initial comprehensive risk score is the maximum value between the comprehensive risk weighted value and the preset attention state risk threshold, so that the key individual risk or traction constraint anomaly marker at least triggers network closure state attention; when the above four single risk thresholds are set When none of the risk items exceed their respective preset individual risk thresholds and the traction constraint risk item is equal to zero, the initial comprehensive risk score is the weighted value of the comprehensive risk. Each preset individual risk threshold is determined based on the protection requirements of the crossed object and historical network closure anomaly records, and is within the effective value range of its corresponding risk item. When a certain risk item has a clear trigger action record in the historical network closure anomaly records, the value of the risk item corresponding to the trigger action record is used as the verification basis for the preset individual risk threshold. The preset anomaly confirmation number is determined based on the preset sampling period, the integrated machine control response time, and the on-site alarm interlock response time, and is an integer greater than or equal to two. Finally, the initial comprehensive risk score is truncated: if the initial comprehensive risk score is less than zero, the comprehensive risk score is zero; if the initial comprehensive risk score is greater than one, the comprehensive risk score is one; otherwise, the comprehensive risk score is the initial comprehensive risk score itself. The comprehensive risk score is an empirical score used for graded prompts and is not used as a statistical probability.

[0121] For example, the initial comprehensive risk score and the comprehensive risk score can be represented by the following formula: , ;in, Indicates the first The initial comprehensive risk score at each sampling time. Indicates the first The comprehensive risk score at each sampling time point, This indicates a preset threshold for abnormal state risks. This indicates a preset risk threshold for a state of concern. Indicates the same traction constraint anomaly marker up to the [number]. The number of consecutive occurrences at each sampling time. Indicates the preset number of exception confirmations. Indicates a key individual risk trigger flag. This indicates that the maximum value among the comparison objects within the parentheses is taken. This indicates taking the minimum value of the comparison objects within the parentheses; when , , and If any of the items exceeds its corresponding preset single-item risk threshold, or hour, ,otherwise .

[0122] The average credibility of multi-source data is obtained by averaging the credibility of each valid data source obtained in step 100. First, the credibility of all non-visual data sources participating in the judgment after validity screening at the current time is added together. When the video module has been configured and formed visual credibility, the visual credibility is added to obtain the total credibility. The total credibility is divided by the number of data sources participating in the judgment to obtain the average credibility of multi-source data. When the video module is not configured or visual credibility has not been formed, the visual data source is not included in the number of data sources participating in the judgment. When the video module has been configured but the visual credibility is less than the preset visual credibility threshold, the visual credibility is still included in the average credibility of multi-source data to reflect the monitoring credibility. When there is no valid non-visual data source at the current time and the video module has not formed visual credibility, the average credibility of multi-source data is marked as invalid, and the monitoring credibility is directly output as insufficient.

[0123] To prevent the failure of key data sources such as vision or all-in-one machine traction from being masked by the average credibility of multi-source data, the minimum credibility of key data sources is obtained separately while calculating the average credibility of multi-source data. Key data sources include vision data sources and all-in-one machine traction data sources that provide traction force to the all-in-one machine and participate in the equivalent tension calculation. The minimum credibility of each data source currently participating in the judgment among the above key data sources is taken as the minimum credibility of the key data source, which is used in step 500 to determine whether the monitoring credibility is sufficient. When there is no data source currently participating in the judgment among the key data sources, the minimum credibility of the key data source is set to zero, and insufficient monitoring credibility is output in step 500. When a configured key data source does not participate in the judgment, the key data source is marked as missing on the display.

[0124] Step 500: Output the network closure status level based on the comprehensive risk score and the average credibility of multi-source data.

[0125] The network closure status level is output based on the comprehensive risk score, average reliability of multi-source data, and traction constraint anomaly markers. When the average reliability of multi-source data is less than the preset insufficient monitoring reliability threshold, or the minimum reliability of the key data source is less than the preset key data source reliability threshold, the output network closure status level is "Insufficient Monitoring Reliability," prompting operators to perform manual verification or switch observation angles. If no "Insufficient Monitoring Reliability" is output, but the comprehensive risk score is greater than or equal to the preset abnormal status risk threshold, the output network closure status level is "Abnormal Network Closure," prompting operators to suspend traction, verify the trolley anchoring status, adjust the integrated machine synchronization status, check the network mesh posture, or perform traction constraint anomaly handling. When the comprehensive risk score is greater than or equal to... When the preset risk threshold for a state of concern is less than the preset risk threshold for an abnormal state, or when the average credibility of multi-source data is less than the preset credibility threshold for a normal state, the output network closure status level is "Network Closure Status: Concerned". When the comprehensive risk score is less than the preset risk threshold for a state of concern, and the average credibility of multi-source data is greater than or equal to the preset credibility threshold for a normal state, the output network closure status level is "Network Closure Status: Normal". When insufficient monitoring credibility and the network closure status being normal, concerned, or abnormal simultaneously meet the output conditions, insufficient monitoring credibility is output first, and the corresponding network closure status level is retained as a reference for manual review. When the comprehensive risk score reaches the preset risk threshold for an abnormal state, action suggestions for an abnormal network closure status are also output.

[0126] The reason for prioritizing output based on insufficient monitoring credibility is that the basis for risk scoring when the data source credibility is insufficient needs to be manually reviewed; the network block status level is retained as a reference for manual review, and the current automatic judgment result can be provided while prompting for review.

[0127] Specifically, the preset thresholds for risk in a state of concern, abnormal state, normal state reliability, insufficient monitoring reliability, and key data source reliability are all determined before operation. The values ​​of the preset thresholds for risk in a state of concern and abnormal state are determined based on historical network closure anomaly records, protection requirements for crossed objects, and on-site alarm classification strategies. The values ​​of the preset thresholds for reliability in a normal state and insufficient monitoring reliability are determined based on the number of valid data sources, data source calibration errors, and manual review triggering conditions. The values ​​of the preset thresholds for reliability of key data sources are determined based on the role of visual data sources and integrated machine-guided data sources in judging network closure status. The preset threshold for risk in a state of concern is greater than or equal to zero and less than the preset threshold for risk in an abnormal state. The preset threshold for risk in an abnormal state is less than or equal to one. The preset threshold for insufficient monitoring reliability is greater than or equal to zero and less than or equal to the preset threshold for reliability in a normal state. The preset threshold for reliability in a normal state is less than or equal to one. The preset threshold for reliability of key data sources is greater than or equal to zero and less than or equal to one.

[0128] The human-computer interaction and alarm unit displays the target blocked area, the real-time estimated coverage area, the location of obstacles or objects being crossed, the all-in-one machine nodes marked by step 200 that have the risk of slippage or insufficient anchoring, and suspected stuck sections or suspected abnormal force sections, suspected sagging sections and suspected lateral offset sections obtained by step 400, as well as data sources with a confidence level lower than the preset normal state confidence level threshold on the display end, so that operators can locate the source of the anomaly; the suspected sagging section is the blocked section with a sagging risk score greater than zero, and the suspected lateral offset section is the blocked section with a lateral offset risk score greater than zero.

[0129] When there is a risk of slippage or insufficient anchoring at a node of the integrated machine, the human-machine interaction and alarm unit outputs the corresponding integrated machine number, its path location, and the bound sealing section, and prompts to suspend the traction output of the integrated machine, maintain or enter self-locking braking, check wheel pressure, the contact status of the drive wheel with the wire or Dyneema rope, clamping mechanism, anchoring status, and data on the source of the traction force of the integrated machine; before the risk of slippage or insufficient anchoring is manually confirmed to be eliminated, the integrated machine will not be used as a normal traction point to generate a suggestion to continue traction.

[0130] When a suspected stuck section is found, the human-machine interaction and alarm unit outputs the corresponding netting section number, the constraint nodes at both ends, and the associated integrated machine, and prompts to suspend the traction of the integrated machine associated with the netting section, maintain the self-locking of the constraint nodes at both ends of the netting section, and verify the contact status between the associated pulley, connection point, net rope, net boundary, and the object being crossed; after manual confirmation that there is no entanglement, hanging, foreign object obstruction, or abnormal release of the hook, the suspected stuck section mark is removed and a traction recovery suggestion is generated.

[0131] When a suspected abnormal stress section is found, the human-machine interaction and alarm unit outputs the corresponding netting section number, equivalent tension, preset minimum effective tension, and associated traction point. It also prompts users to stop increasing the traction force of the associated integrated machine, verify the traction path, preset force transmission node friction reduction coefficient, anchor point, traction point, and the equivalent tension transmission relationship of the netting section. Furthermore, it suggests eliminating the equivalent tension anomaly by unloading, re-tensioning, adjusting the force transmission node, or adjusting the anchor point. Before the suspected abnormal stress section is manually confirmed to be resolved, the equivalent tension of the netting section is only used for judging the stress trend, and the traction constraint risk item continues to be included in the comprehensive risk score.

[0132] To further establish a closed-loop record of operations, the status of the integrated machine, the status of the trolley, the environmental status, the visual credibility, the traction constraint anomaly markers, the net coverage integrity index, the sagging risk score, the lateral offset risk score, the comprehensive risk score, the handling suggestions, and the manual confirmation results at each sampling moment are stored as a net status record; the environmental status includes wind speed, wind direction, temperature, humidity, and rainfall or fog indicators; the traction constraint anomaly markers include integrated machine nodes with slippage or insufficient anchoring risk, suspected jamming sections, and suspected abnormal force sections.

[0133] The above-mentioned network closure status records are stored because offline correction, model training, and accident tracing all require the corresponding saving of inputs, outputs, handling suggestions, and manual confirmation results at the same sampling time.

[0134] After a net closure operation is completed, the preset adjustable parameters are corrected offline based on the manual confirmation results and the net closure status record. For preset adjustable parameters with quantifiable deviations, a limited-amplitude update method is used. The preset learning coefficient is multiplied by the quantification deviation and then added to the preset adjustable parameters before the update to obtain the updated preset adjustable parameters. The preset adjustable parameters are general preset adjustable parameters in the equivalent tension estimation model, wind load correction function, or lateral offset correction function, specifically including the preset motor torque constant, preset wind load correction coefficient, and preset wind-induced offset weight. The quantification deviation is the deviation between the system estimation result and the manual confirmation result for the corresponding preset adjustable parameters. For the preset motor torque constant, the quantification deviation is converted from the traction force deviation of the integrated machine obtained by manual confirmation or traction load sensor calibration. For the preset wind load correction coefficient, the quantification deviation is converted from the deviation between the manually confirmed sag and the system estimated sag. For the preset wind-induced offset weight, the quantification deviation is converted from the deviation between the manually confirmed lateral offset and the system estimated lateral offset. The preset learning coefficient is a value greater than... The preset update parameters, which are in the range of zero and less than one, are determined based on the stability of the manually confirmed deviation during historical offline correction. Since the preset learning coefficient is in the range of greater than zero and less than one, the magnitude of a single update does not exceed the absolute value of the quantization deviation, thus avoiding excessive adjustment in a single instance. The corresponding preset adjustable parameters are corrected in the same direction only when the direction of the manually confirmed deviation is consistent across multiple network closure operations. The trend of the quantization deviation is monitored between adjacent corrections. When the quantization deviation no longer decreases with correction, or when the updated preset adjustable parameter exceeds the preset allowable range of the parameter, the update of the parameter is stopped, so that the preset adjustable parameter remains within the preset allowable range and remains stable. The preset allowable range is determined based on the equipment factory data, the available range of material parameters, historical qualified operation records, and on-site safety control strategies, and is saved before operation. The preset learning coefficient being less than one is not used as a basis for the inevitable convergence of parameter updates. During offline correction, the preset adjustable parameters under the corresponding temperature and humidity conditions are corrected according to the temperature, humidity, and rainfall or fog indicators in the network closure status record.

[0135] The limited-amplitude update method is adopted because the result of a single manual confirmation may be affected by on-site handling, observation angle or temporary working conditions. By limiting the preset learning coefficient to a range greater than zero and less than one, and combining it with the quantitative deviation direction of multiple net sealing operations, the preset adjustable parameters can be gradually corrected according to the operation records.

[0136] When the results of manual verification cannot form a quantitative bias, the record of this operation will only be used as a data sample for subsequent preset parameter calibration or model training, and no real-time parameter updates will be performed.

[0137] Through steps 100 to 500 above, without relying on continuous sensors for the netting ropes or netting sheets, the integrated machine, pulley, environmental and local visual data are used to calculate and output in real time the integrity index of the power transmission line netting coverage, sagging risk score, lateral offset risk score, comprehensive risk score and average reliability of multi-source data, so as to realize real-time monitoring of netting coverage integrity, sagging risk, lateral offset risk and monitoring reliability.

[0138] Example 2: See Figure 3 As shown, a real-time monitoring system for the closure status of transmission lines is provided, used to execute the aforementioned real-time monitoring method for the closure status of transmission lines. The system includes: The multi-source data module 101 is used to acquire multi-source data according to a preset sampling period from the start of the self-sealing operation, establish a virtual state chain and sealing segment based on the multi-source data, and calculate the credibility of each data source.

[0139] The anomaly marking module 102 is used to estimate the traction force of the integrated machine based on multi-source data, estimate the equivalent tension of each sealing segment based on the traction force of the integrated machine and the virtual state chain of the sealing network, and form a traction constraint anomaly mark.

[0140] The coverage evaluation module 103 is used to calculate the estimated sag and estimated lateral position of each netting segment based on the netting segment, multi-source data and equivalent tension, and to calculate the netting coverage integrity index based on the estimated sag and estimated lateral position.

[0141] The risk scoring module 104 is used to visually correct the inversion state composed of the estimated droop and the estimated lateral position according to the credibility of each data source, to obtain the corrected netting state, calculate the mean of the credibility of each data source to obtain the average credibility of the multi-source data, and calculate the comprehensive risk score based on the corrected netting state, the average credibility of the multi-source data, the traction constraint anomaly marker and the netting coverage integrity index.

[0142] The network closure level module 105 is used to output the network closure status level based on the comprehensive risk score and the average credibility of multi-source data.

[0143] The embodiments of this application have been described above, but these embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments based on the guidance of these embodiments, and all of them are within the protection scope of these embodiments.

Claims

1. A method for real-time monitoring of the closure status of transmission lines, characterized in that, Includes the following steps: From the start of the network closure operation, multi-source data is acquired according to the preset sampling period. Based on the multi-source data, a virtual state chain and network closure segment are established, and the credibility of each data source is calculated. The traction force of the integrated machine is estimated based on multi-source data. The equivalent tension of each sealing segment is estimated based on the traction force of the integrated machine and the virtual state chain of the sealing network, and anomaly markers for traction constraints are formed. Based on the netting segments, multi-source data, and equivalent tension, the estimated sagging and estimated lateral position of each netting segment are calculated, and the netting coverage integrity index is calculated based on the estimated sagging and estimated lateral position. Based on the credibility of each data source, the inversion state composed of the estimated droop and the estimated lateral position is visually corrected to obtain the corrected netting state. The mean value of the credibility of each data source is calculated to obtain the average credibility of the multi-source data. Based on the corrected netting state, the average credibility of the multi-source data, the traction constraint anomaly marker, and the netting coverage integrity index, a comprehensive risk score is calculated. The network closure status level is output based on the comprehensive risk score and the average credibility of multi-source data.

2. The method for real-time monitoring of the network closure status of a transmission line according to claim 1, characterized in that, Methods for establishing virtual state chains and blocking segments for network blocking include: Based on the numbering and binding relationships of the integrated machine, pulley, traction point, anchor point and fixed connection point in the multi-source data, a virtual state chain for net enclosure is established with the integrated machine, pulley, traction point, anchor point and fixed connection point as nodes and the boundaries of the net rope, net sheet or traction boundary as connecting edges. The locked all-in-one machine, the anchored trolley, and the fixed connection point are recorded as constraint nodes, and the corresponding connection edges are updated when unhooking or release occurs; Network segments are divided based on constraint nodes.

3. The method for real-time monitoring of the network closure status of a transmission line according to claim 2, characterized in that, Methods for calculating the credibility of various data sources include: Multi-source data is filtered for validity based on communication status, communication delay, and power consumption. For the multi-source data that has been filtered for validity, the credibility of each data source is calculated, including the credibility of non-visual data sources and visual credibility. For non-visual data sources, a predicted value is generated based on the closed virtual state chain. The credibility of the non-visual data source is calculated based on the normalized squared residual between the current measurement value and the predicted value, as well as the data time delay of the non-visual data source from the current sampling time. For video data, visual credibility is calculated based on image clarity, occlusion ratio, image shake intensity, and lighting anomaly degree obtained from the video recognition module.

4. The method for real-time monitoring of the network closure status of a transmission line according to claim 1, characterized in that, Methods for estimating the equivalent tension of each netting segment include: Estimate the traction force of the integrated machine based on the traction load sensor or motor current; For any network segment, the set of traction points bound to the corresponding network segment is determined according to the network virtual state chain, and the traction force of the integrated machine transmitted through the trolley is reduced according to the preset force transmission node friction reduction coefficient to obtain the effective traction force of each traction point. Determine the angle between the effective traction force direction of the traction point and the projection direction of the sealing section onto the horizontal plane, and multiply the effective traction force of each traction point by the cosine value of the corresponding angle to obtain the horizontal projection component of the traction point. The equivalent tension of each sealing segment is obtained by summing the horizontal projection components whose cosine value of the included angle is greater than zero.

5. The method for real-time monitoring of the network closure status of a transmission line according to claim 4, characterized in that, Methods for generating traction constraint anomaly markers include: Calculate the product of the preset friction coefficient or preset equivalent friction parameter and the wheel pressure of the integrated machine, and then subtract the traction force of the integrated machine to obtain the friction margin of the integrated machine. When the friction margin is less than the preset friction margin threshold, it is determined that the corresponding integrated machine has the risk of slippage or insufficient anchoring. When the equivalent tension of the sealing section is less than the preset minimum effective tension, the sealing section is marked as a suspected abnormal stress section; When the increase in the traction force of the integrated machine corresponding to the blocked section is greater than the preset threshold for the increase in the traction force of the integrated machine in adjacent sampling times, and the increase in the travel distance of the integrated machine is less than the preset threshold for the increase in travel distance in adjacent sampling times, the blocked section will be marked as a suspected stuck section. When it is determined that the corresponding integrated machine has a risk of slippage or insufficient anchoring, or is marked as a suspected jamming section or a suspected abnormal force section, a traction constraint abnormality mark is formed.

6. The method for real-time monitoring of the network closure status of a transmission line according to claim 4, characterized in that, Methods for calculating the estimated sag and estimated lateral position of the netting segment include: The horizontal distance and azimuth of the netting section are calculated based on the coordinates of the constraint nodes at both ends of the netting section, and the wind speed component perpendicular to the direction of the netting section is determined based on the wind speed, wind direction and azimuth of the netting section. Based on the estimated sag and horizontal distance of the previous moment, the preset parabolic approximation conditions are determined. When the preset parabolic approximation conditions are met, the parabolic approximation model is adopted, and the estimated sag of the netting section is calculated based on the wind speed component, horizontal distance and equivalent tension. When the preset parabolic approximation conditions are not met, the catenary model is adopted, and the estimated sag of the netting section is calculated based on the wind speed component, horizontal distance and equivalent tension. Project the constraint nodes at both ends of the enclosure segment onto the plane of the preset target enclosure area, and use the average of the lateral coordinates of the two projection points as the estimated lateral position.

7. The method for real-time monitoring of the network closure status of a transmission line according to claim 6, characterized in that, Methods for obtaining network coverage integrity metrics include: The estimated sag and estimated lateral position of each netting segment are mapped to the plane of the preset target netting area to generate the boundary curve of each netting segment. The gap between the boundary curves of adjacent netting segments is filled in according to the preset boundary filling gap threshold to obtain the real-time estimated coverage area. The effective coverage area is obtained by subtracting the area of ​​the void area located inside the real-time estimated coverage area and the area corresponding to the uncovered gaps that have not been filled in the boundary from the overlapping area between the real-time estimated coverage area and the target closed area. Divide the effective coverage area by the target netting area to obtain the netting coverage integrity index.

8. The method for real-time monitoring of the network closure status of a transmission line according to claim 1, characterized in that, Methods for obtaining the corrected network closure status include: Arrange the estimated sag and estimated lateral position of each netting segment in the order of the netting segments to form the inversion state before visual correction; When the visual credibility of each data source is greater than or equal to the preset visual credibility threshold, coordinate registration is performed based on the position of the conductor obtained by video recognition. The visual observation value is composed of the registered net rope edge, net sheet edge or pulley position, and the visual observation noise matrix is ​​adjusted according to the visual credibility. The correction gain is calculated based on the adjusted visual observation noise matrix. The inversion state before visual correction is corrected according to the correction gain and the visual observation value to obtain the corrected netting state. When the visual credibility is less than the preset visual credibility threshold, the inversion state before visual correction is taken as the corrected network closure state.

9. A method for real-time monitoring of the network closure status of a transmission line according to claim 1, characterized in that, Methods for calculating the comprehensive risk score include: Based on the corrected netting status, the estimated sag and estimated lateral position of each netting segment are obtained; The sag risk score is calculated based on the estimated sag amount and the preset maximum allowable sag amount. The lateral offset risk score is calculated based on the estimated lateral position. The net coverage integrity risk score is calculated based on the net coverage integrity index. The traction constraint risk item is determined based on the traction constraint anomaly marker. The comprehensive risk score is obtained by multiplying the network coverage integrity risk score, the maximum value of the sagging risk score of each network segment, the maximum value of the lateral offset risk score of each network segment, and the result after subtracting the average confidence of multi-source data from one, along with the traction constraint risk item, by the corresponding preset risk weights and summing them.

10. A method for real-time monitoring of the network closure status of a transmission line according to claim 1, characterized in that, Methods for outputting the network closure status level include: Calculate the minimum credibility of the key data source based on the credibility of each data source; When the average credibility of multi-source data is less than the preset threshold for insufficient monitoring credibility, or when the minimum credibility of the key data source is less than the preset threshold for the credibility of the key data source, the output network closure status level is insufficient monitoring credibility. In the absence of output monitoring credibility, when the comprehensive risk score is greater than or equal to the preset abnormal state risk threshold, the output network closure status level is abnormal. When the comprehensive risk score is greater than or equal to the preset risk threshold for the state of concern, but less than the preset risk threshold for the state of abnormality, or when the average credibility of multi-source data is less than the preset credibility threshold for the state of normality, the output network closure status level is network closure status of concern. When the comprehensive risk score is less than the preset risk threshold for the state of concern, and the average credibility of the multi-source data is greater than or equal to the preset credibility threshold for the state of normal, the output network closure status level is normal.

11. A real-time monitoring system for the closure status of transmission lines, characterized in that, A method for real-time monitoring of the grid closure status of a transmission line as described in any one of claims 1-10, comprising: The multi-source data module is used to acquire multi-source data according to a preset sampling period from the start of the net sealing operation, establish a virtual state chain and net sealing segment based on the multi-source data, and calculate the credibility of each data source. The anomaly marking module is used to estimate the traction force of the integrated machine based on multi-source data, estimate the equivalent tension of each sealing segment based on the traction force of the integrated machine and the virtual state chain of the sealing network, and form traction constraint anomaly markings. The coverage evaluation module is used to calculate the estimated sagging and estimated lateral position of each netting segment based on the netting segment, multi-source data and equivalent tension, and to calculate the netting coverage integrity index based on the estimated sagging and estimated lateral position. The risk scoring module is used to visually correct the inversion state composed of the estimated droop and the estimated lateral position based on the credibility of each data source, to obtain the corrected netting state, calculate the mean of the credibility of each data source to obtain the average credibility of the multi-source data, and calculate the comprehensive risk score based on the corrected netting state, the average credibility of the multi-source data, the traction constraint anomaly marker and the netting coverage integrity index. The network closure level module is used to output the network closure status level based on the comprehensive risk score and the average credibility of multi-source data.

Images