Wind power plant unmanned aerial vehicle inspection data fault identification system based on artificial intelligence

Abstract

The invention relates to the technical field of fault identification, and provides a wind power plant unmanned aerial vehicle inspection data fault identification system based on artificial intelligence, which comprises an unmanned aerial vehicle, a server, a data acquisition module, an interaction module, an analysis module, a decision module and an inspection control module, the data acquisition module acquires appearance image data of a wind driven generator in a wind power plant, the interaction module interactively transmits the appearance image data acquired by the data acquisition module and ground data receiving equipment, and the inspection control module acquires the position of the wind driven generator and real-time position data of an unmanned aerial vehicle. The data acquisition module acquires the position of the wind driven generator, evaluates the inspection route of the unmanned aerial vehicle according to the position of the wind driven generator and the real-time position data of the unmanned aerial vehicle, adjusts the acquisition posture of the data acquisition module according to the evaluation result, and analyzes the wind driven generator according to the appearance image data acquired by the data acquisition module to form an analysis result. And the decision module evaluates the state decision of the wind driven generator according to the analysis result.

Description

Technical Field

[0001] This invention relates to the field of fault identification technology, and in particular to an artificial intelligence-based fault identification system for wind farm drone inspection data. Background Technology

[0002] With the continuous breakthroughs in my country's wind power technology, wind power generation is receiving more and more attention, and the scale of wind farms is gradually expanding. If the unit fails and shuts down, it will lead to a reduction in power generation and generate high maintenance costs. Therefore, it is necessary to conduct regular inspections of wind turbine units. Traditionally, the inspection of wind turbine blades is mostly done manually, such as using telescopes for observation and high-altitude inspection by workers. This method has the following disadvantages: (1) Most wind farms are widely distributed and have complex terrain, which makes traditional manual inspections very difficult. (2) Wind farm inspection tasks are heavy, traditional manual inspection is inefficient, time-consuming, and results in significant downtime losses. (3) Inspection personnel work at heights, which poses a high safety risk.

[0003] Meanwhile, existing inspection methods are not only time-consuming and labor-intensive, but also inefficient, failing to achieve large-scale, real-time fault monitoring. Furthermore, manual inspections struggle to cover all components of wind turbines, especially high-altitude and difficult-to-observe areas. Additionally, manual inspections are subject to subjectivity in data collection and analysis, potentially leading to misjudgments or omissions.

[0004] For example, Chinese patent CN110609229B discloses a deep learning-based method for detecting wind turbine blade imbalance faults. Although it proposes to use the collected data to build a model to identify faults in the wind farm, it lacks real-time identification of the wind farm, resulting in low identification efficiency and reduced identification accuracy, which is not conducive to high-precision identification scenarios.

[0005] Meanwhile, existing fault diagnosis methods rely on pre-set rules and experience, making it difficult to handle complex and nonlinear fault modes. Data analysis and processing are typically performed offline, failing to provide real-time feedback and rapid response.

[0006] In addition, the data collection and inspection angles cannot be scheduled, collected, and judged in real time, which is not conducive to the identification and evaluation of faults.

[0007] This invention was developed to address the common problems in the field, such as low efficiency of operation and maintenance inspections, poor real-time performance, limited inspection data, poor control over inspection locations, inability to provide real-time fault assessment, poor assessment capabilities, and poor inspection flexibility. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings of current systems by proposing an artificial intelligence-based fault identification system for wind farm drone inspection data.

[0009] To overcome the shortcomings of the prior art, the present invention adopts the following technical solution:

[0010] An AI-based wind farm drone inspection data fault identification system includes a drone and a server. The system further includes a data acquisition module, an interaction module, an analysis module, a decision-making module, and an inspection control module. The server is connected to the data acquisition module, the interaction module, the analysis module, the decision-making module, and the inspection control module.

[0011] The data acquisition module collects external image data of wind turbines in the wind farm. The interaction module transmits the external image data collected by the data acquisition module to the ground data receiving device. The inspection control module collects the location of the wind turbine and the real-time location data of the UAV, evaluates the inspection route of the UAV based on the location of the wind turbine and the real-time location data of the UAV, and adjusts the acquisition posture of the data acquisition module based on the evaluation results. The analysis module analyzes the wind turbine based on the external image data collected by the data acquisition module to form an analysis result. The decision module evaluates the status of the wind turbine based on the analysis results to determine the fault level of the wind turbine.

[0012] The inspection control module includes a positioning unit, an environmental perception unit, and an attitude adjustment unit. The positioning unit is used to locate the real-time position of the UAV. The environmental perception unit senses the inspection environment of the UAV to obtain the inspection position data of the wind turbine currently being inspected. The attitude adjustment unit adjusts the acquisition attitude of the data acquisition module according to the inspection position data and the real-time position of the UAV.

[0013] Optionally, the environmental perception unit includes a stereo vision sensor, a gyroscope, and a weather sensor. The stereo vision sensor measures the distance and relative position between the UAV and the wind turbine. The gyroscope is used to monitor the attitude of the UAV. The weather sensor collects data on wind speed, wind direction, temperature, and humidity in the environment where the UAV is located.

[0014] The environmental perception unit is mounted on the UAV.

[0015] Optionally, the data acquisition module includes a data acquisition unit and a storage unit. The data acquisition unit acquires the appearance of the inspected wind turbine to form appearance image data, and the storage unit stores the appearance image data acquired by the data acquisition unit.

[0016] The data acquisition unit includes an acquisition probe and a support base. The acquisition probe is used to acquire external image data of the wind turbine, and the support base is used to place the acquisition probe.

[0017] The data acquisition module is mounted on the inspection control module and moves with the drone to perform inspections of the wind turbines in the wind farm.

[0018] Optionally, the inspection control module further includes an evaluation unit, which acquires the location of the wind turbine and the real-time location data of the UAV, and determines the inspection route of the UAV according to the following steps:

[0019] S1. Establish a spatial coordinate system XYZ based on the real-time position of the UAV, and calculate the position vector D of the UAV relative to the wind turbine:

[0020] D = P drone -P turbine ;

[0021] In the formula, P drone The real-time position coordinates (x) of the UAV drone ,y drone, z drone ), P turbine The position coordinates of the wind turbine (x turbine ,y turbine ,z turbine );

[0022] S2. Update the drone's position vector:

[0023] D adjusted =D+f(V) wind C drone );

[0024] In the formula, f(V) wind C drone ) is an environmental impact function, the value of which is related to the wind speed and attitude of the environment in which the UAV is located;

[0025] S3. Calculate the new position coordinates of the UAV, Pnew(x). n y n , z n ):

[0026]

[0027] In the formula, D adjusted,x Let D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component in the X-axis direction, D adjusted,yLet D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component along the Y-axis, Δθ, represents the angular increment of the UAV's rotation around the wind turbine.

[0028] S4. Perform a rewinding inspection on the wind turbine currently being inspected, and during the rewinding process, adjust the data acquisition unit through the posture adjustment unit to acquire the appearance image data of the wind turbine.

[0029] S5. Replace other wind turbines in the wind farm and perform the operations in steps S1-S4.

[0030] Optionally, the interaction module includes an interaction unit and a transmission unit. The interaction unit is used for the transmission unit to interact and communicate with the ground data receiving device, so that the ground data receiving device sends an interaction request to the transmission unit. After receiving the interaction request, the transmission unit transmits the appearance image data of the wind turbine acquired by the data acquisition module to the ground data receiving device.

[0031] Optionally, the analysis module acquires the appearance image data collected by the data acquisition module and processes the appearance image data. The processing includes grayscale conversion and edge extraction to extract the histogram and fault area of ​​the wind turbine region, and calculates the state index Worn of the wind turbine according to the following formula:

[0032]

[0033] In the formula, C is the normalization constant corresponding to the historical maximum fault map of the wind turbine, and its value is determined based on historical data; E is the abnormal index of the wind turbine, and its value is calculated according to the following formula:

[0034]

[0035] In the formula, H image (i) is the histogram value of the current image at the i-th gray level, H reference (i) represents the histogram value of the normal state image at the i-th gray level;

[0036] A normalized The fault area index of the wind turbine is calculated according to the following formula:

[0037]

[0038] In the formula, A is the area of ​​the fault region. total It is the total area of ​​the image;

[0039] If the state index Worn of the wind turbine exceeds the lower limit M1 of the set monitoring threshold range, it indicates that the currently detected wind turbine is abnormal.

[0040] Optionally, the decision-making module includes a decision-making unit and an early warning unit. The decision-making unit determines the fault level of the wind turbine based on the evaluation results of the evaluation module, and the early warning unit triggers an early warning notification to the manager based on the decision-making unit's decision results.

[0041] Optionally, the early warning unit includes an early warning information generator and a communication transmitter. The early warning information generator obtains the fault level of the decision-making unit and triggers the communication transmitter to send an early warning to the manager about the analyzed fault level.

[0042] Optionally, the decision unit compares the state index Worn of the wind turbine with a set monitoring threshold range to determine the fault level of the wind turbine.

[0043] Optionally, the interaction unit verifies the identity information of both parties during interactive communication.

[0044] The beneficial effects achieved by this invention are:

[0045] 1. Through the coordination between the decision-making module and the inspection control module, the UAV can circle the wind turbine at the inspection location, improving the reliability of fault identification and ensuring that the entire system has the advantages of high inspection efficiency, high inspection flexibility, multiple inspection locations, high inspection capability and precise control capability.

[0046] 2. By coordinating the inspection control module and the data acquisition module, the drone's attitude can be adjusted according to the real-time environment during the inspection process, thereby improving the inspection efficiency and accuracy of the drone and ensuring that the entire system has the advantages of strong control capability and high inspection reliability.

[0047] 3. Through the interaction unit between the drones, pairing relationships can be established between the drones, improving the inspection efficiency of the entire system and ensuring the fault identification capability and accuracy of the entire system;

[0048] 4. Through the cooperation between the data acquisition module and the analysis module, the entire system can identify anomalies in wind turbines more efficiently, and ensure that the entire system has the advantages of high accuracy and flexibility in anomaly identification.

[0049] 5. Through the cooperation of the decision-making module and the analysis module, the abnormalities of the wind turbines in the wind farm can be predicted and warned, which changes the existing deficiency of not being able to analyze in real time, improves the automatic warning capability of the entire system, and ensures that the entire system has the advantages of high operation and maintenance inspection efficiency, strong real-time performance, reliable evaluation capability, and real-time provision of evaluation warnings and results. Attached Figure Description

[0050] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate the same parts.

[0051] Figure 1 This is a schematic diagram of the overall block shape of the present invention.

[0052] Figure 2 This is a block diagram of the inspection control module, analysis module, decision-making module, and manager of the present invention.

[0053] Figure 3 This is a schematic diagram illustrating the application scenarios of the wind turbine and drone of the present invention.

[0054] Figure 4 This is a schematic diagram of the structure of the drone, data acquisition module, and attitude adjustment unit of the present invention.

[0055] Figure 5 This is a front view schematic diagram of the drone of the present invention.

[0056] Figure 6 This is a top view of the drone of the present invention.

[0057] Figure 7 This is a schematic diagram of the posture adjustment unit and acquisition probe of the present invention.

[0058] Figure 8 for Figure 7 A schematic diagram at EE.

[0059] Explanation of reference numerals in the attached diagram: 1. Unmanned Aerial Vehicle (UAV); 2. Data Acquisition Module; 3. Attitude Adjustment Unit; 4. Interaction Unit; 5. Environmental Perception Unit; 6. Position Marker; 7. Adjustment Seat; 8. Horizontal Steering Seat; 9. Steering Drive Mechanism; 10. Limiting Rod; 11. Pitch Adjustment Plate; 12. Acquisition Probe; 13. Positioning Probe; 14. Positioning Marker; 15. Pitch Adjustment Drive Mechanism; 16. Limiting Track; 17. Position Recognition Probe. Detailed Implementation

[0060] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Furthermore, the accompanying drawings of the present invention are for simple illustrative purposes only and are not depictions of actual dimensions; this is stated beforehand. The following embodiments will further describe the relevant technical content of the present invention in detail, but the disclosed content is not intended to limit the scope of protection of the present invention.

[0061] Example 1: According to Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 This embodiment provides an artificial intelligence-based wind farm drone inspection data fault identification system. The wind farm drone inspection data fault identification system includes a drone and a server (not shown in the figure). The wind farm drone inspection data fault identification system also includes a data acquisition module, an interaction module, an analysis module, a decision-making module, and an inspection control module. The server is connected to the data acquisition module, interaction module, analysis module, decision-making module, and inspection control module respectively, and stores the intermediate data and process data of the data acquisition module, interaction module, analysis module, decision-making module, and inspection control module in the database of the server.

[0062] The data acquisition module collects external image data of wind turbines in the wind farm. The interaction module transmits the external image data collected by the data acquisition module to the ground data receiving device. The inspection control module collects the location of the wind turbine and the real-time location data of the UAV, evaluates the inspection route of the UAV based on the location of the wind turbine and the real-time location data of the UAV, and adjusts the acquisition posture of the data acquisition module based on the evaluation results. The analysis module analyzes the wind turbine based on the external image data collected by the data acquisition module to form an analysis result. The decision module evaluates the status of the wind turbine based on the analysis results to determine the fault level of the wind turbine.

[0063] The wind farm drone inspection data fault identification system also includes a central processing unit (CPU). The CPU is connected to the data acquisition module, interaction module, analysis module, decision-making module, and inspection control module. The CPU provides centralized control over these modules, and the control data of the CPU is stored in the server's database. This enables the entire system to efficiently and accurately identify wind farm faults.

[0064] In this embodiment, the data acquisition module, interaction module, analysis module, decision-making module, and inspection control module are all mounted on the UAV;

[0065] The inspection control module includes a positioning unit, an environmental perception unit, and an attitude adjustment unit. The positioning unit is used to locate the real-time position of the UAV. The environmental perception unit senses the inspection environment of the UAV to obtain the inspection position data of the wind turbine being inspected. The attitude adjustment unit adjusts the acquisition attitude of the data acquisition module according to the inspection position data and the real-time position of the UAV, so that the UAV can acquire the appearance image data of the wind turbine from multiple angles when flying around the wind turbine.

[0066] Optionally, the environmental perception unit includes a stereo vision sensor, a gyroscope, and a weather sensor. The stereo vision sensor measures the distance and relative position between the UAV and the wind turbine. The gyroscope is used to monitor the attitude of the UAV. The weather sensor collects data on wind speed, wind direction, temperature, and humidity in the environment where the UAV is located.

[0067] The environmental perception unit is mounted on the UAV.

[0068] Meanwhile, the weather sensor is used to collect humidity and wind data at the location of the drone;

[0069] The positioning unit includes a locator and a positioning data storage device. The locator collects real-time positioning data of the UAV, and the positioning data storage device stores the real-time positioning data collected by the locator.

[0070] In this embodiment, the positioning data refers to the three-dimensional coordinate data of the UAV;

[0071] The attitude adjustment unit includes an adjustment seat, a horizontal steering component, and a pitch adjustment component. The horizontal steering component adjusts the horizontal acquisition attitude of the data acquisition module so that the data acquisition module can follow the horizontal rotation of the UAV as it rotates around the wind turbine, thereby realizing the inspection of the wind turbine and collecting image data of the external damage of the wind turbine during the inspection process. The pitch adjustment component adjusts the pitch acquisition angle of the data acquisition module so that the data acquisition module can collect external image data of multiple pitch angles at the same inspection point.

[0072] The pitch adjustment component is mounted on the horizontal steering component and performs horizontal steering during the horizontal steering process of the horizontal steering component.

[0073] The adjustment seat is mounted on the UAV;

[0074] The horizontal steering component includes a horizontal steering seat, a steering drive mechanism, a position recognition probe, and at least three position markers for marking. The horizontal steering seat has a passage hole and is nested around the outer periphery of the adjusting seat and hinged to the adjusting seat. The steering drive mechanism is driven to the horizontal steering seat so that the horizontal steering seat rotates along the axis of the hinged position. At least three position markers are disposed on the horizontal steering seat and are evenly distributed along the axis of the horizontal steering seat. The position recognition probe is used to identify the position markers to obtain the horizontal steering angle of the horizontal steering seat.

[0075] In this embodiment, a closed loop is formed between the central processing unit, the position recognition probe 17, and at least one position marker 6, so that the central processing unit can compare the position data collected by the position recognition probe 17 with the required position data, and control the steering drive mechanism 9 through the central processing unit, thereby achieving precise control of the pitch adjustment component and the horizontal steering of the data acquisition module 2.

[0076] The pitch adjustment component includes a pitch adjustment drive mechanism 15, a pitch adjustment plate 11, an adjustment gear, and a limiting rod 10. One end of the limiting rod 10 is connected to the pitch adjustment plate 11, and the other end of the limiting rod 10 extends away from the pitch adjustment plate 11. The adjustment gear is disposed on one end face of the pitch adjustment plate 11. The pitch adjustment drive mechanism 15 is disposed on the horizontal steering seat 8 and meshes with the adjustment gear so that the pitch adjustment plate rotates along the axis of the adjustment gear.

[0077] The adjusting gear is located on the same side as the limiting rod 10;

[0078] In addition, the horizontal steering seat 8 is provided with a limiting rail 16, and the end of the limiting rod 10 away from the pitch adjustment plate 11 is slidably connected to the limiting rail 16, and the position of the pitch adjustment plate is slidably limited during the rotation of the pitch adjustment plate 11. Figure 8 To Figure 7 A half-section view of the components below the mid-position marker without a data acquisition probe. The dashed lines in the figure represent... Figure 7 The view line at the corresponding position, i.e. Figure 7 The limiting rails, limiting rods, and other components are all from... Figure 8 It was observed at the dotted line.

[0079] Meanwhile, in this embodiment, the data acquisition module 2 is mounted on the pitch adjustment plate 11 and rotates along with the pitch adjustment plate.

[0080] In addition, the pitch adjustment component also includes a positioning probe 13 and at least one positioning mark 14. The at least one positioning mark 14 is set along the rotation path of the pitch adjustment plate and is set on the end face of the pitch adjustment plate facing the horizontal steering seat 8. The positioning probe 13 is set on the horizontal steering seat 8 and extends towards one side of the pitch adjustment plate to identify the at least one positioning mark 14 and obtain the angle adjustment amount of the pitch adjustment plate.

[0081] In this embodiment, a closed loop is formed between the central processing unit, the positioning probe 13, and at least one positioning device, so that the central processing unit can compare the position data collected by the positioning probe 13 with the required position data, and control the steering drive mechanism 9 through the central processing unit, thereby achieving precise control of the pitch angle of the data acquisition module 2.

[0082] Optionally, the data acquisition module 2 includes a data acquisition unit and a storage unit. The data acquisition unit acquires the appearance of the inspected wind turbine to form appearance image data, and the storage unit stores the appearance image data of the wind turbine acquired by the data acquisition unit.

[0083] The data acquisition unit includes an acquisition probe 12 and a support base. The acquisition probe 12 is used to acquire external image data of the wind turbine, and the support base is used to place the acquisition probe 12.

[0084] The data acquisition module 2 is mounted on the inspection control module and moves with the drone 1 to perform inspections of the wind turbines in the wind farm.

[0085] The storage unit includes a portable storage disk, which is used to store the appearance image data, and can also store the flight data and process data of the UAV 1.

[0086] Optionally, the inspection control module further includes an evaluation unit, which acquires the location of the wind turbine and the real-time location data of the UAV, and determines the inspection route of the UAV according to the following steps:

[0087] S1. Establish a spatial coordinate system XYZ based on the real-time position of the UAV, and calculate the position vector D of the UAV relative to the wind turbine:

[0088] D = P drone -P turbine ;

[0089] In the formula, P drone The real-time position coordinates (x) of the UAV drone y drone , z drone ), P turbine The position coordinates of the wind turbine (x turbine y turbine , z turbine In this embodiment, the so-called position coordinates are based on the established spatial coordinate system, therefore, the following exists:

[0090] D=(x drone -x turbine y drone -y turbine , z drone -z turbine );

[0091] Here, vector D represents the specific position of the drone relative to the wind turbine in three-dimensional space. It can be used to calculate distance and direction, and for further flight planning or data analysis. This means that the position vector encompasses the horizontal and vertical coordinates, as well as the altitude coordinate in three-dimensional space applications.

[0092] S2. Update the drone's position vector:

[0093] D adjusted =D+f(V) wind C drone );

[0094] In the formula, f(V) wind C drone ) is the environmental impact function, whose value is related to the wind speed and attitude of the UAV in its environment. The environmental impact function f(V) wind C droneThe output is a vector;

[0095] Wherein, the environmental impact function f(V) wind C drone Calculate according to the following formula:

[0096] f(V wind C drone ) = K drag ·R(C drone )·V wind ;

[0097] In the formula, V wind Wind speed, i.e., the wind speed of the surrounding environment, K drog Let R(C) be a drag coefficient vector representing the intensity of the wind's influence on the drone. Its value is determined by the drone's aerodynamic design and is equivalent to a known value (determined by the inherent parameters of the selected drone, such as its wind resistance rating). The wind resistance rating can be determined through dynamic wind load testing and functional testing. Dynamic wind load testing involves placing the drone in a wind tunnel with adjustable speed to simulate flight conditions under different wind speeds, observing the drone's flight stability, handling performance, and the completion of flight maneuvers such as climb, descent, and turn. Functional testing tests the drone's various functions to ensure they function correctly in environments simulating different wind speeds, such as the camera, GPS positioning system, and remote control signal reception. drone () is a rotation matrix that takes into account the directional adjustment of the UAV's attitude to the influence of wind force;

[0098] The specific calculation method depends on the rotation order and the definition of the rotation angle. In this example, assuming the ZYX Euler angle order is used (yaw first, then pitch, and finally roll), then:

[0099] R(C drone ) = R z (Ψ)·R y (φ)·R x (θ);

[0100] In the formula, R x Ry(φ) is the rotation matrix of the UAV rotating around the X-axis by an angle θ, Rz(ψ) is the rotation matrix of the UAV rotating around the Y-axis by an angle θ, and Rz(ψ) is the rotation matrix of the UAV rotating around the Z-axis by an angle θ. These are all basic rotation matrices around the corresponding axes and can be obtained from the standard rotation matrix formula.

[0101] Additionally, the wind speed vector V wind Anemometers can be measured directly on the drone or obtained from meteorological data. Drone attitude C drone Obtained through the UAV's internal inertial measurement unit (IM∪).

[0102] S3. Calculate the new position coordinates of the UAV, Pnew(x). n y n , z n ):

[0103]

[0104] In the formula, D adjusted，x Let D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component in the X-axis direction, D adjusted，y Let D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component in the Y-axis direction, (x t y t , z d Let X, Y, and Z be the coordinates of the wind turbine in three-dimensional space. Δθ represents the angular increment of the UAV's rotation around the wind turbine, calculated using the following formula:

[0105]

[0106] In the formula, v is the flight speed of the UAV, R is the inspection radius, Δt is the time required for the UAV to collect one data point, and Error is the environmental impact index, which is calculated according to the following formula:

[0107] Error = V wind / V base ;

[0108] In the formula, V wind V represents the wind speed in the environment where the drone is located. base The reference wind speed can be either the safe wind speed or the average wind speed determined during drone flight testing.

[0109] When V wind equals V base When Error = 1, it indicates that environmental conditions have no additional impact on the drone's flight performance. When V wind Greater than V base If the error value is greater than 1, it indicates that wind speed has a negative impact on the drone's flight performance; conversely, if the error value is less than 1, it may indicate that the wind speed conditions are relatively ideal.

[0110] S4. Perform a rewinding inspection on the wind turbine currently being inspected, and during the rewinding process, adjust the data acquisition unit through the posture adjustment unit to acquire the appearance image data of the wind turbine.

[0111] S5. Replace other wind turbines in the wind farm and perform the operations in steps S1-S4.

[0112] In this embodiment, it is assumed that the wind farm is equipped with at least two wind turbines, and the above steps are used to inspect at least two turbines.

[0113] In this embodiment, the drag coefficient vector K drog This can be determined according to the following steps:

[0114] 1) For drones, the drag coefficient C D (one with K) drog The relevant parameters can be approximated using the following equations:

[0115]

[0116] In the formula, where F D ρ is the drag force, ρ is the air density, v is the speed of the drone relative to the air, and A is the frontal area of ​​the drone.

[0117] 2) Determine the parameters:

[0118] Driving force F D The value is preset by the staff and can be obtained from historical experimental data, such as through wind tunnel experiments or computational fluid dynamics (CFD) simulations.

[0119] Air density ρ: Approximately 1.225 kg / m³ under standard atmospheric conditions. 3 .

[0120] Speed ​​v: The flight speed of the drone.

[0121] Frontal area A: can be obtained from the design parameters of the drone.

[0122] 3) Calculate K drag Once C is obtained D K can be obtained through appropriate transformation. drag It depends on how K is defined. drag And its role in applications.

[0123] In this embodiment, K drag It is considered as a normalized representation of a drag coefficient related to the application scenario;

[0124] By coordinating the inspection control module and the data acquisition module, the drone's posture can be adjusted according to the real-time environment during the inspection process, thereby improving the inspection efficiency and accuracy of the drone and ensuring that the entire system has the advantages of strong control capability and high inspection reliability.

[0125] Optionally, the interaction module includes an interaction unit and a transmission unit. The interaction unit is used for the transmission unit to interact and communicate with the ground data receiving device, so that the ground data receiving device sends an interaction request to the transmission unit. After receiving the interaction request, the transmission unit transmits the appearance image data of the wind turbine acquired by the data acquisition module to the ground data receiving device.

[0126] The interaction unit includes an interactor and a pairer. The pairer is used to pair the UAV with the currently inspected UAV. Once a pairing is established, interactive communication is performed through the interactor.

[0127] The transmission unit includes a transmitter and a feedback unit. The transmitter transmits the data collected by the data acquisition module to the ground data receiving device. The feedback unit sends a feedback signal to the transmitter after the ground data receiving device has finished receiving the data, so that the transmitter terminates the transmission operation.

[0128] After the transmission unit transmits the inspection-obtained appearance image data to the ground data receiving device, it disconnects from the UAV, allowing the UAV to proceed to the location of the next wind turbine for inspection.

[0129] In this embodiment, each drone is independently paired with and transmits data to the transmission unit, and the pairing with the drone is released after the appearance image data transmission is completed.

[0130] Optionally, the interaction unit verifies the identity information of both parties during interactive communication. The identity verification process includes obtaining the other party's device identification code and establishing a pairing relationship.

[0131] Obtaining the device identification code of the other party and establishing a pairing relationship are techniques well-known to those skilled in the art. These techniques can be found in relevant technical manuals, and therefore will not be elaborated upon in this embodiment. Optionally, the analysis module acquires the appearance image data collected by the data acquisition module and processes the appearance image data. The processing includes grayscale conversion and edge extraction to extract the histogram and fault area of ​​the wind turbine region, and calculates the state index Worn of the wind turbine according to the following formula:

[0132]

[0133] In the formula, C is the normalization constant of the wind turbine image data under extreme conditions, that is, the normalization constant corresponding to the historical maximum fault map of the wind turbine, and its value is determined based on historical experimental data. E is the abnormal index of the wind turbine, and its value is calculated according to the following formula:

[0134]

[0135] In the formula, H image (i) is the histogram value of the current image at the i-th gray level. Its value can be directly obtained through image processing techniques, which will not be elaborated upon in this embodiment. H reference (i) is the histogram value of the normal state image at the i-th gray level. Its value can be directly obtained through image processing technology. In this embodiment, it will not be described in detail.

[0136] A normalized The fault area index of the wind turbine is calculated according to the following formula:

[0137]

[0138] In the formula, A is the area of ​​the fault region of the wind turbine obtained after image processing. total It is the total area of ​​the entire captured image;

[0139] If the state index Worn of the wind turbine exceeds the lower limit M1 of the set monitoring threshold range, it indicates that the currently detected wind turbine is abnormal.

[0140] If the state index Worn of the wind turbine is lower than the abnormal lower limit M1 of the set monitoring threshold range, it indicates that the wind turbine is in a normal state and no abnormality has occurred.

[0141] The abnormal lower limit M1 of the set monitoring threshold range is set by the system or administrator according to the actual situation. This is a technical means well known to those skilled in the art. Those skilled in the art can consult relevant technical manuals to learn about this technology, so it will not be described in detail in this embodiment.

[0142] By coordinating the data acquisition module and the analysis module, the entire system can identify anomalies in wind turbines more efficiently, and ensure that the entire system has the advantages of high accuracy and flexibility in anomaly identification.

[0143] Interaction is achieved through the interaction units between the drones, enabling them to establish pairing relationships, thereby improving the inspection efficiency of the entire system and ensuring the fault identification capability and accuracy of the entire system.

[0144] The normalization constant C for wind turbine image data under extreme conditions is determined based on historical experimental data. In this example, a step is provided for calculating the normalization constant C:

[0145] S11. Identify extreme cases

[0146] That is, a series of wear index Worn results were obtained through image analysis, and the maximum value Worn among these results was identified. max ;

[0147] S12. Consider possible extreme scenarios in the future.

[0148] To ensure that C remains effective even in more extreme future scenarios, we can use the known maximum value Worn. max Add a safety boundary ratio (safety) to the existing structure; for example, C can be defined as Worn. max 110% or 120% to ensure that the calculated Worn can be reasonably normalized even in the event of more severe wear.

[0149] S13, Calculation formula:

[0150] Based on the above considerations, the formula for calculating C can be:

[0151] C = Worn max •(1+safety);

[0152] For example, if the safety boundary ratio is 20% (i.e., 0.20), then:

[0153] C = Worn max ·1.20;

[0154] The safety boundary ratio is determined according to the following steps:

[0155] 1. Collect historical data:

[0156] Collect a series of image data of wind turbines and their corresponding wear index Worn;

[0157] Determine the maximum wear index Worn from these data. max ;

[0158] 2. Statistical Analysis:

[0159] Calculate the mean μ and standard deviation σ of the wear index;

[0160] Statistical methods can be used to estimate extreme wear conditions. For example, it can be assumed that extreme cases fall within the range of the mean plus three times the standard deviation (i.e., μ + 3σ).

[0161] 3. Determine the safety boundary ratio:

[0162] The safety margin ratio can be determined based on the difference between the extreme case estimate and the known maximum value.

[0163] Calculation formula:

[0164]

[0165] For example: Suppose historical data analysis shows that:

[0166] Worn max =0.8

[0167] The average wear index μ = 0.5

[0168] The standard deviation of the wear index σ = 0.1

[0169] Then: Safety boundary ratio

[0170] In this scenario, if managers believe that historical data already covers all possible extreme cases, they may not need to add a safety margin. However, if they wish to consider extreme cases beyond the existing data, they can set a positive value based on professional judgment, such as 10% (0.10).

[0171] Optionally, the decision-making module includes a decision-making unit and an early warning unit. The decision-making unit determines the fault level of the wind turbine based on the evaluation results of the evaluation module, and the early warning unit triggers an early warning notification to the manager based on the decision-making unit's decision results.

[0172] Optionally, the early warning unit includes an early warning information generator and a communication transmitter. The early warning information obtains the fault level of the decision-making unit and triggers the communication transmitter to send an early warning to the manager about the analyzed fault level.

[0173] Optionally, the decision unit compares the state index Worn of the wind turbine with a set monitoring threshold range to determine the fault level of the wind turbine. Specifically,

[0174] If Worn < M1, then the wind turbine is normal;

[0175] If M1 < Worn < M2, then the wind turbine has a minor fault.

[0176] If M1 < Worn < M2, then the wind turbine is in a moderate fault condition.

[0177] If Worn ≥ M2, then the wind turbine is in a severe fault.

[0178] In this embodiment, the upper and lower limits (M1, M2, M3 and M4) of the monitoring threshold range are set by the system or the administrator according to the actual situation of the wind turbine. This is a technical means well known to those skilled in the art. Those skilled in the art can consult relevant technical manuals to learn about this technology, so it will not be described in detail in this embodiment.

[0179] Through the cooperation of the decision-making module and the analysis module, the abnormalities of wind turbines in the wind farm can be predicted and warned, which changes the existing deficiency of not being able to analyze in real time, improves the automatic early warning capability of the entire system, and ensures that the entire system has the advantages of high operation and maintenance inspection efficiency, strong real-time performance, reliable evaluation capability, and real-time provision of evaluation warnings and results.

[0180] Example 2: This example should be understood as including all the features of any of the foregoing examples, and further improving upon them, according to... Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 , Figure 8 Furthermore, the wind farm drone inspection data fault identification system also includes an inspection monitoring module, which acquires the real-time altitude data of the drone and the inspection altitude data set by the manager, and calculates the pitch acquisition angle of the data acquisition module.

[0181] The inspection and monitoring module acquires the real-time altitude data of the drone and the inspection altitude data set by the manager, and calculates the pitch angle θ of the data acquisition module at each inspection point according to the following formula. pitch :

[0182]

[0183] In the formula, H turbine It is the height of the wind turbine, H drone This is the current flight altitude of the drone, C. wind D is the wind direction influence coefficient. horizontal The horizontal distance from the drone to the wind turbine base is determined by the following formula:

[0184]

[0185] In the formula, (lat drone ,lon drone The coordinates () represent the real-time location of the drone, which are its GPS coordinates.

[0186] (lat turbine ,lonturbine ) represents the position coordinates of the wind turbine, which are the GPS coordinates of the wind turbine base, where △lat=lat turbine -lat drone ,△lon=lon turbine -lon drone , where r is the average radius of the Earth;

[0187] For the wind direction influence coefficient C wind Calculate according to the following formula:

[0188]

[0189] In the formula, V wind For wind speed, α wind This is the wind direction angle, representing the angle between the wind direction and the drone's direction of travel. Its value is directly measured by wind speed and direction sensors mounted on the drone itself. V drone The flight speed of the drone is determined based on the actual situation of the drone. k is the adjustment weight, which is used to adjust the weight of the wind force influence according to the actual situation. The value is taken according to the following conditions.

[0190] Specifically, when considering the need to improve the impact of wind, k should be taken as a larger value.

[0191] If the drone is less sensitive to wind, or flies under relatively calm conditions, k can take a smaller value;

[0192] In this embodiment, an example of the value of the adjustment weight k is also provided:

[0193] Assuming that flight tests of the drone in light wind conditions show that for every 1 m / s increase in wind speed, the drone's yaw angle increases by 0.5°, a suitable value for k can be estimated using the following steps:

[0194] 1. Determine the benchmark for wind force influence: Select a benchmark wind speed V base For example, 5 m / s is the wind speed at which the drone is expected to fly comfortably;

[0195] 2. Observe the impact during actual flight: If the actual wind speed V wind At 10 m / s, the flight performance of the UAV (e.g., stability or path deviation) is reduced by 10% compared to the reference wind speed;

[0196] 3. Calculate the k value: This reduction can be achieved through C. wind To adjust;

[0197] in,

[0198] If Cwind It directly affects the percentage decrease in performance; you can find k by solving this equation in reverse.

[0199] This calculation example simplifies the practical process; in reality, determining k requires more experimental data and detailed analysis. For example, k = -0.02 means that for every 1 m / s increase in wind speed, the actual reduction in drone performance relative to the baseline wind speed is k times the wind speed difference. This value can be adjusted according to the specific performance of the drone and flight conditions.

[0200] In this embodiment, those skilled in the art can calculate the accurate value of k based on the actual situation and by analogy with the above example, so it will not be described in detail in this example.

[0201] The inspection and monitoring module transmits the calculated pitch angle to the adjustment module, so that the adjustment module can make pitch adjustments at the inspection points and collect appearance image data at multiple pitch angles.

[0202] Through the cooperation of the inspection monitoring module and the inspection control module, the data acquisition module can acquire data at a more precise angle, increasing the amount of data collected at each inspection point. This ensures the inspection accuracy of the entire system for each wind turbine in the wind farm, giving the system the advantages of strong inspection angle control, abundant data acquisition, strong evaluation capabilities, and high inspection flexibility.

[0203] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A wind farm drone inspection data fault identification system based on artificial intelligence, the wind farm drone inspection data fault identification system comprising a drone and a server, characterized in that, The wind farm drone inspection data fault identification system also includes a data acquisition module, an interaction module, an analysis module, a decision-making module, and an inspection control module. The server is connected to the data acquisition module, the interaction module, the analysis module, the decision-making module, and the inspection control module respectively. The data acquisition module collects external image data of wind turbines in the wind farm. The interaction module transmits the external image data collected by the data acquisition module to the ground data receiving device. The inspection control module collects the location of the wind turbine and the real-time location data of the UAV, evaluates the inspection route of the UAV based on the location of the wind turbine and the real-time location data of the UAV, and adjusts the acquisition posture of the data acquisition module based on the evaluation results. The analysis module analyzes the wind turbine based on the external image data collected by the data acquisition module to form an analysis result. The decision module evaluates the status of the wind turbine based on the analysis results to determine the fault level of the wind turbine. The inspection control module includes a positioning unit, an environmental perception unit, and an attitude adjustment unit. The positioning unit is used to locate the real-time position of the UAV. The environmental perception unit senses the inspection environment of the UAV to obtain the inspection position data of the wind turbine currently being inspected. The attitude adjustment unit adjusts the acquisition attitude of the data acquisition module according to the inspection position data and the real-time position of the UAV.

2. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 1, characterized in that, The environmental perception unit includes a stereo vision sensor, a gyroscope, and a weather sensor. The stereo vision sensor measures the distance and relative position between the UAV and the wind turbine. The gyroscope is used to monitor the attitude of the UAV. The weather sensor collects data on wind speed, wind direction, temperature, and humidity in the environment where the UAV is located. The environmental perception unit is mounted on the UAV.

3. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 2, characterized in that, The data acquisition module includes a data acquisition unit and a storage unit. The data acquisition unit acquires the appearance of the inspected wind turbine to form appearance image data, and the storage unit stores the appearance image data acquired by the data acquisition unit. The data acquisition unit includes an acquisition probe and a support base. The acquisition probe is used to acquire external image data of the wind turbine, and the support base is used to place the acquisition probe. The data acquisition module is mounted on the inspection control module and moves with the drone to perform inspections of the wind turbines in the wind farm.

4. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 3, characterized in that, The inspection control module also includes an evaluation unit, which acquires the location of the wind turbine and the real-time location data of the UAV, and determines the inspection route of the UAV according to the following steps: S1. Establish a spatial coordinate system XYZ based on the real-time position of the UAV, and calculate the position vector D of the UAV relative to the wind turbine: D=P drone -P turbine ; In the formula, P drone The real-time position coordinates (x) of the UAV drone ,y drone ,z drone ), P turbine The position coordinates of the wind turbine (x turbine ,y turbine ,z turbine ); S2. Update the drone's position vector: D adjusted =D+f(V wind ,C drone ); In the formula, f(V) wind C drone ) is an environmental impact function, the value of which is related to the wind speed and attitude of the environment in which the UAV is located; S3. Calculate the new position coordinates of the UAV, Pnew(x). n y n , z n ): In the formula, D adjusted,x Let D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component in the X-axis direction, D adjusted,y Let D be the adjusted position vector of the UAV relative to the wind turbine. adjusted The component along the Y-axis, Δθ, represents the angular increment of the UAV's rotation around the wind turbine. S4. Perform a rewinding inspection on the wind turbine currently being inspected, and during the rewinding process, adjust the data acquisition unit through the posture adjustment unit to acquire the appearance image data of the wind turbine. S5. Replace other wind turbines in the wind farm and perform the operations in steps S1-S4.

5. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 4, characterized in that, The interaction module includes an interaction unit and a transmission unit. The interaction unit is used for the transmission unit to communicate with the ground data receiving device, so that the ground data receiving device sends an interaction request to the transmission unit. After receiving the interaction request, the transmission unit transmits the appearance image data of the wind turbine acquired by the data acquisition module to the ground data receiving device.

6. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 5, characterized in that, The analysis module acquires the appearance image data collected by the data acquisition module and processes the appearance image data. The processing includes grayscale conversion and edge extraction to extract the histogram and fault area of ​​the wind turbine region, and calculates the state index Worn of the wind turbine according to the following formula: In the formula, C is the normalization constant corresponding to the historical maximum fault map of the wind turbine, and its value is determined based on historical data; E is the abnormal index of the wind turbine, and its value is calculated according to the following formula: In the formula, H image (i) is the histogram value of the current image at the i-th gray level, H reference (i) represents the histogram value of the normal state image at the i-th gray level; A normalized The fault area index of the wind turbine is calculated according to the following formula: In the formula, A is the area of ​​the fault region. total It is the total area of ​​the image; If the state index Worn of the wind turbine exceeds the lower limit M1 of the set monitoring threshold range, it indicates that the currently detected wind turbine is abnormal.

7. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 6, characterized in that, The decision-making module includes a decision-making unit and an early warning unit. The decision-making unit determines the fault level of the wind turbine based on the evaluation results of the evaluation module, and the early warning unit triggers an early warning notification to the manager based on the decision-making unit's decision results.

8. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 7, characterized in that, The early warning unit includes an early warning information generator and a communication transmitter. The early warning information generator obtains the fault level of the decision-making unit and triggers the communication transmitter to send an early warning to the manager about the analyzed fault level.

9. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 8, characterized in that, The decision-making unit compares the state index Worn of the wind turbine with a set monitoring threshold range to determine the fault level of the wind turbine.

10. The artificial intelligence-based wind farm drone inspection data fault identification system according to claim 9, characterized in that, When conducting interactive communication, the interaction unit verifies the identity information of both parties.

Images