Multi-mode perception and hierarchical active intervention method and system for external damage risk of power distribution network

Abstract

The application discloses a kind of multi-mode perception and hierarchical active intervention method and system of distribution network external damage risk, and it relates to distribution network safety protection technical field.The method includes: through millimeter wave radar, the distance, azimuth and pitch angle data of moving target are collected and processed to obtain radar point cloud, and the image data of moving target is synchronously collected by camera;Through edge computing unit, the moving target is classified based on image data using visual recognition algorithm to distinguish ordinary targets from targets with long arm structure;Through edge computing unit, based on the spatial distribution density and geometric shape of radar point cloud, the independent point cloud cluster of arm support component is separated from the overall radar point cloud of target with long arm structure;Through edge computing unit, based on the independent point cloud cluster of arm support component, the trajectory prediction and collision probability prediction of arm tip of arm support component are carried out using Kalman filtering.The application realizes the external damage risk prediction and early warning of distribution network line.

Description

Technical Field

[0001] This invention relates to the field of power distribution network security protection technology, specifically to a multi-mode sensing and hierarchical proactive intervention method and system for external damage risks to power distribution networks. Background Technology

[0002] Distribution networks (including overhead power lines, power cables, and auxiliary equipment such as pole-mounted switches and prefabricated substations) are widely distributed and operate in complex environments, frequently intersecting with municipal construction and transportation activities. External damage (hereinafter referred to as "external damage") is one of the main causes of power outages and equipment damage in distribution networks. Currently, external damage protection for distribution networks mainly relies on manual inspections, static warning signs, and simple audible and visual alarm devices, which are insufficient to meet the increasingly complex requirements of urban construction environments for the safe operation of distribution networks. Existing technologies have the following main shortcomings.

[0003] (a) Early warning of external damage to overhead power lines

[0004] 1. Existing external damage identification schemes for power distribution networks lack the ability to predict future trajectories.

[0005] Existing technologies mostly utilize millimeter-wave radar to dynamically monitor the spatial environment of the power distribution network, combine it with lidar for tracking and monitoring, and identify hazard categories through integrated intelligent models. Although they can identify existing hazard sources and issue warnings, their technical approach focuses on matching and identifying the characteristics of potential hazards at the current moment: after obtaining real-time detection information through radar, they extract hazard characteristic parameters and input them into the model to identify the hazard category, and then generate warning information. However, they do not make any form of prediction or extrapolation of the future trajectory of the target.

[0006] 2. Trajectory prediction schemes for transmission lines are not applicable to distribution network scenarios.

[0007] In the field of power transmission lines, trajectory prediction technology has begun to be introduced. Its prediction logic is to judge the overall movement trend of the target, that is, to determine whether the target will intrude into the warning zone of the line. Power transmission lines have tall towers and large spans, so the main threat comes from the spatial intrusion of large machinery. However, the safety distance between overhead lines and the ground in distribution networks is relatively low (usually only 6-9 meters). The main threat is not the intrusion of the entire target, but rather the contact between the boom tip and the conductor during the lifting or rotation of long-arm machinery (such as crane booms, concrete pump truck booms, and aerial work platform booms). In this case, the vehicle chassis may be far beyond the safety distance, with only the boom portion intruding into the danger zone. Existing overall trajectory prediction schemes in the field of power transmission lines cannot specifically model and predict the independent lifting / rotation movements of robotic arms. Directly transplanting these schemes to the distribution network scenario would result in a large number of missed detections.

[0008] 3. Existing solutions lack specific sensing and prediction capabilities for the motion characteristics of "boom-type" devices.

[0009] Currently, there is no solution specifically addressing the unique risk of overhead power lines in distribution networks being struck by long-armed machinery, which involves independently extracting the motion state of the target's movable components (the boom), modeling its spatial trajectory, and predicting collision probabilities. Existing technologies either only perform static ranging or only assess the intrusion of the target as a whole, neither of which can solve the unique external damage problem of distribution networks where "the vehicle has not crossed the boundary, but the boom has already touched the line."

[0010] (ii) Early warning of external damage to underground cables

[0011] 1. "Touch alarms" remain the mainstream, lacking early proximity warnings.

[0012] Currently, the mainstream solutions for preventing external damage to underground cables are still "touch-based" or "proximity-based" alarms. An alarm is triggered when excavation tools damage the cable. Underground power cable damage prevention early warning systems based on fiber optic vibration sensing technology install fiber optic vibration sensors on or near the cable, triggering an alarm upon detecting vibration. The essence of this type of solution is "alarm only after damage occurs": when the cable is broken or the vibration sensor senses a strong impact, the cable itself is often already damaged or about to be damaged.

[0013] 2. The voice fingerprint recognition solution is limited to "recognition" and does not achieve "proximity warning".

[0014] Existing solutions can detect and issue early warnings in real time using acoustic fingerprints of excavation activities through acoustic positioning sensors. They can identify excavation activities and pinpoint the direction of the sound source. However, their early warning logic is simply "early warning upon detection of excavation," without dynamically comparing the real-time location of the excavation point with the cable route, or implementing tiered early warnings based on the speed and trend of the excavation point approaching the cable. In other words, existing technology can only tell you "someone is digging," but cannot determine "how far away they are from the cable and how fast they are approaching," thus failing to issue differentiated warnings matching the risk level when the excavation point is still far away.

[0015] 3. The problem of false alarms of vibration signals in complex ground environments of power distribution networks is prominent.

[0016] Distribution network cables are typically buried directly beneath urban roads or in shallow conduits, making ground vibration interference sources extremely complex: vehicle traffic, pedestrian trampling, and rail transit can all generate vibration signals. Existing vibration monitoring solutions have limited accuracy in identifying weak excavation signals amidst strong background noise, resulting in a high false alarm rate that leads to a "crying wolf" effect among maintenance personnel, ultimately causing the system to be abandoned.

[0017] (III) Active intervention methods

[0018] Existing warning methods are mostly limited to on-site audible and visual alarms (flashing warning lights, blaring horns). When the cab is enclosed, the construction machinery is noisy, or the operator's attention is distracted, the on-site audible and visual signals are difficult to perceive effectively. Current technology lacks a means to directly send strong warning information to the cab / operator's cab, making it impossible to effectively intervene against the perpetrator during the golden window of opportunity when risks arise. Summary of the Invention

[0019] This invention addresses the problems existing in the identification and early warning of external damage risks in distribution networks by providing a multi-mode perception and hierarchical proactive intervention method and system for external damage risks in distribution networks, thereby enabling the prediction and early warning of external damage risks to distribution network lines.

[0020] Firstly, a multi-mode sensing and hierarchical proactive intervention method for external damage risks in distribution networks is provided, the method comprising:

[0021] The range, azimuth and elevation angle data of the moving target are collected and processed by millimeter-wave radar to obtain radar point cloud, and image data of the moving target are collected synchronously by camera. The millimeter-wave radar and the camera are installed on the poles of the overhead power line of the power distribution network.

[0022] Using an edge computing unit and a visual recognition algorithm, the moving target is classified based on the image data to distinguish between ordinary targets and targets with long-arm structures.

[0023] Using the edge computing unit, based on the spatial distribution density and geometry of the radar point cloud, independent point cloud clusters of the boom component are separated from the overall radar point cloud of the target with the long-arm structure.

[0024] Using the edge computing unit, based on the independent point cloud clusters of the boom component, Kalman filtering is used to predict the trajectory and collision probability of the boom tip.

[0025] In some embodiments, the edge computing unit, based on the spatial distribution density and geometry of the radar point cloud, separates independent point cloud clusters of the boom component from the overall radar point cloud of the target with the long-arm structure, including:

[0026] Based on the spatial distribution density and geometric shape of the overall radar point cloud of the target with the long arm structure, identify the movable parts of the target with the long arm structure that are long strip-shaped, fixed at one end to the vehicle body, and can rotate around the fixed point.

[0027] Extract the independent motion parameters of the movable component, wherein the independent motion parameters include at least: boom lifting angular velocity, boom rotation angular velocity, current spatial coordinates of the boom tip, and boom extension length.

[0028] In some embodiments, the edge computing unit, based on the independent point cloud clusters of the boom component, uses Kalman filtering to perform trajectory prediction and collision probability prediction on the boom tip of the boom component, including:

[0029] Establish the motion state equation of the arm tip, wherein the state vector of the motion state equation includes: the three-dimensional position, three-dimensional velocity and three-dimensional acceleration of the arm tip;

[0030] Based on the arm tip coordinate data measured by the millimeter-wave radar, the arm tip observation equation is constructed using the motion state equation.

[0031] The data of the independent motion parameters are input into the arm tip observation equation, and the prediction step and update step of the Kalman filter are used to iterate and output the predicted spatial position and confidence interval of the arm tip at several future moments in real time.

[0032] The highest point of the boom tip in the predicted spatial location is compared with the lowest point of the sag of the power distribution line by spatial vertical distance. At the same time, the horizontal position of the boom tip is compared with the horizontal projection of the power distribution line.

[0033] When the predicted vertical distance in space is less than the first safety threshold and the horizontal projection enters the protection range on both sides of the line, it is determined as a potential risk and an early warning is issued; when the predicted vertical distance in space is less than the second safety threshold, it is determined as a high-risk risk and an early warning is issued, wherein the first safety threshold is greater than the second safety threshold.

[0034] In some embodiments, the method further includes:

[0035] Vibration signals are collected by multiple vibration sensing nodes and sent to the edge computing unit. The multiple vibration sensing nodes are set at preset intervals along the route of the underground cables of the power distribution network, and each of the multiple vibration sensing nodes has a built-in vibration sensor.

[0036] Using the edge computing unit and a pre-stored vibration signal classification model, the construction vibration fingerprint is identified and separated from the vibration signal based on the vibration fingerprint database. The construction vibration fingerprint includes at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations.

[0037] The separated construction vibration fingerprints are processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach speed is determined based on the rate of change of distance.

[0038] Based on the shortest horizontal distance and the approach speed, a graded early warning system is implemented.

[0039] In some embodiments, calculating the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network includes:

[0040] The planar coordinates of the construction excavation point are calculated using the time difference of arrival algorithm for receiving signals from the same seismic source from the multiple vibration sensing nodes.

[0041] The planar coordinates of the construction excavation point are superimposed and compared with the pre-stored coordinates of the cable route geographic information system to calculate the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network in real time.

[0042] In some embodiments, a graded early warning system is implemented based on the shortest horizontal distance and the approach velocity, including:

[0043] If a construction vibration fingerprint is detected and the distance between the construction excavation point and the underground cable of the power distribution network is greater than the first threshold, it is marked as a warning of concern.

[0044] If the distance between the construction excavation point and the underground cable of the power distribution network is less than or equal to the first threshold and it is determined that the construction excavation is continuously approaching, it is marked as a warning level alert.

[0045] When the distance between the construction excavation point and the underground cable of the power distribution network is less than or equal to a second threshold, it is marked as an emergency warning, wherein the second threshold is less than the first threshold.

[0046] In some embodiments, on-site visual and auditory warnings are achieved through LED flashing lights and loudspeakers installed on the poles or smart ground spikes of the overhead power distribution lines.

[0047] A parametric array directional loudspeaker installed on the towers of the overhead power distribution line forms a highly directional sound beam pointing towards the driver's cab or operator's cab for early warning.

[0048] In some embodiments, early warning information is pushed to the person in charge of equipment operation and maintenance unit via wireless communication for early warning purposes. The early warning information pushed includes: a cable routing diagram and real-time distance data.

[0049] Secondly, a multi-mode sensing and hierarchical proactive intervention system for external damage risks in a power distribution network is provided. The system includes a millimeter-wave radar, a camera, and an edge computing unit. The millimeter-wave radar and the camera are mounted on the towers of overhead power lines in the power distribution network.

[0050] The millimeter-wave radar collects and processes the range, azimuth, and elevation data of the moving target to obtain a radar point cloud, and the camera simultaneously collects image data of the moving target.

[0051] Using an edge computing unit and a visual recognition algorithm, the moving target is classified based on the image data to distinguish between ordinary targets and targets with long-arm structures. Based on the spatial distribution density and geometric shape of the radar point cloud, independent point cloud clusters of the boom component are separated from the overall radar point cloud of the target with a long-arm structure. Based on the independent point cloud clusters of the boom component, Kalman filtering is used to predict the trajectory and collision probability of the boom tip.

[0052] In some embodiments, the system further includes: a plurality of vibration sensing nodes, wherein the plurality of vibration sensing nodes are arranged at preset intervals along the route of underground cables in the power distribution network, and each of the plurality of vibration sensing nodes has a built-in vibration sensor to form a distributed sensing network; wherein,

[0053] Vibration signals are collected through the multiple vibration sensing nodes and sent to the edge computing unit;

[0054] Using the edge computing unit and a pre-stored vibration signal classification model, the construction vibration fingerprint is identified and separated from the vibration signal based on the vibration fingerprint database. The construction vibration fingerprint includes at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations.

[0055] The separated construction vibration fingerprints are processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach speed is determined based on the rate of change of distance.

[0056] Based on the shortest horizontal distance and the approach speed, a graded early warning system is implemented.

[0057] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0058] For overhead power distribution lines, a "boom tip trajectory prediction" method for long-arm machinery in power distribution networks was designed. Unlike existing technologies that track and predict the entire target, this invention addresses the unique risk of overhead power distribution lines being struck by long-arm machinery. It independently isolates the boom component from the overall point cloud of the target and performs independent Kalman filter trajectory prediction on its boom tip. This directly predicts the spatial distance between the boom tip and the conductor sag, completely solving the unique external damage problem of "vehicles not crossing the boundary, but booms already touching the line" that is unsolvable by existing technologies in power distribution networks.

[0059] For underground cables in power distribution networks, this invention represents a qualitative leap from "touch-based alarm" to "proximity-based tiered early warning." Unlike existing induction cable solutions (alarm only after damage) and acoustic fingerprint recognition solutions (only identifying excavation, without distance judgment), this invention uses TDOA to accurately locate the coordinates of the excavation point and compare the distance with the cable route in real time. Based on the distance and approach speed, it implements tiered early warning, advancing the warning window from "the instant of touch" to "the dangerous approach stage (minutes)," allowing sufficient response time for equipment maintenance personnel.

[0060] By using an AI classification model to comprehensively identify multi-dimensional vibration characteristics (time domain, frequency domain, energy attenuation), it effectively distinguishes between construction excavation and background interference such as vehicle rolling and pedestrian trampling. This solves the practical problem that existing vibration monitoring solutions have high false alarm rates and are eventually abandoned in complex ground environments of power distribution networks, and significantly reduces the false alarm rate of underground cable vibration monitoring.

[0061] By using parametric array directional sound waves (penetrating cab noise) and 4G / 5G mobile phone SMS push, and utilizing the two-way intervention mechanism of "directly reaching the operator", the warning information is forcibly delivered to the operator and the equipment operation and maintenance unit is notified. This fundamentally overcomes the defect of traditional on-site sound and light alarms failing in closed cabs and strong mechanical noise environments, and significantly improves the actual effectiveness of the warning.

[0062] With a unified "cloud-edge-device" architecture and hierarchical early warning logic, it covers two completely different external damage scenarios: overhead power lines and underground cables in the power distribution network. This avoids the repeated investment by power companies in deploying multiple heterogeneous systems for different scenarios, and has good economic efficiency and scalability. Attached Figure Description

[0063] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart of a multi-mode sensing and hierarchical proactive intervention method for external damage risks in a distribution network according to an embodiment of the present invention.

[0065] Figure 2 This is a flowchart of another method for multi-mode sensing and hierarchical proactive intervention of external damage risks in distribution networks according to an embodiment of the present invention.

[0066] Figure 3 This is a schematic diagram of a multi-mode sensing and hierarchical active intervention system for external damage risk in a power distribution network according to an embodiment of the present invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0068] On the one hand, the present invention provides a multi-mode sensing and hierarchical proactive intervention method for external damage risks in power distribution networks. Figure 1 This is a flowchart illustrating a multi-mode sensing and hierarchical proactive intervention method for external damage risks in a distribution network according to an embodiment of the present invention. (Reference) Figure 1 The multi-mode perception and hierarchical proactive intervention method for external damage risks to the power distribution network includes: S10 to S16.

[0069] In S10, the range, azimuth and elevation angle data of moving targets are collected and processed by millimeter-wave radar to obtain radar point cloud, and image data of moving targets are collected synchronously by camera. The millimeter-wave radar and camera are installed on the poles of the overhead power line of the power distribution network.

[0070] Specifically, a (low-power) millimeter-wave radar and a (high-definition) camera are installed on the towers of overhead power lines in the power distribution network. The millimeter-wave radar uses a fan-beam scanning method to focus on covering the vehicle traffic area below and to the side of the overhead lines. It continuously collects distance, azimuth, and elevation angle data of moving targets at a fixed frequency (e.g., 10~20Hz), and processes this data to obtain point cloud data of the moving targets. Simultaneously, the camera synchronously collects image data of the moving targets.

[0071] In S12, the edge computing unit uses a visual recognition algorithm to classify moving targets based on image data, in order to distinguish between ordinary targets and targets with long-arm structures.

[0072] Specifically, edge computing units can utilize existing visual recognition algorithms to classify moving targets, distinguishing between ordinary targets (i.e., vehicles) and targets with long-arm structures (i.e., vehicles).

[0073] In S14, the edge computing unit separates the independent point cloud clusters of the boom component from the overall radar point cloud of the target with a long-arm structure, based on the spatial distribution density and geometry of the radar point cloud.

[0074] Specifically, unlike existing technologies that track the entire target, this invention targets targets with movable parts (including but not limited to truck cranes, concrete pump trucks, aerial work platforms, truck-mounted cranes, etc.) and separates independent point cloud clusters of boom components from the overall point cloud of the target by using the spatial distribution characteristics and temporal variation characteristics of radar point clouds.

[0075] Separating independent point cloud clusters of boom components from the overall target point cloud includes: identifying elongated, movable components in the target with a long arm structure that are fixed to the vehicle body at one end and can rotate around a fixed point, based on the spatial distribution density and geometry of the overall radar point cloud of the target with a long arm structure; and then extracting the independent motion parameters of the movable component, including but not limited to: boom lifting angular velocity, boom slewing angular velocity, current spatial coordinates of the boom tip, and boom extension length.

[0076] In S16, the edge computing unit uses Kalman filtering to predict the trajectory and collision probability of the boom tip based on the independent point cloud clusters of the boom component.

[0077] Specifically, the edge computing unit uses a Kalman filter to recursively estimate and predict the spatial coordinate sequence of the boom tip of the boom component, so as to realize the trajectory prediction and collision probability prediction of the boom tip of the boom component.

[0078] In some embodiments, the edge computing unit performs trajectory prediction and collision probability prediction on the boom tip of the boom component, including: establishing a motion state equation for the boom tip, wherein the state vector of the motion state equation includes: the three-dimensional position, three-dimensional velocity, and three-dimensional acceleration of the boom tip; then, based on the boom tip coordinate data measured by millimeter-wave radar, the motion state equation is used to construct a boom tip observation equation; next, the data of independent motion parameters are input into the boom tip observation equation, and the prediction step and update step of Kalman filtering are used for iteration to output the boom tip in real time at several moments in the future (such as the next 3 seconds, 5 seconds, etc.). The predicted spatial location and confidence interval are calculated over 10 seconds. Finally, the vertical distance between the highest point of the arm tip and the lowest point of the sag of the distribution line in the predicted spatial location is compared. At the same time, the horizontal position of the arm tip is compared with the horizontal projection of the distribution line. When the predicted vertical distance is less than the first safety threshold (e.g., 3 meters) and the horizontal projection enters the protection range on both sides of the line, it is judged as a potential risk and an early warning is issued. When the predicted vertical distance is less than the second safety threshold (e.g., 1.5 meters), it is judged as a high-risk risk. The first safety threshold is greater than the second safety threshold and an early warning is issued.

[0079] In this invention, for overhead power distribution lines, the boom components of long-arm machinery are independently tracked in three-dimensional space (based on Kalman filtering trajectory prediction) and collision probability is predicted (using the horizontal and vertical distances between the boom tip and the sag of the power distribution line). This can solve the unique problem of power distribution networks where "the vehicle has not crossed the boundary but the boom has already touched the line".

[0080] In some embodiments, the multi-mode perception and hierarchical proactive intervention method for external damage risk of the power distribution network also includes the perception and prediction of external damage risk to underground cables in the power distribution network. Figure 2This is a flowchart of another method for multi-mode perception and hierarchical proactive intervention of external damage risk in distribution networks according to an embodiment of the present invention. Specifically, the perception and prediction of external damage risk to underground cables in the distribution network includes: S20 to S26.

[0081] In S20, vibration signals are collected by multiple vibration sensing nodes and sent to the edge computing unit. The multiple vibration sensing nodes are set at preset intervals along the route of the underground cables of the power distribution network, and each of the multiple vibration sensing nodes has a built-in vibration sensor.

[0082] Specifically, vibration sensing nodes can be deployed along the route of underground cables in the power distribution network (buried at a depth of approximately 0.5 to 1.5 meters) at preset intervals (e.g., 20 to 50 meters). Each node has a built-in vibration sensor, such as a triaxial high-precision MEMS (Micro-Electro-Mechanical Systems) accelerometer. Each node connects to the edge computing unit via a self-organizing network communication method (such as BLE Mesh or LoRa), featuring low power consumption, wireless deployment, and no need for road excavation.

[0083] In S22, the edge computing unit uses a pre-stored vibration signal classification model to identify and separate construction vibration fingerprints from vibration signals based on a vibration fingerprint database. The construction vibration fingerprints include at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations.

[0084] Specifically, the edge computing unit can have a built-in AI (artificial intelligence) classification model trained on massive samples. This model, based on pre-stored vibration fingerprints in a vibration fingerprint database, can identify and separate the following construction vibration fingerprints from complex road surface background vibration signals:

[0085] ① Temporal impact characteristics and frequency domain distribution of manual excavation (scraping);

[0086] ② Vibration waveform characteristics of mechanical excavation (bucket impacting the ground);

[0087] ③ The high-frequency vibration characteristics of the hydraulic breaker impact;

[0088] ④ Continuous vibration characteristics of drilling / pipe jacking operations.

[0089] In S24, the separated construction vibration fingerprint is processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach speed is determined based on the distance change rate.

[0090] Specifically, AI classification models can be used to make comprehensive judgments based on multi-dimensional features (time domain peak, frequency domain main frequency, energy decay curve, etc.), effectively filtering out background interference such as vehicle crushing and pedestrian trampling, and significantly reducing the false alarm rate.

[0091] In some embodiments, calculating the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network includes: calculating the planar coordinates of the construction excavation point using a Time Difference of Arrival (TDOA) algorithm based on the same source signal received by multiple vibration sensing nodes; then, superimposing and comparing the planar coordinates of the construction excavation point with the pre-stored cable routing geographic information system (GIS) coordinates to calculate the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network in real time.

[0092] In S26, graded early warnings are implemented based on the shortest horizontal distance and approach speed.

[0093] In some embodiments, a tiered early warning system is implemented based on the shortest horizontal distance and approach speed, including: a "highlight" warning is issued when a construction vibration fingerprint is detected and the distance between the excavation point and the underground cable of the power distribution network is greater than a first threshold (e.g., 5 meters); a "warning" warning is issued when the distance between the excavation point and the underground cable of the power distribution network is less than or equal to the first threshold and it is determined that the excavation is continuously approaching; and an "emergency" warning is issued when the distance between the excavation point and the underground cable of the power distribution network is less than or equal to a second threshold (e.g., 2 meters). The second threshold is less than the first threshold.

[0094] For example, different warning measures can be set for the aforementioned attention-level, warning-level, and emergency-level warnings. For attention-level warnings, event logs can be recorded and notifications can be pushed to the monitoring platform. For warning-level warnings, on-site audible and visual alarms can be activated, and a warning and cable location map can be pushed to the mobile phone of the person in charge of equipment operation and maintenance. For emergency-level warnings, the strongest audible and visual alarms can be activated, and an emergency signal can be sent to the distribution automation master station.

[0095] In this invention, for underground cables in power distribution networks, early identification of construction excavation activities is performed, and a proximity-based hierarchical early warning is achieved by comparing the dynamic distance between the construction excavation point and the route of the underground cables in the power distribution network. This advances the early warning window from the "moment of contact" to the "danger approach stage", reducing the false alarm rate.

[0096] In some embodiments, for early warning of external damage risks to distribution network lines, whether overhead lines or underground cables, on-site visual and auditory warnings can be achieved through LED flashing lights and loudspeakers installed on poles or smart ground nails of the overhead distribution network lines.

[0097] Meanwhile, for high-risk scenarios, parametric array directional loudspeakers installed on poles can be activated. These loudspeakers utilize the strong directivity of ultrasonic carrier waves to modulate audio signals onto ultrasonic frequencies for transmission. The signals then demodulate into audible sound in the air, forming a highly directional sound beam. This beam can be precisely directed towards the driver's cab or operator's compartment, ensuring that the warning sound reaches a high volume (e.g., above 90dB) only at that location, while the surrounding ambient noise level remains relatively constant. This ensures that the operator receives the warning without disturbing residential areas.

[0098] In addition to providing on-site visual and auditory warnings and directional loudspeaker warnings, the system can also automatically push early warning information to the person in charge of equipment operation and maintenance through the 4G / 5G (fourth generation / fifth generation) communication unit. The pushed content may include: cable routing diagram and real-time distance data.

[0099] Existing warning methods for external damage to power distribution lines all involve "sounding to the environment – ​​flashing warning lights and broadcasting through horns in all directions," and whether the operator notices it depends on their attention. However, in this invention, "sounding to the operator" is achieved – directional sound waves reach directly to the driver's cab, ensuring that the warning information is received by the operator, fundamentally solving the problem of traditional warnings failing in high-noise environments.

[0100] In this invention, a "boom tip trajectory prediction" method for long-arm machinery in power distribution networks is designed for overhead lines. Unlike existing technologies that track and predict the entire target, this invention addresses the unique risk of overhead lines being struck by long-arm machinery. It independently isolates the boom component from the overall point cloud of the target and performs independent Kalman filter trajectory prediction on its boom tip, directly predicting the spatial distance between the boom tip and the conductor sag. This completely solves the unique external damage problem of power distribution networks, where "the vehicle has not crossed the boundary but the boom has already touched the line," which is unsolvable by existing technologies.

[0101] Meanwhile, for underground cables in power distribution networks, this invention achieves a qualitative leap from "touch-based alarm" to "proximity-based graded early warning." Unlike existing induction cable solutions (alarm only after damage) and acoustic fingerprint recognition solutions (only identifying excavation, without distance judgment), this invention uses TDOA to accurately locate the coordinates of the excavation point and compare the distance with the underground cable route in real time. Based on the distance and approach speed, it implements graded early warning, advancing the early warning window from "the instant of touch" to "the dangerous approach stage (minutes)," allowing sufficient response time for equipment maintenance personnel.

[0102] By using an AI classification model to comprehensively identify multi-dimensional vibration characteristics (time domain, frequency domain, energy attenuation), it effectively distinguishes between construction excavation and background interference such as vehicle rolling and pedestrian trampling. This solves the practical problem that existing vibration monitoring solutions have high false alarm rates and are eventually abandoned in complex ground environments of power distribution networks, and significantly reduces the false alarm rate of underground cable vibration monitoring.

[0103] By using parametric array directional sound waves (which penetrate cab noise) and 4G / 5G mobile phone SMS push, and utilizing a two-way intervention mechanism that "directly reaches the operator," the warning information is forcibly delivered to the operator and the equipment maintenance unit is also notified. This fundamentally overcomes the shortcomings of traditional on-site sound and light alarms that fail in enclosed cabs and environments with strong mechanical noise, and significantly improves the actual effectiveness of the warning.

[0104] This invention adopts a unified "cloud-edge-device" architecture and hierarchical early warning logic, covering two completely different external damage scenarios: overhead power lines and underground cables in power distribution networks. This avoids the repeated investment of power companies in deploying multiple heterogeneous systems for different scenarios, and has good economic efficiency and scalability.

[0105] On the other hand, the present invention provides a multi-mode sensing and hierarchical proactive intervention system for external damage risks in power distribution networks. Figure 3 This is a schematic diagram of a multi-mode sensing and hierarchical proactive intervention system for external damage risks in a distribution network, according to an embodiment of the present invention. (Reference) Figure 3 The system includes: millimeter-wave radar, camera, and edge computing unit. The millimeter-wave radar and camera are installed on the poles of the overhead power lines of the power distribution network.

[0106] Range, azimuth, and elevation angle data of moving targets are acquired and processed using millimeter-wave radar to obtain radar point clouds, while image data of moving targets is simultaneously acquired via a camera. Using an edge computing unit and a visual recognition algorithm, moving targets are classified based on the image data to distinguish between ordinary targets and targets with long-arm structures. Based on the spatial distribution density and geometry of the radar point cloud, independent point cloud clusters of the boom components are separated from the overall radar point cloud of targets with long-arm structures. Based on these independent point cloud clusters of the boom components, Kalman filtering is used to predict the trajectory and collision probability of the boom tip.

[0107] To predict and warn of external damage risks to underground cables in power distribution networks, this multi-mode sensing and hierarchical proactive intervention system for external damage risks includes: multiple vibration sensing nodes set at preset intervals along the route of the underground cables, with each node having a built-in vibration sensor, thus forming a distributed sensing network. Each node connects to an edge computing unit via self-organizing network communication for data transmission.

[0108] Vibration signals are collected by multiple vibration sensing nodes and sent to the edge computing unit. The edge computing unit uses a pre-stored vibration signal classification model to identify and separate construction vibration fingerprints from the vibration signals. The construction vibration fingerprints include at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations. The separated construction vibration fingerprints are processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach velocity is determined based on the rate of change of distance. Based on the shortest horizontal distance and the approach velocity, a graded early warning is implemented.

[0109] In addition, the system is also equipped with a communication unit to send early warning information to the corresponding early warning devices, such as LED flashing lights and loudspeakers on the poles or smart ground nails of the power distribution network, and / or parametric array directional speakers. It can also push early warning information to the person in charge of equipment operation and maintenance unit through wireless communication.

[0110] The specific implementation functions of each structural unit in the multi-mode sensing and hierarchical active intervention system for external damage risks of the distribution network are described in the relevant description of the multi-mode sensing and hierarchical active intervention method for external damage risks of the distribution network, and will not be repeated here.

[0111] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-mode sensing and hierarchical proactive intervention method for external damage risks in distribution networks, characterized in that, The method includes: The range, azimuth and elevation angle data of the moving target are collected and processed by millimeter-wave radar to obtain radar point cloud, and image data of the moving target are collected synchronously by camera. The millimeter-wave radar and the camera are installed on the poles of the overhead power line of the power distribution network. Using an edge computing unit and a visual recognition algorithm, the moving target is classified based on the image data to distinguish between ordinary targets and targets with long-arm structures. Using the edge computing unit, based on the spatial distribution density and geometry of the radar point cloud, independent point cloud clusters of the boom component are separated from the overall radar point cloud of the target with the long arm structure. Using the edge computing unit, based on the independent point cloud clusters of the boom component, Kalman filtering is used to predict the trajectory and collision probability of the boom tip.

2. The method according to claim 1, characterized in that, Using the edge computing unit, based on the spatial distribution density and geometry of the radar point cloud, independent point cloud clusters of the boom component are separated from the overall radar point cloud of the target with the long-arm structure, including: Based on the spatial distribution density and geometric shape of the overall radar point cloud of the target with the long arm structure, identify the movable parts of the target with the long arm structure that are long strip-shaped, fixed at one end to the vehicle body, and can rotate around the fixed point. Extract the independent motion parameters of the movable component, wherein the independent motion parameters include at least: boom lifting angular velocity, boom rotation angular velocity, current spatial coordinates of the boom tip, and boom extension length.

3. The method according to claim 2, characterized in that, Using the edge computing unit, based on the independent point cloud clusters of the boom component, and employing Kalman filtering, trajectory prediction and collision probability assessment are performed on the boom tip of the boom component, including: Establish the motion state equation of the arm tip, wherein the state vector of the motion state equation includes: the three-dimensional position, three-dimensional velocity and three-dimensional acceleration of the arm tip; Based on the arm tip coordinate data measured by the millimeter-wave radar, the arm tip observation equation is constructed using the motion state equation. The data of the independent motion parameters are input into the arm tip observation equation, and the prediction step and update step of the Kalman filter are used to iterate and output the predicted spatial position and confidence interval of the arm tip at several future moments in real time. The highest point of the boom tip in the predicted spatial location is compared with the lowest point of the sag of the power distribution line by spatial vertical distance. At the same time, the horizontal position of the boom tip is compared with the horizontal projection of the power distribution line. When the predicted vertical distance in space is less than the first safety threshold and the horizontal projection enters the protection range on both sides of the line, it is determined as a potential risk and an early warning is issued; when the predicted vertical distance in space is less than the second safety threshold, it is determined as a high-risk risk and an early warning is issued, wherein the first safety threshold is greater than the second safety threshold.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Vibration signals are collected by multiple vibration sensing nodes and sent to the edge computing unit. The multiple vibration sensing nodes are set at preset intervals along the route of the underground cables of the power distribution network, and each of the multiple vibration sensing nodes has a built-in vibration sensor. Using the edge computing unit and a pre-stored vibration signal classification model, the construction vibration fingerprint is identified and separated from the vibration signal based on the vibration fingerprint database. The construction vibration fingerprint includes at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations. The separated construction vibration fingerprints are processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach speed is determined based on the rate of change of distance. Based on the shortest horizontal distance and the approach speed, a graded early warning system is implemented.

5. The method according to claim 4, characterized in that, Calculate the shortest horizontal distance between the construction excavation point and the underground cables of the power distribution network, including: The planar coordinates of the construction excavation point are calculated using the time difference of arrival algorithm for receiving signals from the same seismic source from the multiple vibration sensing nodes. The planar coordinates of the construction excavation point are superimposed and compared with the pre-stored coordinates of the cable route geographic information system to calculate the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network in real time.

6. The method according to claim 5, characterized in that, Based on the shortest horizontal distance and the approach speed, a tiered early warning system is implemented, including: If construction vibration fingerprints are detected and the distance between the construction excavation point and the underground cable of the power distribution network is greater than the first threshold, it is marked as a warning of concern. If the distance between the construction excavation point and the underground cable of the power distribution network is less than or equal to the first threshold and it is determined that the construction excavation is continuously approaching, it is marked as a warning level alert. When the distance between the construction excavation point and the underground cable of the power distribution network is less than or equal to a second threshold, it is marked as an emergency warning, wherein the second threshold is less than the first threshold.

7. The method according to claim 6, characterized in that, By installing LED flashing lights and loudspeakers on the poles or smart ground spikes of the overhead power distribution lines, on-site visual and auditory warnings are achieved, and A parametric array directional loudspeaker installed on the towers of the overhead power distribution line forms a highly directional sound beam pointing towards the driver's cab or operator's cab for early warning.

8. The method according to claim 7, characterized in that, Warning information is pushed to the person in charge of equipment operation and maintenance unit via wireless communication. The warning information includes: cable route diagram and real-time distance data.

9. A multi-mode sensing and hierarchical proactive intervention system for external damage risks in a distribution network, characterized in that, The system includes: a millimeter-wave radar, a camera, and an edge computing unit. The millimeter-wave radar and the camera are mounted on the towers of overhead power lines in the power distribution network. The millimeter-wave radar collects and processes the range, azimuth, and elevation data of the moving target to obtain a radar point cloud, and the camera simultaneously collects image data of the moving target. Using an edge computing unit and a visual recognition algorithm, the moving target is classified based on the image data to distinguish between ordinary targets and targets with long-arm structures. Based on the spatial distribution density and geometric shape of the radar point cloud, independent point cloud clusters of the boom component are separated from the overall radar point cloud of the target with a long-arm structure. Based on the independent point cloud clusters of the boom component, Kalman filtering is used to predict the trajectory and collision probability of the boom tip.

10. The system according to claim 9, characterized in that, The system further includes: multiple vibration sensing nodes and a communication unit. The multiple vibration sensing nodes are arranged at preset intervals along the route of the underground cables of the power distribution network, and each of the multiple vibration sensing nodes has a built-in vibration sensor to form a distributed sensing network. The communication unit is used to send early warning information to early warning equipment. Vibration signals are collected through the multiple vibration sensing nodes and sent to the edge computing unit; Using the edge computing unit and a pre-stored vibration signal classification model, the construction vibration fingerprint is identified and separated from the vibration signal based on the vibration fingerprint database. The construction vibration fingerprint includes at least: manual excavation, mechanical excavation, hydraulic breaker impact, and drilling / pipe jacking operations. The separated construction vibration fingerprints are processed to eliminate background interference, and the shortest horizontal distance between the construction excavation point and the underground cable of the power distribution network is calculated. The approach speed is determined based on the rate of change of distance. Based on the shortest horizontal distance and the approach speed, a graded early warning system is implemented.

Images