A method and system for unmanned aerial vehicle (UAV) power line inspection based on multimodal fusion

Abstract

This invention relates to a method and system for unmanned aerial vehicle (UAV) power line inspection based on multimodal fusion, belonging to the field of partial discharge detection. The system includes a meteorological station along the power line, a quadcopter UAV, a remote cloud server, and a mobile terminal. The meteorological station collects environmental meteorological information; the quadcopter UAV collects UHF and infrared temperature signals, receives flight control commands, and completes power line inspection tasks; the remote cloud server uses a multimodal fusion method to determine the type and location of partial discharge, calculates the discharge risk hazard index according to the discharge risk hazard index formula, incorporates it into the power system maintenance work plan, generates disposal suggestions, and sends them to the mobile terminals of maintenance personnel to complete the maintenance tasks. This invention integrates multimodal data from the power line environment, UHF data, and infrared thermal imaging data to detect partial discharge type and location, calculate the discharge hazard index, and generate disposal suggestions, achieving integrated management of power line inspection and maintenance.

Description

Technical Field

[0001] This invention belongs to the field of partial discharge detection, specifically relating to a method and system for unmanned aerial vehicle (UAV) power line inspection based on multimodal fusion. Background Technology

[0002] Power lines are characterized by their wide distribution, complex geographical environment, high voltage level, long transmission distance, diverse fault types, and high maintenance safety risks. Traditional manual inspections, which rely on personnel climbing towers or walking to inspect, are inefficient and unsafe. Existing drone inspections mostly use a single sensor (such as visible light or infrared), which cannot simultaneously acquire multi-dimensional information such as insulator corona discharge, abnormal equipment temperature, and mechanical structural damage. Furthermore, the response to dynamic factors such as sudden changes in the weather environment and temporary obstacles is delayed, resulting in a high rate of missed fault detection and significant safety hazards.

[0003] With the development of science and technology, new detection methods have emerged in power line inspection. For example, Chinese patent CN119247059A discloses a UHF partial discharge detection system and method mounted on a drone. This system mainly relies on UHF sensor signals mounted on the drone for detection, without considering data accuracy, the influence of meteorological environment, or multi-sensor collaborative optimization and intelligent analysis methods. Therefore, ensuring data accuracy and real-time performance, considering the influence of meteorological factors, and using multi-sensor collaborative optimization to detect partial discharge type and location, evaluate discharge hazards, and provide disposal recommendations are urgent technical problems that need to be solved. Summary of the Invention

[0004] The purpose of this invention is to solve the problem of accuracy in existing power line inspection and provide a UAV power line inspection method and system based on multimodal fusion.

[0005] To achieve the above objectives, the technical solution of the present invention is: a UAV power line inspection method based on multimodal fusion, comprising the following steps:

[0006] S1. The quadcopter drone flies along the power transmission line in a zigzag trajectory with a predetermined lateral swing amplitude. It uses a wide-area scanning mode to simultaneously collect UHF signals and infrared temperature matrix signals and send them to a remote cloud server.

[0007] S2. The remote cloud server acquires data from the nearest power line meteorological station to the quadcopter drone, corrects the data using UHF correction formula and infrared temperature correction formula, performs multi-scale decomposition of the UHF signal using Daubechies wavelet basis, extracts the energy proportion of the 1.5GHz band, and displays the UHF data and infrared thermal image.

[0008] S3. The remote cloud server calculates and adjusts the flight altitude and speed of the quadcopter drone based on the flight altitude control formula and flight speed control formula. When a pulse group of UHF pulses lasting ≥3 cycles and with a single cycle pulse number ≥50 is detected, it is suspected that there is partial discharge and controls the quadcopter drone to reduce its altitude for detailed detection.

[0009] S4. The remote cloud server controls the quadcopter drone to perform a spiral descent in the target area with an initial radius of R, a final radius of r, and a descent rate of v0. At each descent height H, a set of UHF data in the 1.2 / 1.5 / 1.8GHz frequency band and a high-resolution infrared thermal image are collected.

[0010] S5. The remote cloud server calculates the azimuth angle of the discharge source using the FDOA frequency difference positioning method, and establishes a partial discharge defect volume model by fusing infrared temperature matrix signals, and calculates the coordinates of the maximum temperature rise point.

[0011] S6. A four-dimensional feature vector is established on a remote cloud server, and the SVM support vector machine classifier rules are used to determine the type of partial discharge.

[0012] S7. The remote cloud server calculates the discharge risk hazard index based on the partial discharge type and the discharge risk hazard index formula, and uses the discharge risk hazard index as the priority of power system maintenance operations to be added to the power system maintenance operation plan.

[0013] S8. The remote cloud server sends information on the location, type, hazard index, and handling suggestions of partial discharge anomalies to the mobile terminals of maintenance personnel according to the power system maintenance operation plan, thereby completing the maintenance task.

[0014] Furthermore, in step S2, the UHF correction formula is: , among which, U correct U is the corrected signal amplitude. raw The original detection signal amplitude is given by T, ambient temperature is given by RH, ambient humidity is given by P, and α0, β0, γ0, and k are given by k. T These are all correction coefficients, determined using the least squares method;

[0015] The infrared temperature correction formula is as follows: , among which, T correct For the corrected infrared temperature, T meas The infrared module detects the temperature, ɛ is the emissivity of the target surface, n is the atmospheric attenuation coefficient, L is the distance to the target, and a, b, c, d, e, and f are correction coefficients determined using the least squares method.

[0016] Furthermore, in step S3, the flight altitude control formula is: , among which, S UHFUHF signal strength, h is flight altitude, S UHF ∈[40dBμV,80dBμV], h∈[3m,7m];

[0017] The flight speed control formula is as follows: Where v is the flight speed, if S UHF A deceleration mechanism was triggered at >60dBμV, reducing the quadcopter drone to 0.5m / s.

[0018] Furthermore, in step S5, the FDOA frequency difference positioning method includes using the mutual fuzzy function formula to calculate the two sets of antenna execution signals and using the target azimuth formula to calculate the target azimuth from the time delay difference and frequency shift difference corresponding to the extracted peak values;

[0019] The formula for the mutual fuzziness function is: , where t represents time, χ(τ, f) is the mutual ambiguity function, s1(t) and s2(t+τ) represent the signals received by the first group of antennas and the second group of antennas, respectively, τ represents the time delay of the signal received by the second group of antennas relative to the first group of antennas, and f represents the frequency shift of the signal received by the second group of antennas relative to the first group of antennas;

[0020] The formula for the target orientation is: Where θ is the target azimuth angle, c is the speed of light, f0 is the center frequency of the signal, d is the antenna spacing, and Δf is the frequency shift difference;

[0021] The partial discharge defect volume model integrates the azimuth coordinate system calculated by the FDOA frequency difference positioning method with the infrared temperature field coordinate system. A rigid body transformation matrix is ​​used to align the data space, and the discharge area is divided into a three-dimensional voxel grid. Each voxel is associated with a temperature value and azimuth weight. Based on the Bayesian criterion, the FDOA positioning probability and the infrared thermal field probability are fused to generate a three-dimensional probability density distribution cloud map of the discharge defect. The output is a comprehensive view model that includes azimuth annotation of the discharge point, thermal field contour overlay, and three-dimensional volume rendering.

[0022] Furthermore, in step S6, the four-dimensional feature vector is: , where Q m This indicates the proportion of wavelet energy in the 1.5GHz band. f represents the rate of temperature rise. peak N represents the main frequency of the UHF signal. pulses Indicates the number of pulses per second;

[0023] The SVM support vector machine classifier rule is: when f peak ∈[300,800]MHz and Q m <0.5 is considered corona discharge, when f peak∈[1200,1800]MHz and ≥5℃ / s is considered surface discharge, when N pulses >200 / s and Q m A value >0.7 indicates internal discharge.

[0024] Furthermore, in step S7, the discharge risk hazard index formula includes: corona discharge hazard index formula, surface discharge hazard index formula, and internal discharge hazard index formula.

[0025] The formula for the corona discharge hazard index is: ,in, RH represents ambient humidity, K a α is the corona resistance correction factor, β is the main frequency weighting index, and β is the high frequency energy weighting index.

[0026] The formula for the surface discharge hazard index is: ,in, K b γ is the surface insulation correction factor, and γ is the temperature rise rate weighting index;

[0027] The formula for the internal discharge hazard index is: ,in, K c λ is the equipment type correction factor, and λ is the pulse density weighting index.

[0028] Furthermore, in step S8, the treatment recommendations include corona discharge treatment recommendations, surface discharge treatment recommendations, and internal discharge treatment recommendations;

[0029] The corona discharge treatment recommendations include: H a ∈ When necessary, the system should be shut down for maintenance and repair, and deteriorated parts should be replaced; H a ∈ When necessary, clean the insulation surface and reduce ambient humidity; otherwise, strengthen inspections and monitor humidity and discharge frequency.

[0030] The recommended surface discharge treatment includes: H b ∈ Immediately stop operation for inspection and repair, and replace damaged parts; H b ∈ In some cases, local cleaning, repair, or application of anti-flashover coating should be performed; in other cases, the frequency of inspections should be increased, and humidity and temperature rise should be monitored.

[0031] The internal discharge handling recommendations include: H c ∈ If the situation occurs, immediately shut down the system and replace the entire insulation system; H c ∈ When necessary, perform local repairs, vacuum drying, or replace the seals; otherwise, periodically monitor pulse density and humidity.

[0032] This invention also provides a multimodal fusion-based unmanned aerial vehicle (UAV) power line inspection system for implementing any of the methods described above, comprising a meteorological station along the power line, a quadcopter UAV, a remote cloud server, and a mobile terminal; wherein,

[0033] The meteorological station along the power line is connected to the remote cloud server via a 4G / 5G network, and is used to collect environmental meteorological information along the power line and send the information to the remote cloud server.

[0034] The quadcopter drone is connected to the remote cloud server via a 4G / 5G network. It is used to synchronously collect UHF signals and infrared temperature matrix signals and send them to the remote cloud server, receive flight control commands from the remote cloud server, and complete power line inspection tasks.

[0035] The remote cloud server is connected to the mobile terminal via a 4G / 5G network. It is used to receive environmental information along the power line sent by the meteorological station along the power line and UHF signals and infrared temperature matrix signals collected synchronously by the quadcopter drone. It uses a multimodal fusion method to determine the type and location of partial discharge, calculates the discharge risk hazard index according to the discharge risk hazard index formula, adds it to the power system maintenance operation plan, generates disposal suggestions, and sends them to the mobile terminal of the maintenance personnel to complete the maintenance task.

[0036] Furthermore, the quadcopter drone includes a GPS-tamed rubidium atomic clock, a UHF probe, a conditioning circuit, an AD circuit, an FPGA controller, an infrared module, a tablet computer, and the drone body;

[0037] The GPS-disciplined rubidium atomic clock is connected to the FPGA controller I / O port to generate a synchronization pulse signal to synchronize the timestamps of the UHF signal and infrared data, and then sends it to the FPGA controller.

[0038] The UHF probe is electrically connected to the conditioning circuit and is used to receive UHF signals and amplify and filter the signals before sending them to the AD circuit.

[0039] The AD circuit is connected to the FPGA controller in parallel and is used to convert the analog signal from the UHF probe after passing through the conditioning circuit into a digital signal and send it to the FPGA controller.

[0040] The FPGA controller is connected to the USB port of the tablet computer and is used to receive digital signals from the UHF probe and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock.

[0041] The infrared module is connected to the RJ45 port of the tablet computer and is used to collect infrared temperature matrix signals and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock.

[0042] The tablet computer is connected to the USB port of the drone body to receive UHF signals and infrared temperature matrix signals, and transmits the UHF signals and infrared temperature matrix signals from the drone body to a remote cloud server via a 4G / 5G network.

[0043] The present invention also provides a computer-readable storage medium having stored thereon computer program instructions executable by a processor, wherein when the processor executes the computer program instructions, it can implement the steps of any of the methods described above.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] (1) Based on the field environmental data of meteorological stations along the power line, UHF correction formula and infrared temperature correction formula were constructed, and flight altitude control formula and flight speed control formula were constructed based on UHF signal strength, which improved the data sampling accuracy;

[0046] (2) The azimuth angle of the discharge source was located using the FDOA frequency difference positioning method, and the coordinates of the maximum temperature rise point were determined by fusing infrared temperature field data. An SVM support vector machine classifier was constructed to determine the type of partial discharge.

[0047] (3) A formula for the discharge risk hazard index was constructed and corresponding disposal suggestions were generated, which were sent to the maintenance personnel via their mobile terminals to complete the maintenance tasks. Attached Figure Description

[0048] Figure 1 This is a structural diagram of a UAV power line inspection system based on multimodal fusion according to the present invention;

[0049] Figure 2 This is a flowchart illustrating the workflow of a UAV power line inspection method based on multimodal fusion according to the present invention. Detailed Implementation

[0050] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1

[0052] like Figure 1 As shown, this example provides a UAV power line inspection system based on multimodal fusion, which includes a meteorological station along the power line, a quadcopter UAV, a remote cloud server, and a mobile terminal.

[0053] The meteorological station along the power line is connected to the remote cloud server via a 4G / 5G network, and is used to collect environmental meteorological information along the power line and send the information to the remote cloud server.

[0054] The quadcopter drone is connected to the remote cloud server via a 4G / 5G network. It is used to synchronously collect UHF signals and infrared temperature matrix signals and send them to the remote cloud server, receive flight control commands from the remote cloud server, and complete power line inspection tasks.

[0055] The remote cloud server is connected to the mobile terminal via a 4G / 5G network. It is used to receive environmental information along the power line sent by the meteorological station along the power line and UHF signals and infrared temperature matrix signals collected synchronously by the quadcopter drone. It uses a multimodal fusion method to determine the type and location of partial discharge, calculates the discharge risk hazard index according to the discharge risk hazard index formula, adds it to the power system maintenance operation plan, generates disposal suggestions, and sends them to the mobile terminal of the maintenance personnel to complete the maintenance task.

[0056] This system utilizes meteorological stations along power lines, quadcopter drones, remote cloud servers, and mobile terminals to implement a multimodal fusion-based drone power line inspection system. This system integrates environmental data, UHF data, and infrared temperature matrix data along power lines to detect partial discharge types and locations, calculate discharge hazard indices, and generate disposal recommendations, achieving integrated management of power line inspection and maintenance. Through design, simulation, and verification, a modular product is formed, enabling rapid portability between different platforms and accelerating the product development process.

[0057] The quadcopter drone includes a GPS-tamed rubidium atomic clock, a UHF probe, a conditioning circuit, an AD circuit, an FPGA controller, an infrared module, a tablet computer, and the drone body.

[0058] The GPS-disciplined rubidium atomic clock is connected to the I / O port of the FPGA controller and is used to generate a synchronization pulse signal to synchronize the timestamps of the UHF signal and infrared data and send them to the FPGA controller.

[0059] The UHF probe is electrically connected to the conditioning circuit and is used to receive UHF signals and amplify and filter the signals before sending them to the AD circuit.

[0060] The AD circuit is connected to the FPGA controller in parallel and is used to convert the analog signal from the UHF probe after passing through the conditioning circuit into a digital signal and send it to the FPGA controller.

[0061] The FPGA controller is connected to the USB port of the tablet computer and is used to receive digital signals from the UHF probe and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock.

[0062] The infrared module is connected to the RJ45 port of the tablet computer and is used to collect infrared temperature matrix signals and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock.

[0063] The tablet computer is connected to the USB port of the drone body and is used to receive UHF data and infrared temperature matrix data, and to send the UHF data and infrared temperature data from the drone body to the remote cloud server via 4G / 5G network.

[0064] The drone body is a DJI quadcopter drone, the infrared module is an X384H produced by Wuhan Gewu Youxin, and the FPGA controller hardware is a Cyclone IV EP4CE10F17C8, which was compiled using Quartus II_13.0 software.

[0065] In this embodiment, a power line meteorological station, a quadcopter drone, a remote cloud server, and a mobile terminal realize a drone power inspection system based on multimodal fusion. This system integrates environmental data along the power line, UHF data, and infrared temperature matrix data to detect the type and location of partial discharge, calculate the discharge hazard index, and generate disposal suggestions, thus realizing integrated management of power inspection and maintenance. Through design, simulation, and verification, a modular product is formed, which can be quickly ported between different platforms, accelerating the product development process.

[0066] Example 2

[0067] like Figure 2 As shown, this example provides a UAV power line inspection method based on multimodal fusion, which adopts a UAV power line inspection system based on multimodal fusion as described in Embodiment 1. The method includes the following steps:

[0068] S1, the quadcopter drone flies along the power transmission line in a zigzag trajectory with a lateral swing amplitude of ±2m. It adopts a wide-area scanning mode to simultaneously collect UHF signals and infrared thermal image temperature signals and send them to a remote cloud server.

[0069] S2. The remote cloud server acquires data from the nearest meteorological station along the power line to the drone, corrects the data using the UHF correction formula and the infrared temperature correction formula, performs multi-scale decomposition of the UHF signal using the Daubechies wavelet basis, extracts the energy proportion of the 1.5GHz band, and displays the UHF data and infrared thermal image.

[0070] S3. The remote cloud server calculates and adjusts the drone's flight altitude and speed based on the flight altitude control formula and flight speed control formula. When a pulse group of UHF pulses lasting ≥3 cycles and with a single cycle pulse number ≥50 is detected, it is suspected that there is partial discharge and controls the drone to reduce its altitude for detailed detection.

[0071] S4. The remote cloud server controls the drone to perform a spiral descent in the target area with an initial radius of 5m, a final radius of 0.5m, and a descent rate of 0.2m / s. Every 0.5m of descent, a set of 1.2 / 1.5 / 1.8GHz band UHF data and high-resolution infrared thermal images are collected.

[0072] S5. The remote cloud server calculates the azimuth angle of the discharge source using the FDOA frequency difference positioning method, integrates infrared temperature field data to establish a partial discharge defect volume model, and calculates the coordinates of the maximum temperature rise point.

[0073] S6. A four-dimensional feature vector is established on a remote cloud server, and the SVM support vector machine classifier rules are used to determine the type of partial discharge.

[0074] S7. The remote cloud server calculates the discharge risk hazard index based on the discharge type and discharge risk hazard index formula, and uses the hazard index as the priority of power system maintenance operations to add them to the power system maintenance operation plan.

[0075] S8. The remote cloud server sends information on the location, type, hazard index, and handling suggestions of partial discharge anomalies to the mobile terminals of maintenance personnel according to the power system maintenance operation plan, thereby completing the maintenance task.

[0076] The UHF correction formula is: , among which, U correct U is the corrected signal amplitude. raw The original detected signal amplitude is given in mV, T is the ambient temperature in °C, RH is the ambient humidity in %, P is the ambient atmospheric pressure in kPa, and α0, β0, γ0 and k T These are all correction coefficients, determined using the least squares method;

[0077] The infrared temperature correction formula is as follows: , among which, T correct For the corrected infrared temperature, T measThe value is the temperature detected by the infrared thermal imager, in °C; ɛ is the emissivity of the target surface, typically taken as 0.95; n is the atmospheric attenuation coefficient, typically taken as 1.41; L is the distance to the target, in meters; P is the ambient atmospheric pressure, in kPa; RH is the ambient humidity, in percent; and a, b, c, d, e, and f are correction coefficients determined using the least squares method.

[0078] The flight altitude control formula is as follows: , among which, S UHF UHF signal strength, in dBμV, h is flight altitude, in meters, and S UHF ∈[40dBμV,80dBμV], h∈[3m,7m];

[0079] The flight speed control formula is as follows: , among which, S UHF UHF signal strength, in dBμV, v is flight speed, in m / s. Here, if S... UHF A deceleration mechanism is triggered at 60dBμV to reduce speed to 0.5m / s.

[0080] The FDOA frequency difference positioning method includes using a mutual fuzzy function formula to calculate the two sets of antenna execution signals and using a target azimuth formula to calculate the target azimuth from the time delay difference and frequency shift difference corresponding to the extracted peak values.

[0081] The formula for the mutual fuzziness function is: , where t represents time, χ(τ, f) is the mutual ambiguity function, s1(t) and s2(t+τ) represent the signals received by the first group of antennas and the second group of antennas, respectively, τ represents the time delay of the signal received by the second group of antennas relative to the first group of antennas, and f represents the frequency shift of the signal received by the second group of antennas relative to the first group of antennas;

[0082] The formula for the target orientation is: Where θ is the target azimuth angle, c is the speed of light, f0 is the center frequency of the signal (where f0 = 1.5 GHz), d is the antenna spacing, and Δf is the frequency shift difference;

[0083] The partial discharge defect volume model integrates the azimuth coordinate system calculated by FDOA with the infrared temperature field coordinate system, and uses a rigid body transformation matrix to achieve data space alignment. The discharge area is divided into a three-dimensional voxel grid, with each voxel associated with a temperature value and azimuth weight. Based on the Bayesian criterion, the model fuses the FDOA positioning probability and the infrared thermal field probability to generate a three-dimensional probability density distribution cloud map of the discharge defect. The output is a comprehensive view model that includes azimuth annotation of the discharge point, thermal field contour overlay, and three-dimensional volume rendering.

[0084] The four-dimensional feature vector is: , where Q m This indicates the proportion of wavelet energy in the 1.5GHz band. This indicates the rate of temperature rise, in °C / s, f. peak Indicates the main frequency of the UHF signal, in MHz, N pulses Indicates the number of pulses per second;

[0085] The SVM support vector machine classifier rule is: when f peak ∈[300,800]MHz and Q m <0.5 is considered corona discharge, when f peak ∈[1200,1800]MHz and ≥5℃ / s is considered surface discharge, when N pulses >200 / s and Q m A value >0.7 indicates internal discharge.

[0086] The discharge risk hazard index formulas include: corona discharge hazard index formula, surface discharge hazard index formula, and internal discharge hazard index formula.

[0087] The formula for the corona discharge hazard index is: ,in, RH represents ambient humidity, K a The corona resistance correction factor is generally between 0.1 and 0.5; α is the main frequency weighting index, generally 1.2; and β is the high-frequency energy weighting index, generally 0.8.

[0088] The formula for the surface discharge hazard index is: ,in, RH represents ambient humidity, K b The surface insulation correction factor is typically between 0.2 and 0.6; α is the dominant frequency weighting index, typically 1.5; and γ is the temperature rise rate weighting index, typically 1.2.

[0089] The formula for the internal discharge hazard index is: ,in, RH represents ambient humidity, K c The equipment type correction factor is generally between 0.5 and 1.5, such as 0.8 for oil-immersed transformers and 1.2 for GIS equipment. λ is the pulse density weighting index, generally 1.5, and β is the high-frequency energy weighting index, generally 1.0.

[0090] The proposed treatments include corona discharge treatment recommendations, surface discharge treatment recommendations, and internal discharge treatment recommendations.

[0091] The corona discharge treatment recommendations include: H a ∈ When necessary, the system should be shut down for maintenance and repair, and deteriorated parts should be replaced; H a ∈ At times, clean the insulation surface to reduce ambient humidity; otherwise, strengthen inspections and monitor humidity and discharge frequency.

[0092] The recommended surface discharge treatment includes: H b ∈ Immediately stop operation for inspection and repair, and replace damaged parts; H b ∈ In some cases, local cleaning, repair, or application of anti-flashover coating is performed; for others, the frequency of inspections is increased, and humidity and temperature rise are monitored.

[0093] The internal discharge handling recommendations include: H c ∈ If the situation occurs, immediately shut down the system and replace the entire insulation system; H c ∈ When necessary, local repairs, vacuum drying, or replacement of seals are performed; otherwise, periodic monitoring of pulse density and humidity is conducted.

[0094] like Figure 2 As shown, in this embodiment of the invention, the quadcopter drone flies in a zigzag trajectory, simultaneously collecting UHF signals and infrared thermal imaging temperature signals. A remote cloud server obtains data from the nearest meteorological station along the power line to the drone, corrects the UHF and infrared temperatures, extracts the energy proportion of the 1.5GHz band, displays the UHF data and infrared thermal image, adjusts the flight altitude and speed, and controls the drone to reduce altitude for detailed detection when partial discharge is suspected. The azimuth angle of the discharge source is determined by the FDOA frequency difference positioning method, the coordinates of the maximum temperature rise point are determined by fusing infrared temperature field data, the type of partial discharge is determined by the SVM support vector machine classifier rules, the discharge risk hazard index is calculated using the discharge risk hazard index formula, and disposal suggestions are generated. Finally, the suggestions are sent to the mobile terminal of the maintenance personnel to complete the maintenance task.

[0095] This invention discloses a UAV power line inspection method based on multimodal fusion, implemented using standard JAVA language and a modular design approach. It utilizes meteorological stations along the power line, quadcopter UAVs, remote cloud servers, and mobile terminals to fuse environmental data, UHF data, and infrared thermal imaging data along the power line in a multimodal manner. This enables the detection of partial discharge types and locations, the calculation of discharge hazard indices, and the generation of disposal recommendations, thereby achieving integrated management of power line inspection and maintenance.

[0096] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for unmanned aerial vehicle (UAV) power line inspection based on multimodal fusion, characterized in that, Includes the following steps: S1. The quadcopter drone flies along the power transmission line in a zigzag trajectory with a predetermined lateral swing amplitude. It uses a wide-area scanning mode to simultaneously collect UHF signals and infrared temperature matrix signals and send them to a remote cloud server. S2. The remote cloud server acquires data from the nearest power line meteorological station to the quadcopter drone, corrects the data using UHF correction formula and infrared temperature correction formula, performs multi-scale decomposition of the UHF signal using Daubechies wavelet basis, extracts the energy proportion of the 1.5GHz band, and displays the UHF data and infrared thermal image. S3. The remote cloud server calculates and adjusts the flight altitude and speed of the quadcopter drone based on the flight altitude control formula and flight speed control formula. When a pulse group of UHF pulses lasting ≥3 cycles and with a single cycle pulse number ≥50 is detected, it is suspected that there is partial discharge and controls the quadcopter drone to reduce its altitude for detailed detection. S4. The remote cloud server controls the quadcopter drone to perform a spiral descent in the target area with an initial radius of R, a final radius of r, and a descent rate of v0. At each descent height H, a set of UHF data in the 1.2 / 1.5 / 1.8GHz frequency band and a high-resolution infrared thermal image are collected. S5. The remote cloud server calculates the azimuth angle of the discharge source using the FDOA frequency difference positioning method, and establishes a partial discharge defect volume model by fusing infrared temperature matrix signals, and calculates the coordinates of the maximum temperature rise point. S6. A four-dimensional feature vector is established on a remote cloud server, and the SVM support vector machine classifier rules are used to determine the type of partial discharge. S7. The remote cloud server calculates the discharge risk hazard index based on the partial discharge type and the discharge risk hazard index formula, and uses the discharge risk hazard index as the priority of power system maintenance operations to be added to the power system maintenance operation plan. S8. The remote cloud server sends information on the location, type, hazard index, and handling suggestions of partial discharge anomalies to the mobile terminals of maintenance personnel according to the power system maintenance operation plan, thereby completing the maintenance task.

2. The UAV power line inspection method based on multimodal fusion according to claim 1, characterized in that, In step S2, the UHF correction formula is: , among which, U correct U is the corrected signal amplitude. raw The original detection signal amplitude is given by T, ambient temperature is given by RH, ambient humidity is given by P, and α0, β0, γ0, and k are given by k. T These are all correction coefficients, determined using the least squares method; The infrared temperature correction formula is as follows: , among which, T correct For the corrected infrared temperature, T meas For the infrared module to detect temperature, ɛ is the emissivity of the target surface, n is the atmospheric attenuation coefficient, L is the distance to the target, and a, b, c, d, ... Both f and f are correction coefficients, determined using the least squares method.

3. The UAV power line inspection method based on multimodal fusion according to claim 1, characterized in that, In step S3, the flight altitude control formula is: , of which S UHF UHF signal strength, h is flight altitude, S UHF ∈[40dBμV,80dBμV], h∈[3m,7m]; The flight speed control formula is: Where v is the flight speed, if S UHF A deceleration mechanism was triggered at >60dBμV, reducing the quadcopter drone to 0.5m / s.

4. The UAV power line inspection method based on multimodal fusion according to claim 1, characterized in that, In step S5, the FDOA frequency difference positioning method includes calculating the two sets of antenna execution signals using the mutual ambiguity function formula and calculating the target azimuth using the target azimuth formula by extracting the time delay difference and frequency shift difference corresponding to the peak values; The formula for the mutual fuzziness function is: , where t represents time, χ(τ, f) is the mutual ambiguity function, s1(t) and s2(t+τ) represent the signals received by the first group of antennas and the second group of antennas, respectively, τ represents the time delay of the signal received by the second group of antennas relative to the first group of antennas, and f represents the frequency shift of the signal received by the second group of antennas relative to the first group of antennas; The formula for the target orientation is: Where θ is the target azimuth angle, c0 is the speed of light, f0 is the center frequency of the signal, d0 is the antenna spacing, and Δf is the frequency shift difference; The partial discharge defect volume model integrates the azimuth coordinate system calculated by the FDOA frequency difference positioning method with the infrared temperature field coordinate system. A rigid body transformation matrix is ​​used to align the data space, and the discharge area is divided into a three-dimensional voxel grid. Each voxel is associated with a temperature value and azimuth weight. Based on the Bayesian criterion, the FDOA positioning probability and the infrared thermal field probability are fused to generate a three-dimensional probability density distribution cloud map of the discharge defect. The output is a comprehensive view model that includes azimuth annotation of the discharge point, thermal field contour overlay, and three-dimensional volume rendering.

5. The UAV power line inspection method based on multimodal fusion according to claim 1, characterized in that, In step S6, the four-dimensional feature vector is: , where Q m This indicates the proportion of wavelet energy in the 1.5GHz band. f represents the rate of temperature rise. peak N represents the main frequency of the UHF signal. pulses Indicates the number of pulses per second; The SVM support vector machine classifier rule is: when f peak ∈[300,800]MHz and Q m <0.5 is considered corona discharge, when f peak ∈[1200,1800]MHz and ≥5℃ / s is considered surface discharge, when N pulses >200 / s and Q m A value >0.7 indicates internal discharge.

6. The UAV power line inspection method based on multimodal fusion according to claim 5, characterized in that, In step S7, the discharge risk hazard index formula includes: corona discharge hazard index formula, surface discharge hazard index formula, and internal discharge hazard index formula; The formula for the corona discharge hazard index is: ,in, RH represents ambient humidity, K a α is the corona resistance correction factor, β is the main frequency weighting index, and β is the high frequency energy weighting index. The formula for the surface discharge hazard index is: ,in, K b γ is the surface insulation correction factor, and γ is the temperature rise rate weighting index; The formula for the internal discharge hazard index is: ,in, K c λ is the equipment type correction factor, and λ is the pulse density weighting index.

7. The UAV power line inspection method based on multimodal fusion according to claim 6, characterized in that, In step S8, the treatment recommendations include corona discharge treatment recommendations, surface discharge treatment recommendations, and internal discharge treatment recommendations; The corona discharge treatment recommendations include: H a ∈ When necessary, the system should be shut down for maintenance and repair, and deteriorated parts should be replaced; H a ∈ When necessary, clean the insulation surface and reduce ambient humidity; otherwise, strengthen inspections and monitor humidity and discharge frequency. The recommended surface discharge treatment includes: H b ∈ Immediately stop operation for inspection and repair, and replace damaged parts; H b ∈ In some cases, local cleaning, repair, or application of anti-flashover coating should be performed; in other cases, the frequency of inspections should be increased, and humidity and temperature rise should be monitored. The recommended internal discharge handling includes: H c ∈ If the situation occurs, immediately shut down the system and replace the entire insulation system; H c ∈ When necessary, perform local repairs, vacuum drying, or replace the seals; otherwise, periodically monitor pulse density and humidity.

8. A UAV power line inspection system based on multimodal fusion that implements the method as described in any one of claims 1-7, characterized in that, This includes weather stations along power lines, quadcopter drones, remote cloud servers, and mobile terminals; among them, The meteorological station along the power line is connected to the remote cloud server via a 4G / 5G network, and is used to collect environmental meteorological information along the power line and send the information to the remote cloud server. The quadcopter drone is connected to the remote cloud server via a 4G / 5G network. It is used to synchronously collect UHF signals and infrared temperature matrix signals and send them to the remote cloud server, receive flight control commands from the remote cloud server, and complete power line inspection tasks. The remote cloud server is connected to the mobile terminal via a 4G / 5G network. It is used to receive environmental information along the power line sent by the meteorological station along the power line and UHF signals and infrared temperature matrix signals collected synchronously by the quadcopter drone. It uses a multimodal fusion method to determine the type and location of partial discharge, calculates the discharge risk hazard index according to the discharge risk hazard index formula, adds it to the power system maintenance operation plan, generates disposal suggestions, and sends them to the mobile terminal of the maintenance personnel to complete the maintenance task.

9. A UAV power line inspection system based on multimodal fusion according to claim 8, characterized in that, The quadcopter drone includes a GPS-docked rubidium atomic clock, a UHF probe, a conditioning circuit, an AD circuit, an FPGA controller, an infrared module, a tablet computer, and the drone body; The GPS-disciplined rubidium atomic clock is connected to the FPGA controller I / O port to generate a synchronization pulse signal to synchronize the timestamps of the UHF signal and infrared data, and then sends it to the FPGA controller. The UHF probe is electrically connected to the conditioning circuit and is used to receive UHF signals and amplify and filter the signals before sending them to the AD circuit. The AD circuit is connected to the FPGA controller in parallel and is used to convert the analog signal from the UHF probe after passing through the conditioning circuit into a digital signal and send it to the FPGA controller. The FPGA controller is connected to the USB port of the tablet computer and is used to receive digital signals from the UHF probe and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock. The infrared module is connected to the RJ45 port of the tablet computer and is used to collect infrared temperature matrix signals and synchronously send them to the tablet computer according to the pulse signals emitted by the GPS-disciplined rubidium atomic clock. The tablet computer is connected to the USB port of the drone body to receive UHF signals and infrared temperature matrix signals, and transmits the UHF signals and infrared temperature matrix signals from the drone body to a remote cloud server via a 4G / 5G network.

10. A computer-readable storage medium having stored thereon computer program instructions executable by a processor, wherein when the processor executes the computer program instructions, it is able to implement the steps of the method as described in any one of claims 1-7.

Images