Substation environment monitoring and risk assessment unmanned aerial vehicle system and unmanned aerial vehicle recovery method

Abstract

The invention relates to the technical field of intelligent operation and maintenance of electric power facilities, in particular to a transformer substation environment monitoring and risk assessment unmanned aerial vehicle system and an unmanned aerial vehicle recovery method, and the transformer substation environment monitoring and risk assessment unmanned aerial vehicle system comprises an unmanned aerial vehicle recovery system, an unmanned aerial vehicle carrier platform and a monitoring system; wherein the unmanned aerial vehicle recovery system is used for accurately positioning and guiding recovery of an unmanned aerial vehicle; wherein the unmanned aerial vehicle carrier platform is used for carrying a monitoring system and related equipment; wherein the monitoring system is used for performing real-time monitoring on environmental parameters and equipment operation states of the transformer substation and analyzing acquired data so as to assess operation risks. According to the invention, centimeter-level positioning in a strong electromagnetic environment is realized, the problem of navigation drifting of a traditional unmanned aerial vehicle is solved, and the recovery success rate is high.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation and maintenance of power facilities, and is an unmanned aerial vehicle (UAV) system for substation environment monitoring and risk assessment and a UAV recovery method. Background Art

[0002] Currently, the monitoring of substation equipment status mainly relies on a mode that combines a fixed sensor network (such as temperature probes, gas detectors) with periodic manual inspections. According to the "State Grid Equipment Operation and Maintenance White Paper (2023)", due to the limited installation positions of fixed sensors, only about 65% of the surface areas of key equipment can be covered, resulting in high-risk parts such as the top of transformer bushings and the joints of GIS equipment gas chambers being in the monitoring blind area for a long time. Although manual inspections have the advantage of flexibility, due to the limited operation frequency (usually once a week) and differences in personnel skills, the missed inspection rate is as high as 18% - 22%. A case of bushing overheating failure in a 500 kV substation in 2022 showed that it took 72 hours from the occurrence of the abnormality to the manual discovery, with a direct economic loss exceeding 8 million yuan. More severely, parameters such as temperature, partial discharge, and gas collected by traditional means are scattered in independent systems, lacking the ability of spatio-temporal alignment and multi-dimensional correlation analysis, and it is difficult to construct a global portrait of the equipment health status.

[0003] In recent years, although UAV inspection technology has been gradually applied in the power industry, its applicability in substation scenarios still has significant bottlenecks. The strong electromagnetic environment in substations (magnetic field intensity > 100 μT) causes UAV navigation signal drift, and the positioning error is generally > 2 m, increasing the collision risk during UAV recovery; the mainstream solutions rely on visible light / infrared dual-light pods and cannot synchronously detect key risk indicators such as SF6 leakage, ultraviolet characteristics of partial discharge, and abnormal equipment noises; in emergency scenarios such as thunderstorms and strong winds, UAVs are extremely prone to crashing, and when the UAV crashes, the high-value tool monitoring system installed on the UAV will be damaged. Summary of the Invention

[0004] The present invention provides an unmanned aerial vehicle (UAV) system for substation environment monitoring and risk assessment and a UAV recovery method, which overcomes the above-mentioned deficiencies of the prior art and can effectively solve the problems of UAV navigation signal drift caused by the strong electromagnetic environment in substations and the extremely high probability of UAV crashing in emergency scenarios such as thunderstorms and strong winds.

[0005] One of the technical solutions of the present invention is achieved by the following measures: An unmanned aerial vehicle (UAV) system for substation environment monitoring and risk assessment includes a UAV recovery system, a UAV vehicle platform, and a monitoring system; Among them, the UAV recovery system is used for accurately positioning and guiding the recovery of the UAV; Among them, the UAV vehicle platform is used for carrying the monitoring system and related equipment; Among them, the monitoring system is used to monitor the environmental parameters and equipment operation status of the substation in real time, and analyze the collected data to evaluate the operation risks.

[0006] The following is a further optimization or / and improvement of one of the above-mentioned inventive technical solutions: The above-mentioned UAV recovery system may include a recovery platform, and the recovery platform includes a landing platform. The upper surface of the landing platform is provided with QR code pasting areas around its perimeter. The QR code pasting areas are used to paste QR codes with specific coding information, and the QR codes can be recognized by the UAV.

[0007] The lower end surface of the above-mentioned landing platform may be provided with a pre-embedded base, and the pre-embedded base is integrally formed by casting with a reinforced concrete structure.

[0008] The center of the upper surface of the above-mentioned landing platform may be provided with a cross-shaped guiding light and a ring-shaped guiding light.

[0009] The above-mentioned UAV vehicle platform uses a multi-rotor UAV, and a monitoring system is mounted on the lower end of the multi-rotor UAV; the monitoring system includes a mounting seat, a second motor is installed on the lower surface of the mounting seat, the output shaft of the second motor is connected to an L-shaped bracket, a sealed box body is installed on the outer side surface of the L-shaped bracket, a high-pressure gas cylinder is installed inside the sealed box body, a mounting frame is installed on the lower end surface of the L-shaped bracket, a U-shaped mounting seat is rotatably connected to the outer side surface of the mounting frame, an integrated detection device is installed on the upper surface of the U-shaped mounting seat, annular shells are provided at both the front and rear ends of the outer side surface of the mounting frame, an annular cabin is provided inside the annular shell, an uninflated airbag is provided inside the annular cabin, an arc-shaped cover plate is snap-connected to the outer side surface of the annular cabin, the high-pressure gas cylinder is connected to the uninflated airbag through a gas conduit, a solenoid valve is installed on the gas conduit, and a first motor is installed on the outer side surface of the mounting frame, and the output shaft of the first motor is connected to the rotating shaft of the U-shaped mounting seat.

[0010] The outer side surface of the above-mentioned sealed box body may be provided with a detachable cover plate.

[0011] The above-mentioned integrated detection device may include a broadband infrared thermal imager, an ultraviolet pulse detection unit, a high-precision SF6 gas sensor, an acoustic vibration sensor array, an acoustic-optic alarm device, and a gyroscope; Among them, the broadband infrared thermal imager is used to monitor the temperature distribution of the equipment and detect potential overheating hazards; Among them, the ultraviolet pulse detection unit is used to detect partial discharge and evaluate the insulation performance of the equipment; Among them, the high-precision SF6 gas sensor is used to detect SF6 gas leakage and concentration changes; Among them, the acoustic vibration sensor array is used to collect the vibration acoustic signals of the equipment and judge the operation status; Among them, the acoustic-optic alarm device is used to emit acoustic-optic signals when the equipment is abnormal to remind personnel to handle; Among them, the gyroscope is used to monitor the attitude of the drone and compensate for the attitude changes during sensor scanning.

[0012] The second technical solution of the present invention is achieved by the following measures: A method for recovering a substation environment monitoring and risk assessment drone, including the following steps: Step S1: QR code detection and corner point positioning based on the YOLOv5m algorithm; Step S2: Perform multi-sensor data fusion, adjust the flight trajectory of the drone according to the fused pose information, and achieve a stable landing.

[0013] The following is a further optimization or / and improvement of the second technical solution of the above invention: The above step S1 may include: Step S11: The drone is equipped with a downward-looking camera to collect RGB images of the takeoff and landing platform area; Step S12: Add an attention mechanism to the Backbone of YOLOv5m; Step S13: Use ZBar or the QRCodeDetector of OpenCV to parse the information embedded in the QR code and locate the coordinates of the four corners of the QR code; Step S14: Use the PnP algorithm to solve the pose and rotation matrix of the drone relative to the takeoff and landing platform.

[0014] The above step S2 may include: Step S21: Synchronize the time series of vision, IMU, and barometer, and normalize the data to a unified coordinate system; Step S22: Design a Kalman filter to fuse visual pose, IMU integrated displacement, attitude angle, and barometer height; Step S23: Fuse vision, IMU, and barometer data, and output the positioning coordinates and attitude of the drone in the takeoff and landing platform coordinate system; Step S24: Adjust the flight trajectory of the drone according to the fused pose information, and combine the cross-shaped guiding lights and the ring-shaped guiding lights to achieve a stable landing.

[0015] The present invention is applicable to the real-time monitoring of substation equipment status, early warning of potential safety hazards, and emergency response. By improving the YOLOv5m model and multi-sensor fusion algorithm, centimeter-level positioning in a strong electromagnetic environment is achieved, solving the problem of navigation drift of traditional drones. The recovery success rate is high. The present invention integrates infrared, ultraviolet, SF6 gas, and vibration sensors, supports 360° rotation scanning and spatio-temporal data fusion, and has a low missed detection rate; the innovative annular airbag ejection system reduces the equipment crash loss rate. Through the "improved YOLOv5m + multi-sensor fusion" algorithm, the present invention reduces the drone recovery positioning error from >2m in the traditional solution to ≤11.5cm, significantly avoiding the navigation drift problem in a strong electromagnetic environment; the visual IMU barometer Kalman filter fusion strategy realizes timestamp alignment and noise suppression (visual noise weight > IMU), dynamically compensating for signal distortion caused by electromagnetic interference; in the embodiments of the present invention, the improved YOLOv5m model enhances the small target detection ability through the attention mechanism, combined with data augmentation in low-light / occluded scenarios, and the QR code recognition success rate is increased to 98.7% (traditional method <85%), ensuring precise recovery under extreme conditions such as thunderstorms and at night; integrating multi-dimensional sensors such as broadband infrared (20°C to 1500°C), ultraviolet pulse (160 - 300nm), SF6 gas (0.1ppm resolution), and acoustic vibration (20Hz - 20kHz), synchronously collecting 12 types of key parameters such as temperature, partial discharge, gas leakage, and mechanical vibration, breaking through the limitation of the single modality of traditional dual-optical pods; the present invention realizes 360° rotation scanning of the sensor through the L-type bracket motor driving the U-shaped mounting seat, combined with gyroscope attitude compensation, constructs a 3D thermal distribution map of the equipment surface and an SF6 diffusion path model, and the potential hazard positioning accuracy reaches the 5cm² level of the equipment surface; the innovative annular airbag ejection system (response time <0.2s), controlled by a high-pressure gas cylinder and an electromagnetic valve, and the air pressure trigger threshold is dynamically adjusted according to the height sensor. Experiments show that it can absorb 87% of the impact energy, increasing the survival rate of the monitoring system from 40% in the traditional solution to 92%; the combination of cross-shaped guiding lights and annular guiding lights provides multi-spectral landing guidance, combined with the wind-resistant structure of the embedded base, realizing the stable recovery of the drone under 8-level wind conditions, and the wind resistance ability is increased by 60% compared with the traditional platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic structural diagram of the recovery platform according to an embodiment of the present invention.

[0017] Figure 2 It is a schematic structural diagram of the multi-rotor drone according to an embodiment of the present invention.

[0018] Figure 3 It is a schematic structural diagram of the monitoring system according to an embodiment of the present invention.

[0019] Figure 4 It is a front view of the monitoring system according to an embodiment of the present invention.

[0020] Figure 5 This is the left view of the monitoring system according to the embodiment of the present invention.

[0021] The codes in the attached drawings are respectively: 1 is the embedded base, 2 is the landing platform, 3 is the cross-shaped guiding light, 4 is the annular guiding light, 5 is the QR code pasting area, 6 is the multi-rotor UAV, 7 is the monitoring system, 71 is the mounting seat, 72 is the L-shaped bracket, 73 is the sealed box body, 74 is the mounting frame, 75 is the integrated detection device, 76 is the first motor, 77 is the annular shell, 78 is the arc-shaped cover plate, 79 is the second motor, 710 is the solenoid valve, 711 is the high-pressure gas cylinder, 712 is the U-shaped mounting seat. Specific embodiments

[0022] The present invention is not limited by the following embodiments, and the specific implementation manners can be determined according to the technical solutions of the present invention and the actual situation. In the description of the present invention, if the orientation description is involved, such as "upper", "lower", "front", "rear", "left", "right", etc., the orientation or positional relationship indicated is based on the Figure 4 orientation or positional relationship shown in the attached drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention. When a certain feature is referred to as "set", "fixed", "connected" to another feature, it can be directly set, fixed, connected to another feature, or indirectly set, fixed, connected to another feature.

[0023] The present invention will be further described below in conjunction with the embodiments: Embodiment 1: As shown in Figure 1 , 2 , 3, 4, 5, the substation environmental monitoring and risk assessment UAV system includes a UAV recovery system, a UAV vehicle platform and a monitoring system; Among them, the UAV recovery system is used for precise positioning and guiding the recovery of the UAV; Among them, the UAV vehicle platform is used for carrying the monitoring system and related equipment; Among them, the monitoring system is used for real-time monitoring of the environmental parameters and equipment operation status of the substation, and analyzing the collected data to evaluate the operation risk.

[0024] In the embodiments of the present invention, various sensors in the monitoring system collect in real time the environmental parameters of the substation (such as temperature, humidity, concentration of harmful gases, etc.) and the equipment operation status data (such as the surface temperature of the equipment, vibration conditions, electrical parameters, etc.). The unmanned aerial vehicle (UAV) platform takes the monitoring system to a suitable position above the substation to ensure that the monitoring system can effectively collect data. The UAV recovery system locates the UAV that has completed the monitoring task and guides it to land and be recovered. After the data is collected, it is transmitted to the analysis module, and specific algorithms are used to analyze and process the data to evaluate the operation risk of the substation. Through this, all-round and real-time monitoring and risk assessment of the substation are achieved. The automated monitoring process significantly reduces the workload of manual inspections and improves the monitoring efficiency. Timely detection of potential risks helps to take measures in advance to avoid accidents and ensure the safe and stable operation of the substation. At the same time, the UAV recovery system ensures the safe recovery of the UAV, improves the use efficiency and service life of the UAV.

[0025] In the embodiments of the present invention, the UAV recovery system includes a recovery platform, and the recovery platform includes a landing platform 2. Four sides of the upper surface of the landing platform 2 are respectively provided with two-dimensional code pasting areas 5 for pasting two-dimensional codes with specific coding information, and the two-dimensional codes can be recognized by the UAV. During the landing process of the UAV, an image of the landing platform area is captured through an image recognition device (such as a camera and related image recognition algorithms) carried by the UAV itself, and the two-dimensional code on the landing platform is recognized. The specific coding information in the two-dimensional code includes information such as the position and direction of the landing platform. The UAV calculates its position and attitude deviation relative to the landing platform according to the recognized two-dimensional code information, and then adjusts its flight trajectory to align with the landing platform for landing. Through this, a precise visual positioning identifier is provided for the UAV to land, significantly improving the accuracy and reliability of the UAV landing. Even in a complex environment (such as light changes, background interference, etc.), the two-dimensional code can still be effectively recognized, ensuring that the UAV can land stably, reducing the probability of recovery failure, and improving the coherence and safety of the UAV operation.

[0026] In the embodiment of the present invention, a pre-embedded base 1 is provided on the lower end surface of the landing platform 2, and the pre-embedded base 1 is integrally cast in a reinforced concrete structure. The pre-embedded base is pre-buried underground during construction. By closely combining with the surrounding soil or foundation, and utilizing its large mass and stable structure, it provides stable support for the landing platform. When the drone lands, the landing platform transmits the impact force to the pre-embedded base. The pre-embedded base integrally cast in a reinforced concrete structure, relying on its high strength and integrity, disperses and bears the impact force, maintaining the stable position of the landing platform. In this way, the stability and impact resistance of the landing platform are greatly enhanced, which can effectively cope with the impact force generated when drones of different weights land, as well as the influence of natural environmental factors (such as strong winds, earthquakes, etc.) on the landing platform. It ensures the long-term stable operation of the recovery platform, reduces the drone landing accidents caused by the instability of the platform, improves the service life of the recovery platform, and reduces the maintenance cost and safety risk.

[0027] In the embodiment of the present invention, a cross-shaped guiding light 3 and a ring-shaped guiding light 4 are provided at the center of the upper surface of the landing platform 2. Under low light or complex weather conditions, the cross-shaped guiding light and the ring-shaped guiding light turn on and emit high-brightness light. The drone identifies the light characteristics of the guiding lights through its own visual sensors (such as cameras), determines the center position of the landing platform according to the cross-shaped guiding light, and the ring-shaped guiding light assists in judging the boundary and angle of the landing platform, so as to determine its precise position and attitude relative to the landing platform, adjust the flight trajectory, and complete the landing action. In this way, it provides reliable visual guidance for the drone to land in adverse environments, compensates for the deficiencies of QR code positioning in poor light conditions, etc., and improves the adaptability and safety of the drone landing. The prominent signs of the guiding lights help the drone quickly and accurately find the landing platform, shorten the landing time, and further improve the recovery efficiency.

[0028] In the embodiment of the present invention, the unmanned aerial vehicle (UAV) platform adopts a multi-rotor UAV, and a monitoring system 7 is mounted at the lower end of the multi-rotor UAV; the monitoring system 7 includes a mounting seat 71, a second motor 79 is mounted on the lower surface of the mounting seat 71, the output shaft of the second motor 79 is connected to an L-shaped bracket 72, a sealed box body 73 is mounted on the outer side surface of the L-shaped bracket 72, a high-pressure gas cylinder 711 is mounted inside the sealed box body 73, a mounting frame 74 is mounted on the lower end surface of the L-shaped bracket 72, a U-shaped mounting seat 712 is rotatably connected to the outer side surface of the mounting frame 74, an integrated detection device 75 is mounted on the upper surface of the U-shaped mounting seat 712, annular shells 77 are provided at both the front and rear ends of the outer side surface of the mounting frame 74, an annular cabin is provided inside the annular shell 77, an uninflated airbag is provided inside the annular cabin, an arc-shaped cover plate 78 is snap-connected to the outer side surface of the annular cabin, the high-pressure gas cylinder 711 is connected to the uninflated airbag through a gas conduit, a solenoid valve 710 is mounted on the gas conduit, a first motor 76 is mounted on the outer side surface of the mounting frame 74, and the output shaft of the first motor 76 is connected to the rotating shaft of the U-shaped mounting seat 712. The multi-rotor UAV has the characteristics of vertical takeoff and landing, hovering in the air, and flexible steering, and can fly freely in the complex environment of the substation, facilitating reaching the positions that need to be monitored, improving the flexibility and coverage of the monitoring, and enabling comprehensive monitoring of all corners of the substation. It can adapt to different weather and environmental conditions. Compared with other types such as fixed-wing UAVs, it is more suitable for operating in substations with limited space and many obstacles. By connecting the L-shaped bracket 72 through the second motor 79, the angle and direction of components such as the sealed box body 73 and the integrated detection device 75 can be adjusted by rotation, which can optimize the monitoring perspective according to the actual monitoring requirements and ensure accurate acquisition of the required data. The mounting seat 71 serves as the basic support structure and provides a stable mounting platform for the entire monitoring system. The reasonable design and connection of components such as the L-shaped bracket 72 and the mounting frame 74 ensure the structural stability of the system during flight, reduce the influence of factors such as the vibration of the UAV during flight on the monitoring equipment, and are beneficial to improving the accuracy of the monitoring data. The sealed box body 73 can protect internal components such as the high-pressure gas cylinder 711, prevent dust, water vapor, sundries, etc. from entering the inside of the box body, avoid damage to equipment such as the high-pressure gas cylinder or affecting its performance caused by these factors, and extend the service life of the equipment. Placing the high-pressure gas cylinder in the sealed box body improves the safety to a certain extent, prevents the high-pressure gas cylinder from being accidentally collided, squeezed, etc., and reduces the safety risk. The U-shaped mounting seat 712 is rotatably connected to the mounting frame 74, and the output shaft of the first motor 76 is connected to the rotating shaft of the U-shaped mounting seat 712, enabling the integrated detection device 75 to rotate, thereby realizing monitoring in different directions. This is beneficial for comprehensively detecting substation equipment and obtaining more comprehensive operating status information. The angle of the integrated detection device 75 can be accurately adjusted according to actual needs, enabling it to more accurately align with the monitoring target and improving the accuracy and effect of the detection.When the drone encounters unexpected situations (such as out of control, collision, etc.) during flight or landing, the solenoid valve 710 opens, and the high-pressure gas cylinder 711 inflates the uninflated airbag through the air duct, causing the airbag to expand and deploy rapidly. This can play a buffering and protective role, reducing the impact on the drone and monitoring equipment, lowering the risk of equipment damage, and ensuring the safety of the equipment and personnel. When the safety protection device is not damaged, it can be reused after simple maintenance and inspection, improving the practicality and economy of the device.

[0029] In the embodiment of the present invention, the mounting seat 71 is fixed on the multi-rotor drone. The mounting seat is firmly connected to the fuselage of the multi-rotor drone through fixing devices such as bolts and buckles. The monitoring system and related equipment are installed on the mounting seat. When the drone is flying, the mounting seat moves with the drone. By virtue of its stable connection with the drone, it ensures that the equipment maintains a relatively stable position and posture during flight and will not shake or displace due to factors such as the vibration and airflow of the drone during flight. In this way, a stable and reliable installation foundation is provided for the monitoring equipment, ensuring the stability of the equipment during flight, enabling the monitoring equipment to accurately collect data. It avoids the influence of equipment shaking on the data collection accuracy, improves the accuracy and reliability of the monitoring data, and provides high-quality data support for subsequent risk assessment. At the same time, it facilitates the installation and disassembly of the equipment, which is conducive to the maintenance and update of the equipment.

[0030] In the embodiment of the present invention, the outer side of the sealed box body 73 is provided with a detachable cover plate. The sealed box body is used to accommodate the equipment or components that need to be protected (such as some environment-sensitive sensors, electronic components, etc.). The detachable cover plate is connected to the box body through devices such as hinges and locks. When it is necessary to repair, maintain or replace the equipment inside the box, open the connection devices such as locks and remove the cover plate, then the internal equipment can be operated. After the operation is completed, reinstall the cover plate and ensure sealing to prevent external impurities such as dust and water vapor from entering the inside of the box body. In this way, it facilitates the maintenance and repair of the equipment inside the sealed box body, improving the maintainability of the equipment. The sealing design protects the internal equipment from the influence of the external harsh environment, extends the service life of the equipment, reduces the monitoring system failure caused by the damage of the equipment due to environmental factors, and improves the stability and reliability of the operation of the entire system.

[0031] In the embodiment of the present invention, the integrated detection device 75 includes a broadband infrared thermal imager, an ultraviolet pulse detection unit, a high-precision SF6 gas sensor, an acoustic vibration sensor array, an acoustic-optical warning device, and a gyroscope; Among them, the broadband infrared thermal imager is used to monitor the temperature distribution of the equipment and discover potential overheating hazards; Among them, the ultraviolet pulse detection unit is used to detect partial discharge and evaluate the insulation performance of the equipment; Among them, a high-precision SF6 gas sensor is used to detect SF6 gas leakage and concentration changes; Among them, an acoustic vibration sensor array is used to collect the vibration acoustic signals of the equipment and judge the operating state; Among them, an acoustic-optic alarm device is used to emit acoustic-optic signals when the equipment is abnormal to remind personnel to handle it; Among them, a gyroscope is used to monitor the attitude of the unmanned aerial vehicle and compensate for the attitude changes during sensor scanning.

[0032] The broadband infrared thermal imager receives the infrared radiation emitted by an object, converts it into a thermal image, and intuitively displays the temperature distribution of the equipment according to the temperature differences represented by different colors. When the temperature exceeds the normal range, potential overheating hazards can be judged. The ultraviolet pulse detection unit detects the ultraviolet pulse signals emitted by partial discharges generated during the operation of the equipment, and evaluates the insulation performance of the equipment according to parameters such as signal intensity and frequency. The high-precision SF6 gas sensor uses a specific sensing technology to detect SF6 gas in the environment. When there is gas leakage, the sensor senses the change in gas concentration and outputs a corresponding signal. The acoustic vibration sensor array is arranged around the equipment to collect the vibration acoustic signals generated during the operation of the equipment, and judges the operating state of the equipment by analyzing the characteristics such as the frequency and amplitude of the signals. When any of the above sensors detects abnormal data, the acoustic-optic alarm device is triggered to emit strong light and an alarm sound to remind the staff. The gyroscope monitors the attitude changes of the unmanned aerial vehicle in real time. When the attitude of the unmanned aerial vehicle changes, it provides attitude compensation data for sensor scanning to ensure that the sensor can always accurately collect target data. By integrating detection equipment with multiple functions in this way, the substation equipment is comprehensively monitored from multiple dimensions, and various potential problems in the operation of the equipment, such as overheating, partial discharge, gas leakage, and mechanical failure, can be discovered in a timely and accurate manner. The acoustic-optic alarm device timely reminds the staff to handle abnormal situations, avoids the expansion of faults, and ensures the safe operation of the substation. The attitude compensation function of the gyroscope improves the accuracy and reliability of the data collected by the sensor and enhances the performance of the entire monitoring system.

[0033] When in use, the usage method of this system is designed based on the closed-loop process of "autonomous inspection, multi-modal data collection, real-time analysis, and safe recovery", and the specific steps are as follows: A1. System initialization and takeoff preparation A1.1. Deployment of the recovery platform: Install and embed the base 1 in the preset safe area of the substation, and fix the landing platform 2 to ensure that its level error ≤ 0.5°. Stick two-dimensional codes 5 around the upper surface of the landing platform. The two-dimensional codes encode the origin, direction, and safety boundary information of the takeoff and landing platform coordinate system. Start the cross-shaped guiding light 3 (red light, stroboscopic mode) and the annular guiding light 4 (blue light, constant-on mode) to provide multi-spectral visual guidance for the unmanned aerial vehicle.

[0034] A1.2. Configuration of the unmanned aerial vehicle carrier platform: Mount the monitoring system 7 to the bottom of the multi-rotor UAV through the mounting base 71, and check the rotational flexibility of the first motor 76 and the second motor 79. Calibrate the sensors of the integrated detection device 75: Broadband infrared thermal imager: Calibrate the temperature measurement error (±1°C) with a standard blackbody radiation source.

[0035] SF6 gas sensor: Calibrate the sensitivity by introducing SF6 gas with a standard concentration (50 ppm).

[0036] Ultraviolet pulse detection unit: Adjust the gain to recognize a partial discharge intensity ≥5 pC.

[0037] A1.3. Task planning and navigation initialization: Import the 3D point cloud map of the substation through the ground control station, and delimit the inspection path (avoiding high-risk areas such as lightning rods and busbars). Start the improved YOLOv5m model, load the pre-trained weights and the anti-interference dataset (including fuzzy and low-light scenarios). Activate the multi-sensor fusion algorithm, and set the Kalman filter parameters (process noise covariance Q = 0.01, observation noise covariance R = 0.1).

[0038] A2. Autonomous inspection and data collection A2.1. Takeoff and navigation: The UAV autonomously takes off to the preset height (10 m), and fuses the data from the IMU (100 Hz) and the barometer (10 Hz) in real time to resist electromagnetic interference (magnetic field strength > 100 μT), and keep the hovering positioning error ≤15 cm.

[0039] A2.2. Multi-modal sensing collaborative scanning: The second motor 79 drives the L-shaped bracket 72 to rotate horizontally, and the first motor 76 controls the vertical swing of the U-shaped mounting base 712 to achieve 360° omnidirectional coverage of the sensor.

[0040] Data synchronous acquisition: Broadband infrared thermal imager: Take pictures of the surface temperature distribution of the equipment (resolution 640×512, frame rate 30 Hz), and identify overheating points (>80°C). Ultraviolet pulse detection unit: Capture the ultraviolet signals of partial discharge (160 - 300 nm), and mark the discharge intensity and position.

[0041] SF6 gas sensor: Detect the leakage concentration (range 0 - 1000 ppm, response time < 3 s), and generate a gas diffusion thermal diagram.

[0042] Acoustic vibration sensor: Collect the abnormal sound spectrum of the equipment (20 Hz - 20 kHz), and identify mechanical looseness or insulation deterioration by combining FFT analysis.

[0043] A2.3. Real-time risk assessment and warning The on-board edge computing unit performs spatio-temporal alignment and correlation analysis on multi-source data: Coupling model: Establish a joint diagnosis model for temperature, partial discharge, and vibration (e.g., trigger a first-level alarm when the temperature gradient anomaly is accompanied by a partial discharge pulse > 100 pC).

[0044] Acousto-optic alarm device: If the SF6 concentration > 50 ppm or the surface temperature rise rate of the equipment > 5 °C / min is detected, start the red strobe and buzzer alarm.

[0045] The data is transmitted back to the substation SCADA system in real time to generate the Equipment Health Index (EHI) and maintenance work orders.

[0046] A3. Emergency Response and Safe Recovery A3.1. Rapid Recovery in Extreme Weather When the meteorological sensor detects a wind speed > 12 m / s or a thunderstorm warning, the ground station sends an emergency recovery command. The UAV switches to the anti-interference navigation mode: Vision-dominated: The downward-looking camera captures the QR code area at 30 fps, and improves the YOLOv5m model to output detection frames in real time (confidence > 0.9).

[0047] Sub-pixel corner optimization: Locate the QR code corners through the ShiTomasi algorithm, and combine PnP to solve the relative pose of the UAV takeoff and landing platform (error ≤ 11.5 cm).

[0048] Multi-sensor fusion: The Kalman filter dynamically weights the vision data (weight 0.7), IMU data (weight 0.2), and barometer data (weight 0.1) to suppress the attitude jitter caused by electromagnetic interference.

[0049] A3.2. Airbag Ejection Protection (Crash Scenario): If the UAV altitude sensor detects a free fall acceleration > 2g and the altitude < 5 m, trigger the ring airbag protection system: The solenoid valve 710 is instantaneously opened (response time < 0.2 s), and the high-pressure gas cylinder 711 releases compressed gas to inflate the uninflated airbag into a ring-shaped buffer layer within 0.5 s. The arc-shaped cover plate 78 is automatically ejected by the airbag expansion pressure to avoid direct collision between the monitoring system 7 and the ground, and the impact energy absorption rate > 87%.

[0050] A3.3. Precise Landing and Data Transmission: In a complex electromagnetic environment, the UAV slowly descends along the combined light track of the cross-shaped guiding light 3 and the ring-shaped guiding light 4, and fuses the vision and IMU data to adjust the attitude angle in real time (pitch / roll angle error < 1°). After landing, the sensor power supply is automatically turned off, and the complete inspection data packet (including infrared thermal images, partial discharge videos, gas concentration curves, etc.) is uploaded through the 5G private network.

[0051] Example 2: This example discloses a method for recovering a UAV for substation environment monitoring and risk assessment, including the following steps: Step S1: QR code detection and corner point localization based on the YOLOv5m algorithm; Step S2: Perform multi-sensor data fusion, adjust the UAV flight trajectory according to the fused pose information, and achieve stable landing.

[0052] In step S1, the camera carried by the UAV captures images of the landing platform area. The YOLOv5m algorithm processes the images, identifies the QR code, and locates the four corner points of the QR code through a specific algorithm. According to the corner point coordinates and the pre-set positional relationship between the QR code and the landing platform, the approximate position information of the UAV relative to the landing platform is determined. In step S2, various sensors on the UAV, such as a vision sensor (obtaining QR code positioning information), an IMU (inertial measurement unit, measuring acceleration, angular velocity, etc.), and a barometer (measuring altitude), collect data in real time. After synchronizing the data of these different types of sensors in time and normalizing them, the data is input into a data fusion algorithm such as a Kalman filter. The algorithm performs weighted fusion on the data according to the characteristics of each sensor's data to obtain more accurate UAV pose information (position and attitude). The UAV flight control system compares the fused pose information with the target landing position and attitude, calculates the deviation, and then adjusts the flight trajectory to achieve stable landing. Through QR code detection and corner point localization based on the YOLOv5m algorithm, the use of advanced image recognition technology improves the accuracy and speed of UAV landing positioning. Multi-sensor data fusion makes full use of the advantages of each sensor, makes up for the limitations of a single sensor, and improves the accuracy of UAV pose information. The accurate pose information enables the UAV to precisely adjust the flight trajectory, greatly improving the success rate and stability of UAV landing, ensuring the safe recovery of the UAV, and guaranteeing the smooth completion of the entire monitoring task.

[0053] In the embodiment of the present invention, step S1 includes: Step S11: The UAV carries a downward-looking camera to collect RGB images of the takeoff and landing platform area; Step S12: Add an attention mechanism to the Backbone of YOLOv5m; Step S13: Use ZBar or the QRCodeDetector of OpenCV to parse the information embedded in the QR code and locate the coordinates of the four corner points of the QR code; Step S14: Use the PnP algorithm to solve the pose and rotation matrix of the UAV relative to the takeoff and landing platform.

[0054] Specifically, step S1 includes the following steps: Step S11: Image acquisition: The UAV carries a downward-looking camera to collect RGB images of the takeoff and landing platform area in real time; Step S12, Object Detection: Add an attention mechanism to the Backbone of YOLOv5m to improve the detection ability for small-sized QR codes; Dataset Enhancement: Add synthetic data for scenarios such as blur, low light, and occlusion to improve robustness.

[0055] Output: Detect the QR code area in the image; Step S13, QR Code Decoding and Corner Extraction: Decoding Algorithm: Use ZBar or the QRCodeDetector in OpenCV to parse the information embedded in the QR code; Corner Detection Algorithm: Use the Shi-Tomasi corner detection algorithm to locate the coordinates of the four corners of the QR code; Further optimize the corner position to sub-pixel accuracy of 0.1 pixel through sub-pixel corner optimization; Step S14, Relative Pose Calculation: PnP Algorithm: Input: Pixel coordinates (2D) of the QR code corners and preset physical coordinates (3D); Output: The pose and rotation matrix of the drone relative to the takeoff and landing platform.

[0056] In Step S11, during the descent of the drone, the downward-facing camera captures real-time images of the takeoff and landing platform area to obtain visual data containing information such as QR codes. In Step S12, an attention mechanism is added to the Backbone network structure of the YOLOv5m algorithm, enabling the model to pay more attention to the feature regions related to the QR code when processing images, suppressing the interference of irrelevant background information, and improving the accuracy and robustness of QR code feature extraction. In Step S13, the QRCodeDetector library in ZBar or OpenCV is used to parse the QR code in the captured image, obtain the encoded information embedded in the QR code, and use its positioning algorithm to determine the pixel coordinates of the four corners of the QR code in the image. In Step S14, based on the known three-dimensional coordinates of the four corners of the QR code in the world coordinate system (takeoff and landing platform coordinate system) and the two-dimensional pixel coordinates obtained in Step S13, the PnP (Perspective-n-Point) algorithm is used to calculate the position (translation vector) and attitude (rotation matrix) of the drone relative to the takeoff and landing platform. Through these steps, Step S11 provides the basic image data for subsequent QR code detection and positioning. Step S12 enhances the algorithm's detection ability for QR codes in complex environments by improving the algorithm structure, enabling accurate QR code recognition even in the presence of partial occlusion, uneven lighting, etc. Step S13 accurately parses the QR code information and locates the corners, providing key data for precise calculation of the drone's pose. Step S14 uses the PnP algorithm to obtain the accurate pose of the drone, providing the core position and attitude information for the drone's landing, significantly improving the accuracy of the drone's landing positioning, and ensuring that the drone can accurately align with the landing platform.

[0057] In an embodiment of the present invention, step S2 includes: Step S21: Synchronize the time series of vision, IMU, and barometer, and normalize the data to a unified coordinate system; Step S22: Design a Kalman filter to fuse the vision pose, IMU integrated displacement, attitude angle, and barometer height; Step S23: Fuse the vision, IMU, and barometer data, and output the positioning coordinates and attitude of the UAV in the coordinate system of the takeoff and landing platform; Step S24: Adjust the flight trajectory of the UAV according to the fused pose information, and combine the cross-shaped guiding light 3 and the annular guiding light 4 to achieve a stable landing.

[0058] Specifically, step S2 includes the following steps: Step S21, Sensor data synchronization and preprocessing: Timestamp alignment: Synchronize the time series of vision data (camera), IMU (100Hz), and barometer (10Hz) through hardware triggering or software interpolation; Data normalization: Convert the acceleration a, angular velocity ω, and barometer height h of the IMU to a unified coordinate system; Step S22, Kalman filter design: State variables: , including position, velocity, and Euler angles; Observation model: Vision observation: The pose (T, R) output by PnP; IMU observation: Calculate the displacement by integrating the acceleration, and calculate the attitude angle by the gyroscope; Barometer observation: Altitude z; Prediction step (state equation): F: State transition matrix (based on the dynamic model of the IMU).

[0059] Uk: Control input (such as motor thrust).

[0060] wk: Process noise (Gaussian distribution).

[0061] Update step (observation equation): H: Observation matrix (fusing vision, IMU, and barometer); vk: Observation noise (vision noise < IMU noise); Kalman gain calculation: Update the state estimate and covariance matrix: Step S23, Output High-Precision Three-Dimensional Coordinates: The fused state estimation Provide the real-time position (x, y, z) and attitude (θ, ϕ, ψ) of the UAV in the landing platform coordinate system, with an accuracy of centimeter level (≤11.5 cm).

[0062] In step S21, since the data collection frequencies and time bases of the vision sensor, IMU, and barometer may be different, the data collected by these sensors are aligned in time through a hardware synchronization circuit or a software time calibration algorithm. At the same time, the data of each sensor is converted from its own measurement coordinate system to a unified UAV flight coordinate system, facilitating subsequent data fusion processing. In step S22, a Kalman filter is designed, which constructs a state equation and an observation equation according to the characteristics of the data of each sensor and the noise statistical model. The pose information obtained by the vision sensor, the displacement and attitude angle information obtained by integrating the IMU, and the altitude information measured by the barometer are used as observation values and input into the filter. The filter uses the state estimation value at the previous moment and the observation value at the current moment, and through two steps of prediction and update, optimally estimates the current state (position and attitude) of the UAV. In step S23, the Kalman filter outputs the fused result, that is, the accurate positioning coordinates (x, y, z) and attitude information (pitch angle, roll angle, yaw angle) of the UAV in the landing platform coordinate system. In step S24, the UAV flight control system compares the fused pose information with the pose information of the pre-set landing point, calculates the deviation value. According to the deviation value, by adjusting the control parameters such as the throttle and rudder surface of the UAV, the flight trajectory is changed. At the same time, combined with the visual guidance information provided by the cross-shaped guiding lights and the annular guiding lights on the landing platform, the flight trajectory is further fine-tuned to finally achieve a stable landing. Step S21 ensures the consistency of different sensor data in time and space, laying a foundation for accurate data fusion. Step S22 effectively fuses multi-source data through the Kalman filter, reduces the noise and error effects of single-sensor data, and improves the accuracy and reliability of the UAV pose information. Step S23 outputs high-precision positioning and attitude information, providing an accurate basis for flight trajectory adjustment. Step S24 comprehensively utilizes the fused pose information and guiding light information to achieve the stable landing of the UAV in a complex environment, improves the success rate and safety of UAV recovery, and ensures the smooth end of the monitoring mission.

[0063] The above technical features constitute the embodiments of the present invention, which have strong adaptability and implementation effects. Non-essential technical features can be added or subtracted according to actual needs to meet the requirements of different situations.

Claims

1. An unmanned aerial vehicle system for substation environment monitoring and risk assessment, characterized in that It includes a drone recovery system, a drone vehicle platform, and a monitoring system; Among them, the drone recovery system is used to accurately locate and guide the recovery of the drone; Among them, the drone vehicle platform is used to carry the monitoring system and related equipment; Among them, the monitoring system is used to monitor the environmental parameters and equipment operation status of the substation in real time, and analyze the collected data to evaluate the operation risk.

2. The substation environment monitoring and risk assessment UAV system according to claim 1, characterized in that The drone recovery system includes a recovery platform, and the recovery platform includes a landing platform. QR code pasting areas are provided around the upper surface of the landing platform. The QR code pasting areas are used to paste QR codes with specific coding information, and the QR codes can be recognized by the drone.

3. The substation environment monitoring and risk assessment UAV system according to claim 2, characterized in that The lower end surface of the landing platform is provided with a pre-embedded base, and the pre-embedded base is integrally cast in a reinforced concrete structure.

4. The substation environment monitoring and risk assessment UAV system according to claim 2 or 3, characterized in that A cross-shaped guiding light and a ring-shaped guiding light are provided in the center of the upper surface of the landing platform.

5. The substation environment monitoring and risk assessment UAV system according to claim 1 or 2 or 3, characterized in that The drone vehicle platform uses a multi-rotor drone, and the monitoring system is mounted on the lower end of the multi-rotor drone; the monitoring system includes a mounting seat, a second motor is mounted on the lower surface of the mounting seat, the output shaft of the second motor is connected to an L-shaped bracket, a sealed box body is mounted on the outer side surface of the L-shaped bracket, a high-pressure gas cylinder is mounted inside the sealed box body, a mounting frame is mounted on the lower end surface of the L-shaped bracket, a U-shaped mounting seat is rotatably connected to the outer side surface of the mounting frame, an integrated detection device is mounted on the upper surface of the U-shaped mounting seat, annular shells are provided at both the front and rear ends of the outer side surface of the mounting frame, an annular cabin is provided inside the annular shell, an uninflated airbag is provided inside the annular cabin, an arc-shaped cover plate is buckled to the outer side surface of the annular cabin through a buckle, the high-pressure gas cylinder is connected to the uninflated airbag through a gas duct, a solenoid valve is mounted on the gas duct, and a first motor is mounted on the outer side surface of the mounting frame, and the output shaft of the first motor is connected to the rotating shaft of the U-shaped mounting seat.

6. The substation environment monitoring and risk assessment UAV system according to claim 5, characterized in that A detachable cover plate is provided on the outer side surface of the sealed box body.

7. The substation environment monitoring and risk assessment UAV system according to claim 1 or 2 or 3 or 6, characterized in that The integrated detection device includes a broadband infrared thermal imager, an ultraviolet pulse detection unit, a high-precision SF6 gas sensor, an acoustic vibration sensor array, an acoustic-optic alarm device, and a gyroscope; Among them, the broadband infrared thermal imager is used to monitor the temperature distribution of the equipment and detect potential overheating hazards; Among them, the ultraviolet pulse detection unit is used to detect partial discharge and evaluate the insulation performance of the equipment; Among them, the high-precision SF6 gas sensor is used to detect SF6 gas leakage and concentration changes; Among them, the acoustic vibration sensor array is used to collect the vibration acoustic signals of the equipment and judge the operation status; Among them, the acoustic-optic alarm device is used to emit acoustic-optic signals when the equipment is abnormal to remind personnel to handle it; Among them, the gyroscope is used to monitor the attitude of the drone and compensate for the attitude changes during sensor scanning.

8. A method for recovering an unmanned aerial vehicle for substation environmental monitoring and risk assessment, characterized in that It includes the following steps: Step S1, QR code detection and corner positioning based on the YOLOv5m algorithm; Step S2, perform multi-sensor data fusion, adjust the flight trajectory of the drone according to the fused pose information, and achieve stable landing.

9. The method for recovering an unmanned aerial vehicle for substation environment monitoring and risk assessment according to claim 8, wherein Step S1 includes: Step S11: The drone carries a downward-looking camera to collect RGB images of the takeoff and landing platform area; Step S12: Add an attention mechanism to the Backbone of YOLOv5m; Step S13: Use ZBar or QRCodeDetector of OpenCV to parse the information embedded in the QR code and locate the coordinates of the four corners of the QR code; Step S14: Use the PnP algorithm to calculate the pose and rotation matrix of the UAV relative to the takeoff and landing platform.

10. The method for recovering an unmanned aerial vehicle for substation environment monitoring and risk assessment according to claim 8 or 9, characterized in that Step S2 includes: Step S21: Synchronize the time series of vision, IMU, and barometer, and normalize the data to a unified coordinate system; Step S22: Design a Kalman filter to fuse visual pose, IMU integrated displacement, attitude angle, and barometer altitude; Step S23: Fuse vision, IMU, and barometer data, and output the positioning coordinates and attitude of the UAV in the takeoff and landing platform coordinate system; Step S24: Adjust the flight trajectory of the UAV according to the fused pose information, and combine the cross-shaped guiding lights and the ring-shaped guiding lights to achieve a stable landing.

Images