Wind turbine blade UAV inspection method and device

Abstract

The present invention discloses a method and device for drone inspection of wind turbine blades, belonging to the technical field of wind turbine generators. The method includes: performing a drone inspection on the front of the wind turbine blade to generate first inspection information, while simultaneously performing a drone inspection on the back of the wind turbine blade to generate second inspection information, wherein the first inspection information includes a first track point data set and a set of blade front imagery, and the second inspection information includes a second track point data set and a set of blade back imagery; obtaining the results of the drone inspection of the wind turbine blades based on the first and second inspection information, combined with wind turbine blade defect judgment conditions; and determining whether to activate a drone re-inspection mode based on the wind turbine blade drone inspection results. If so, the drone re-inspection mode is activated to update the wind turbine blade drone inspection results; if not, the re-inspection mode is not activated. The present invention can reduce the impact on wind farm operation and power generation efficiency, and improve the accuracy of wind turbine blade drone inspection results.

Description

Technical Field

[0001] The present invention relates to the technical field of wind turbines, and more specifically, to a method and device for unmanned aerial vehicle inspection of wind turbine blades. Background Art

[0002] Wind power generation is a key energy sector in the development of new energy and a crucial strategic direction for promoting energy structure adjustment and carbon neutrality. The blades of wind turbines play a critical role in wind power generation. Defective or damaged blades can directly impact the entire turbine component, potentially disrupting normal operation. This places higher demands on blade inspections.

[0003] Drone inspection technology has begun to be used in wind turbine blade inspections. Autonomous drone inspections typically utilize rotorcraft with stable hovering capabilities. The process typically involves flying the drone to the vicinity of the wind turbine blades along a planned trajectory while the wind turbine is shut down. Based on its own positioning and the blade's position, the drone follows a trajectory consisting of multiple track points to capture a complete surface image of the blade at a specified location. Compared to traditional manual inspections of wind turbine blades, drone inspections have significantly improved the efficiency of wind turbine blade inspections.

[0004] However, due to the high wind speeds in wind farms, drones need to inspect wind turbine blades at close range, which places high demands on the drone's flight stability, control reliability, and safety. This increases the cost of the drone's hardware and software algorithms, leading to a decrease in the drone's endurance performance. Insufficient endurance often forces the wind turbine inspection to stop midway, forcing the drone to return to its takeoff position, which greatly affects the efficiency of drone inspections of wind turbine blades. Moreover, existing drone inspection technologies for wind turbine blades must be performed while the wind turbine is shut down, and wind turbine shutdown inspections can affect the wind farm's operating and power generation efficiency. Therefore, it is of great significance to research and develop drone inspection technologies for wind turbine blades to monitor the operating status of wind turbine blades, collect defect data on wind turbine blades, determine the defect type of wind turbine blades, improve the efficiency and accuracy of drone inspections, reduce the impact on wind farm operation and power generation efficiency, and realize automated drone inspections of wind turbine blades in wind farms. Summary of the Invention

[0005] Based on this, the present invention provides a method and device for UAV inspection of wind turbine blades, aiming to partially or completely solve the above technical problems. The present invention is achieved through the following technical solutions:

[0006] In a first aspect, the present invention provides a method for inspecting wind turbine blades using a drone, comprising the following steps:

[0007] Step S100, obtaining structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, wherein the structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the altitude and wind speed;

[0008] Step S200: Based on structural parameters, motion parameters, and environmental parameters of at least one wind turbine in a wind farm, a drone inspection is performed on the front of a wind turbine blade to generate first inspection information, and simultaneously a drone inspection is performed on the back of the wind turbine blade to generate second inspection information, wherein the first inspection information includes a first track point data set and a blade front image set, and the second inspection information includes a second track point data set and a blade back image set; the blade front image set includes a plurality of overall front images of the wind turbine blade; and the blade back image set includes a plurality of overall back images of the wind turbine blade.

[0009] Step S300: Obtain the drone inspection result of the wind turbine blade based on the first inspection information and the second inspection information, combined with the wind turbine blade defect judgment conditions; determine whether to start the drone re-inspection mode based on the wind turbine blade drone inspection result. If yes, start the drone re-inspection mode and update the wind turbine blade drone inspection result; if not, do not start the re-inspection mode.

[0010] Optionally, starting the drone re-inspection mode includes the following steps:

[0011] Step S401: obtaining a sequence of defect track points corresponding to defective wind turbine blades according to the inspection results of the wind turbine blade drone;

[0012] Step S402: acquiring defect track points corresponding to the defective wind turbine blades according to the defect track point image sequence, wherein the defect track points form a first re-inspection track point set and / or a second re-inspection track point set;

[0013] Step S403: planning a drone re-inspection route according to the first re-inspection track point set and / or the second re-inspection track point set, and starting the drone re-inspection of the wind turbine blades.

[0014] Optionally, updating the wind turbine blade drone inspection results includes the following steps:

[0015] Step S501: Performing a UAV re-inspection of a wind turbine blade according to the UAV re-inspection route to obtain re-inspection information of the wind turbine blade, the re-inspection information including a blade front re-inspection image set and a blade back re-inspection image set; the blade front re-inspection image set includes a plurality of wind turbine blade front re-inspection images; and the blade front re-inspection image set includes a plurality of wind turbine blade back re-inspection images.

[0016] Step S502: Obtain the wind turbine blade drone re-inspection result according to the re-inspection information, and update the wind turbine blade drone inspection result according to the wind turbine blade drone re-inspection result and the wind turbine blade drone inspection result.

[0017] Optionally, step S502 includes:

[0018] Step S5021: determine whether the wind turbine blade drone re-inspection result is consistent with the wind turbine blade drone inspection result. If so, update the wind turbine blade drone re-inspection result as the wind turbine blade drone inspection result. If not, proceed to step S5022.

[0019] Step S5022: at least based on the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the corresponding multiple back re-inspection images of the wind turbine blades, use a first classifier, a second classifier, and a third classifier to respectively classify the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the multiple back re-inspection images of the wind turbine blades, and obtain results of a first defect type set C1 and a first probability set P1 of the first classifier, a second defect type set C2 and a second probability set P2 of the second classifier, and a third defect type set C3 and a third probability set P3 of the third classifier;

[0020] Step S5023: Based on the first defect type set C1, the second defect type set C2 and the third defect type set C3, the first probability set P1, the second probability set P2 and the third probability set P3, calculate the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, evaluate the results of the defect types of the wind turbine blades based on the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, and update the results of evaluating the defect types of the wind turbine blades as the wind turbine blade drone inspection results.

[0021] Optionally, the defect track point includes a first track point and a second track point, the UAV re-inspection route includes a first re-inspection route, a re-inspection switching route and a second re-inspection route, the re-inspection switching route connects the first re-inspection route and the second re-inspection route, the first re-inspection track point set is used to form a first re-inspection route, the first re-inspection route is used to perform a front UAV inspection of the wind turbine blade at the defect first track point to form a first re-inspection information, the second re-inspection track point set is used to form a second re-inspection route, the second re-inspection route is used to perform a back UAV inspection of the wind turbine blade at the defect second track point to form a second re-inspection information; obtain the defect first track point 1 at the start of the first re-inspection route and the defect first track point P at the end, The second re-inspection route starts with the defective second track point 1 and ends with the defective second track point Q, and the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, and the fourth distance D4 between the defective first track point P and the defective second track point Q are calculated; based on the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, the minimum distance Dmin among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4 is obtained, and the drone re-inspection route is planned according to the minimum distance Dmin.

[0022] In a second aspect, the present invention provides a wind turbine blade drone inspection device, which implements any wind turbine blade drone inspection method described in the first aspect, including a parameter acquisition unit, an inspection unit, a display unit, a re-inspection unit, and a server:

[0023] a parameter acquisition unit, which acquires structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm and sends them to the inspection unit, wherein the structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the altitude and wind speed;

[0024] The inspection unit performs a drone inspection on the front of a wind turbine blade based on structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, generates first inspection information, and transmits it to a server. Simultaneously, it performs a drone inspection on the back of the wind turbine blade, generates second inspection information, and transmits it to the server. The first inspection information includes a first track point data set and a blade front image set, and the second inspection information includes a second track point data set and a blade back image set. The blade front image set includes a plurality of overall front images of the wind turbine blades, and the blade back image set includes a plurality of overall back images of the wind turbine blades.

[0025] The server obtains the wind turbine blade drone inspection result based on the first inspection information and the second inspection information and the wind turbine blade defect judgment condition, and sends the result to the re-inspection unit and the display unit, and the display unit displays the wind turbine blade drone inspection result;

[0026] The re-inspection unit provides an interactive interface for starting the drone re-inspection mode based on the drone inspection results of the wind turbine blades through the display unit; if yes, the drone re-inspection mode is started and the drone inspection results of the wind turbine blades are updated; if not, the re-inspection mode is not started.

[0027] Optionally, the re-inspection unit includes a defect image search unit, a defect track point forming unit, a re-inspection route planning unit and a starting unit. The defect image search unit obtains a defect track point image sequence corresponding to the defect type of the wind turbine blade based on the UAV inspection result of the wind turbine blade; the defect track point forming unit obtains the defect track points corresponding to the defect type of the wind turbine blade based on the defect track point image sequence, and forms the defect track points into a first re-inspection track point set and / or a second re-inspection track point set; the re-inspection route planning unit plans the UAV re-inspection route based on the first re-inspection track point set and / or the second re-inspection track point set; the starting unit starts the UAV re-inspection of the wind turbine blade.

[0028] Optionally, the UAV re-inspection route includes a first re-inspection route, a re-inspection switching route T and a second re-inspection route. The re-inspection switching route connects the first re-inspection route and the second re-inspection route. The first re-inspection track point set is used to form the first re-inspection route. The first re-inspection route is used to perform a frontal UAV inspection on the defect type of the wind turbine blade to form a first re-inspection information. The second re-inspection track point set is used to form a second re-inspection route. The second re-inspection route is used to perform a back UAV inspection on the defect type of the wind turbine blade to form a second re-inspection information.

[0029] Optionally, the re-inspection route planning unit includes a distance calculation sub-unit, which obtains the defective first track point 1 at the start and the defective first track point P at the end of the first re-inspection route, the defective second track point 1 at the start and the defective second track point Q at the end of the second re-inspection route, calculates the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, and the fourth distance D4 between the defective first track point P and the defective second track point Q, obtains the minimum distance Dmin = min[D1, D2, D3, D4] among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, and sends the minimum distance Dmin to the re-inspection route planning unit.

[0030] Optionally, the re-inspection unit also includes a re-inspection information acquisition unit, a first evaluation unit and a second evaluation unit. The re-inspection information acquisition unit re-inspects the wind turbine blades by drone according to the drone re-inspection route to obtain re-inspection information of the wind turbine blades; the first evaluation unit obtains the wind turbine blade drone re-inspection results based on the re-inspection information; the second evaluation unit updates the wind turbine blade drone inspection results based on the wind turbine blade drone re-inspection results and the wind turbine blade drone inspection results.

[0031] In the third aspect, the present invention application provides a wind turbine blade drone inspection device, which includes a processor, a memory, and a computing program stored on the memory and executable by the processor, wherein when the computing program is executed by the processor, the steps of the wind turbine blade drone inspection method as described in any one of the first aspects are implemented.

[0032] In a fourth aspect, the present invention application provides a computer-readable storage medium having a computing program stored thereon, wherein when the computing program is executed by a processor, the steps of the wind turbine blade drone inspection method as described in any one of the first aspects are implemented.

[0033] In summary, the beneficial technical effects of the present invention are:

[0034] (1) In the present invention, firstly, the UAV inspection technology for wind turbine blades does not require the wind turbine to be shut down. When the wind turbine is running, the UAV inspection work of the wind turbine blades can be carried out synchronously, and the working conditions of multiple wind turbine blades of at least one wind turbine in the wind farm can be photographed, and whether multiple wind turbine blades have defects can be judged, which will not affect the operation and power generation efficiency of the wind farm. In addition, the first track point 1, the first track point 2, ..., the first track point M, the second track point 1, the second track point 2, ..., the second track point M can be planned and set in advance. As long as a UAV is able to take pictures of the front or back of the blade, even if the endurance performance of the UAV is reduced or insufficient, the spare UAV can take off in advance or immediately take off to replace the existing UAV with poor endurance performance, and continue to take pictures of the front and back of the wind turbine blades at the first track point 1, the first track point 2, ..., the first track point M, the second track point 1, the second track point 2, ..., the second track point M, thereby improving the efficiency of the UAV inspection of wind turbine blades.

[0035] (2) In the present invention, the re-inspection route of the UAV is planned according to the minimum distance Dmin. With the help of the UAV's advanced navigation and positioning technology, the UAV can accurately fly according to the planned re-inspection route and obtain high-resolution images and data. By setting appropriate shooting angles and segmented shooting images, the integrity and consistency of the data are ensured, which facilitates comparative analysis and timely detection of subtle changes. The planned re-inspection route enables the UAV to cover the target area in the shortest time, greatly improving the inspection efficiency. It does not require a large amount of manpower to conduct long-term inspections on site, reducing labor costs and labor intensity, and reducing equipment loss. Reasonable route planning can reduce unnecessary flights and operations of the UAV, extend its service life, and reduce equipment maintenance and replacement costs. According to the planned re-inspection route, the consistency and comparability of the data are guaranteed, which helps to obtain more accurate and comprehensive data, improve the accuracy and reliability of data analysis, and provide strong support for the decision-making of subsequent wind turbine blade defect detection results.

[0036] (3) In the present invention, the conflict coefficient CT, the first classifier, the second classifier, the third classifier, the confidence expectation and the true voting expectation are set. When CT≥0.8, the confidence expectation and the voting expectation can be comprehensively considered to determine the defect type. By introducing a comparison mechanism of product and average expectation, when the product of the confidence expectation and the true voting expectation of different defect types is still less than the average expectation, the defect types corresponding to these expectations are used as the results of evaluating the defect types of wind turbine blades. In this way, reliable defect types can be screened out more rigorously, the classification errors caused by the conflict of classifiers can be reduced, and the evaluation of the defect types of wind turbine blades can be made more accurate and stable. It also helps to improve the accuracy of wind turbine blade defect identification under complex classification conditions, reduce the misleading of subsequent maintenance decisions due to misjudgment, reduce unnecessary maintenance costs and resource waste caused by the classifier's incorrect judgment of the defect type, ensure the safe operation of wind turbine blades, and ensure that the wind farm does not stop, thereby improving the operating efficiency of the entire wind farm. When CT<0.8, the true voting expectation is focused. When the true voting expectations of different defect types are still greater than the average true voting expectation, the defect types corresponding to these true voting expectations are used as the results of evaluating the defect types of wind turbine blades. In this way, the defect type of the wind turbine blade can be quickly identified, the complex multi-factor calculation process is avoided, and the efficiency and timeliness of the wind turbine blade defect assessment are improved. This is especially important for the inspection of large-scale wind turbine blades. A large amount of data can be processed in a short time and reliable results can be obtained. At the same time, when the classifier consistency is high, the advantages of multiple classifiers can be fully utilized to reduce the misclassification caused by individual factors or accidental errors, avoid unnecessary repairs or miss the best repair time due to misjudgment, reduce maintenance costs and wind turbine blade downtime, and improve the availability and power generation efficiency of wind power equipment. Moreover, with the accumulation of data and the improvement of the deep learning performance of the classifier, the accuracy and reliability are expected to be further improved, which can promote the development of intelligent drone inspections in the wind power industry towards a more refined and precise direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of the wind turbine blade drone inspection method applied by the present invention;

[0038] Figure 2 This is a schematic diagram of the principle of the wind turbine blade drone inspection method applied by the present invention. Figure 1 ;

[0039] Figure 3 is a flow chart of step S400 of the present invention;

[0040] Figure 4 This is a schematic diagram of the principle of the wind turbine blade drone inspection method applied by the present invention. Figure 2 ;

[0041] Figure 5 This is a schematic diagram of the principle of the wind turbine blade drone inspection method applied by the present invention. Figure 3 ;

[0042] Figure 6 This is a schematic diagram of a wind turbine blade drone inspection device applied for by the present invention;

[0043] Figure 7 This is a schematic diagram of the composition of the multiple inspection unit of the present invention;

[0044] Description of the drawings: first blade - 101, second blade - 102, third blade - 103, first unmanned unit - 201, second drone unit one - 2021, second drone unit two - 2022, third flight unit - 203, parameter acquisition unit - 100, inspection unit - 200, display unit - 300, re-inspection unit - 400, server - 500, defect image search unit - 401, defect track point formation unit - 402, re-inspection route planning unit - 403, start-up unit - 404, re-inspection information acquisition unit - 405, first evaluation unit - 406, second evaluation unit - 407, judgment sub-unit - 4071, processing sub-unit - 4072, evaluation update sub-unit - 4073. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure more clearly understood, the embodiments of the present disclosure are further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present disclosure and are not intended to limit the embodiments of the present disclosure.

[0046] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are intended only to facilitate understanding and reading by persons skilled in the art. They are not intended to limit the conditions under which the present invention can be implemented, and therefore have no substantial technical significance. Any structural modifications, changes in proportions, or adjustments in sizes, without affecting the efficacy and objectives of the present invention, should still fall within the scope of the technical contents disclosed in the present invention. Furthermore, terms such as "and" and "or" used in this specification are intended only for clarity of description and are not intended to limit the scope of implementation. Changes or adjustments in their relative relationships, without substantially changing the technical contents, should also be considered within the scope of implementation of the present invention. Furthermore, the various embodiments of the present invention are not independent of each other, but can be combined.

[0047] The terms "first", "second" and "third" are used for descriptive purposes only and should not be understood as indicating or suggesting relative importance or implicitly indicating the number of the indicated technical features; thus, the features defined as "first", "second" and "third" may explicitly or implicitly include one or more of such features. In the description of the present invention application, unless otherwise specified, "plurality" means two or more.

[0048] In the existing field of wind power generation technology, a horizontal-axis wind turbine refers to a wind turbine whose blades' rotation axis is parallel to the ground and perpendicular to the turbine blades. A horizontal-axis wind turbine typically includes an impeller, a nacelle, and a tower. The impeller portion is primarily composed of three blades, the nacelle is typically a horizontal nacelle containing key operating components of the horizontal-axis wind turbine (such as the impeller), and the tower typically includes a tower column perpendicular to the ground. Due to the simple structure and large blade rotation range of horizontal-axis wind turbines, the ease of trajectory planning in drone inspection technology when the horizontal-axis wind turbine is in a stopped state, and the high power generation efficiency, horizontal-axis wind turbines have been widely used in most wind farms for power generation. The operating state of horizontal-axis wind turbines also greatly affects the operation and maintenance efficiency and power generation revenue of most wind farms. These are technical knowledge that ordinary technicians in the existing field of wind power generation technology possess.

[0049] First, as Figure 1-Figure 5 As shown, the present invention provides a method for inspecting wind turbine blades using a drone, comprising the following steps:

[0050] Step S100, obtaining structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, wherein the structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the altitude and wind speed;

[0051] In some embodiments, the structural parameters, motion parameters, and environmental parameters of the wind turbine can be obtained in advance or by measurement. The structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the terrain and wind speed. For example, a drone can be used to take multiple photos of at least one wind turbine to obtain multiple images of the at least one wind turbine. The multiple images can be used to obtain the structural parameters and altitude of the wind turbine. A wind speed sensor can also be used to measure the wind speed around the wind turbine, thereby measuring the structural parameters, motion parameters, and environmental parameters of the wind turbine.

[0052] For example, in the application of the present invention, the first drone unit 201 can be used to measure and obtain some or all structural parameters and some environmental parameters (altitude) of the wind turbine. The wind turbine includes three wind turbine blades, namely the first wind turbine blade 101, the second wind turbine blade 102 and the third wind turbine blade 103, the impeller center height is 75m, the blade length is 44m, the maximum chord length is 3.7m, the impeller diameter is about 90m, the wind turbine speed is about 25r / min, the altitude is about 1500 meters, the wind speed is about 4m / s, etc.

[0053] Step S200: Based on the structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, a drone inspection is performed on the front of the wind turbine blade to form first inspection information, and simultaneously a drone inspection is performed on the back of the wind turbine blade to form second inspection information, wherein the first inspection information includes a first track point data set and a blade front image set, and the second inspection information includes a second track point data set and a blade back image set; the blade front image set includes a plurality of complete front images of the entire wind turbine blade; and the blade back image set includes a plurality of complete back images of the entire wind turbine blade.

[0054] In some embodiments, the plurality of wind turbine blades are the three aforementioned wind turbine blades. The wind turbine blades of a wind turbine generator may include: a front surface, a back surface, a blade root, and a blade tip. The front surface can be understood as at least the windward side or the leeward side, and correspondingly, the back surface can be understood as at least the leeward side or the windward side. Persons skilled in the art may reasonably understand this based on actual circumstances. When a horizontal-axis wind turbine generator is operating clockwise, the front and back surfaces of the blades are generally considered to be areas of the wind turbine blade that are most susceptible to environmental factors such as wind, sand, lightning, and rain. Accordingly, it is considered necessary to acquire images of these areas of the wind turbine blade.

[0055] In some embodiments, a drone inspection of the front face of a wind turbine blade generates first inspection information, the first inspection information including a first track point dataset and a first blade front image dataset. The drone inspection typically includes a drone and a camera, the camera being mounted on the drone. The drone can take off based on the structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, and the camera is positioned above the blade. The first track point dataset includes a first track point 1, a first track point 2, ..., and a first track point M, where M is a positive integer greater than or equal to 3. The drone hovers over the first track point 1, the first track point 2, ..., and the first track point M, and the cameras each take images downwardly facing the blade, thereby acquiring blade front image 1 of first track point 1, blade front image 2 of first track point 2, ..., and blade front image M of first track point M. It should be noted that, since the wind turbine blades are long and the camera shooting range is limited, the blade front image 1 of the first track point 1, the blade front image 2 of the first track point 2, ..., the blade front image M of the first track point M are segmented images obtained by photographing the front of the wind turbine blade at each first track point. The number of segments of the segmented image can correspond to the total number of first track points, that is, the blade front image 1, the blade front image 2, ..., the blade front image M can be combined to form the overall front image of the wind turbine blade. Accordingly, the blade front image 1 of the first track point 1, the blade front image 2 of the first track point 2, ..., the blade front image M of the first track point M can be processed, and then a blade front image set of the wind turbine blade can be obtained by combining them. The blade front image set includes a complete plurality of overall front images of the wind turbine blades.

[0056] For example, Figure 2 As shown, in the application of the present invention, the first track point 1, the first track point 2, ..., the first track point M can be planned and set in advance, as long as the drone can appear at the first track point 1, the first track point 2, ..., the first track point M to take pictures of the front of the wind turbine blade; at the same time, the second drone unit 2021 can be used to perform drone inspections on the front of the first wind turbine blade 101, the front of the second wind turbine blade 102, and the front of the third wind turbine blade 103 to form first inspection information, and the second drone unit 2021 hovers over the first track point 1P11, the first track point 2P12, ..., the first track point iP1i, ..., the first track point MP1M, i = 1, 2, 3, ..., M.

[0057] Obtain the blade front image 1 of the first track point 1P11, the blade front image 2 of the first track point 2P12, ..., the blade front image M of the first track point MP1M, the blade front image 1 includes the first wind turbine blade 101 front image 1, the second wind turbine blade 102 front image 1, and the third wind turbine blade 103 front image 1, the blade front image 2 includes the first wind turbine blade 101 front image 2, the second wind turbine blade 102 front image 2, and the third wind turbine blade 103 front image 2, ..., the blade front image M includes the first wind turbine blade 101 front image M, the second wind turbine blade 102 front image M, and the third wind turbine blade 103 front image M. On this basis, blade front image 1, blade front image 2, ..., blade front image M can be combined to form an overall front image of the wind turbine blade. The blade front image set will include a complete overall front image of the first wind turbine blade 101, a complete overall front image of the second wind turbine blade 102, and a complete overall front image of the third wind turbine blade 103, that is, the blade front image set can include multiple complete overall front images of wind turbine blades.

[0058] In some embodiments, a drone inspection of the back of a wind turbine blade generates second inspection information, the second inspection information including a second track point dataset and a second blade back image dataset. Similarly, a drone inspection typically includes a drone and a camera, the camera being mounted on the drone. The drone can take off based on the structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, and the camera is positioned above the blade. The second track point dataset includes second track point 1, second track point 2, ..., and second track point N, where N is a positive integer greater than or equal to 3. The drone hovers over second track point 1, second track point 2, ..., and second track point N, and the cameras each take photos downwardly facing the blade, thereby acquiring blade back image 1 of second track point 1, blade back image 2 of second track point 2, ..., and blade back image N of second track point N. It should also be noted that, because the wind turbine blades are long and the camera shooting range is limited, the blade back image 1 of the second track point 1, the blade back image 2 of the second track point 2, ..., the blade back image N of the second track point N are segmented images obtained by photographing the back of the wind turbine blade at each second track point. The number of segments of the segmented images can correspond to the total number of second track points, that is, the blade back image 1, the blade back image 2, ..., the blade back image N can be combined to form the overall back image of the wind turbine blade. Accordingly, the blade back image 1 of the second track point 1, the blade back image 2 of the second track point 2, ..., the blade back image N of the second track point M can be image processed, and then they can be combined to form a blade back image set of the wind turbine blade. The blade back image set includes a complete plurality of overall back images of the wind turbine blades.

[0059] For example, Figure 2 As shown, in the application of the present invention, the second track point 1, the second track point 2, ..., the second track point N can be planned and set in advance, as long as the second track point 1, the second track point 2, ..., the second track point N can appear for the drone to take pictures of the back of the wind turbine blade; at the same time, the second drone unit 2022 can be used to perform drone inspections on the back of the first wind turbine blade 101, the back of the second wind turbine blade 102, and the back of the third wind turbine blade 103 to form second inspection information, and the second drone unit 2022 hovers at the second track point 1P21, the second track point 2P22, ..., the second track point jP2j, ..., the second track point NP2N, j = 1, 2, 3, ..., N.

[0060] Obtain the blade back image 1 of the second track point 1P21, the blade back image 2P22 of the second track point 2, ..., the second track point jP2j, ..., and the blade back image N of the second track point NP2N. The blade back image 1 includes the back image 1 of the first wind turbine blade 101, the back image 1 of the second wind turbine blade 102, and the back image 1 of the third wind turbine blade 103. The blade back image 2 includes the back image 2 of the second wind turbine blade 101, the back image 2 of the second wind turbine blade 102, and the back image 2 of the third wind turbine blade 103, ..., and the blade back image N includes the back image N of the second wind turbine blade 101, the back image N of the second wind turbine blade 102, and the back image N of the third wind turbine blade 103. On this basis, blade back image 1, blade back image 2, ..., blade back image N can be combined to form an overall back image of the wind turbine blade. The blade back image set will include a complete overall back image of the first wind turbine blade 101, a complete overall back image of the second wind turbine blade 102, and a complete overall back image of the third wind turbine blade 103, that is, the blade back image set includes multiple complete overall back images of wind turbine blades.

[0061] In the present application, the camera is used to take pictures with the direction of shooting downward toward the blade. The foreground in the front image or the back image is usually the ground or the sea surface, so the front image or the back image of the blade can be easily extracted. At the same time, generally speaking, the camera starts taking pictures only when the first wind turbine blade 101, the second wind turbine blade 102, and the third wind turbine blade 103 are basically level with the ground and enter the shooting range of the camera. The camera can rotate or not rotate to ensure a sufficient shooting range, so that the spines of the wind turbine blades in the front image or the back image of the first wind turbine blade 101, the second wind turbine blade 102, and the third wind turbine blade 103 occupy more pixels, and the front image or the back image of the first wind turbine blade 101, the second wind turbine blade 102, and the third wind turbine blade 103 have a high degree of similarity and a low difficulty in image stitching, which facilitates the subsequent combination to form a blade front image set or a blade back image set, and then the complete overall front image of the first wind turbine blade 101, the second wind turbine blade 102, and the third wind turbine blade 103 and the overall back image of the first wind turbine blade 101, the second wind turbine blade 102, and the third wind turbine blade 103 can be obtained.

[0062] Step S300: Obtain the drone inspection result of the wind turbine blade based on the first inspection information and the second inspection information, combined with the wind turbine blade defect judgment conditions; determine whether to start the drone re-inspection mode based on the wind turbine blade drone inspection result. If yes, start the drone re-inspection mode and update the wind turbine blade drone inspection result; if not, do not start the re-inspection mode.

[0063] In some embodiments, the defect types of wind turbine blades generally include: (1) fatigue defects (which can be named C-1). Due to fatigue, wind turbine blades have tiny cracks, long and thin cracks, etc. The cracks or cracks are narrow in width and short in length and can be directly observed on the surface of the wind turbine blades; 2) impact defects (which can be named C-2). When the wind turbine blades are subjected to external impact or collision, such as lightning strikes, bird collisions, etc., burns or open cracks occur on the surface of the wind turbine blades, and ruptures, cracks and fractures occur; 3) erosion defects (which can be named C-3). When the surface of the wind turbine blades is eroded by factors such as chemicals, wind and sand, seawater or air pollution, erosion damage will occur, such as sand holes (i.e., a continuous area of the blade has relatively dense and discrete small holes); 4) shedding defects (which can be named C-4). Due to unreliable bonding of the wind turbine blades, the internal fiber cloth of the wind turbine blades is exposed, and surface shedding occurs.

[0064] Based on the above four common types of wind turbine blade defects, after learning and training, multiple samples of wind turbine blade defects can usually form a classifier for detecting wind turbine blade defect types. Usually, the image of the wind turbine blade can be input into the classifier to detect and classify the defects of the photographed wind turbine blade (for example, neural network technology, deep learning technology, etc. can be used). That is, combined with the wind turbine blade defect judgment conditions, the classifier is used to obtain the wind turbine blade drone inspection results. These are all common technical knowledge of existing wind turbine blades. Therefore, the wind turbine blade defect judgment conditions applied by the present invention include: judging whether the wind turbine blade has fatigue defects, impact defects, erosion defects and shedding defects, that is, judging whether the wind turbine blade has one or more of fatigue defects, impact defects, erosion defects and shedding defects based on the blade front image set and the blade back image set, that is, judging whether the wind turbine blade has defects and the type of wind turbine blade defects. More specifically, based on the overall front image or the overall back image of the first wind turbine blade, the overall front image or the overall back image of the second wind turbine blade, and the overall front image or the overall back image of the third wind turbine blade, it can be judged whether the first wind turbine blade, the second wind turbine blade, and the third wind turbine blade have defects and the corresponding defect types, which are one or more of fatigue defects, impact defects, erosion defects, and shedding defects.

[0065] In some embodiments, obtaining a drone inspection result for a wind turbine blade includes: a result indicating that the wind turbine blade has defects and the type of defect, or a result indicating that the wind turbine blade has no defects and the type of defect is none. Apparently, upon detecting that the wind turbine blade has defects and the type of defect is none, it is determined that a drone re-inspection mode can be initiated to conduct another drone inspection of the wind turbine blade, thereby improving the accuracy and reliability of the wind turbine blade defect detection results. Upon detecting that the wind turbine blade has no defects and the type of defect is none, it is determined that the drone re-inspection mode can be deactivated.

[0066] Therefore, in the application of the present invention, firstly, the wind turbine blade drone inspection technology does not require the wind turbine to be shut down. When the wind turbine is running, the wind turbine blade drone inspection work can be carried out synchronously to photograph the working conditions of the wind turbine blades of at least one wind turbine in the wind farm, and judge whether there are defects in the wind turbine blades, which will not affect the power generation efficiency of the wind farm; in addition, the first track point 1, the first track point 2, ..., the first track point M, the second track point 1, the second track point 2, ..., the second track point M can be planned and set in advance, and as long as the drone can inspect the blades, It only needs to take pictures of the front or back of the wind turbine blade. Even if the endurance performance of the drone decreases or is insufficient, the backup drone can take off in advance or quickly take off to replace the existing drone with poor endurance performance, and continue to take pictures of the front and back of the wind turbine blade at the first track point 1, the first track point 2, ..., the first track point M, the second track point 1, the second track point 2, ..., the second track point M. This does not affect the drone inspection efficiency of the wind turbine blade at all. Performing drone inspection on the wind turbine blade again can further improve the accuracy and reliability of the wind turbine blade defect detection results.

[0067] Optionally, the first inspection information includes a first track point image sequence one, a first track point image sequence two, and a first track point image sequence three;

[0068] The first track point image sequence 1 at least includes: a first track point front image sequence SNPF11, a first track point front image sequence SNPF12; ...; a first track point front image sequence SNPF1 i; ...; a first track point front image sequence SNPF1 M; the first track point front image sequence SNPF1 i includes: [first track point i P1 i, first wind turbine blade 101 front image i];

[0069] The first track point image sequence 2 at least includes: a first track point front image sequence SNPF21, a first track point front image sequence SNPF22; ...; a first track point front image sequence SNPF2 i; ...; a first track point front image sequence SNPF2M; the first track point front image sequence SNPF2 i includes: [first track point i P1 i, second wind turbine blade 102 front image i];

[0070] The first track point image sequence three includes at least: the first track point front image sequence SNPF31, the first track point front image sequence SNPF32; ...; the first track point front image sequence SNPF3i; ...; the first track point front image sequence SNPF3M; the first track point front image sequence SNPF3i includes: [first track point i P1 i, third wind turbine blade 103 front image i], i=1, 2, 3, ..., M.

[0071] In the present invention, the first inspection information includes the first track point image sequence 1, the first track point image sequence 2 and the first track point image sequence 3.

[0072] The first track point image sequence 1 may include: [first track point 1P11, front image 1 of the first wind turbine blade 101; first track point 1P12, front image 2 of the first wind turbine blade 101; ...; first track point 1P1 M, front image M of the first wind turbine blade 101; overall front image of the first wind turbine blade 101];

[0073] The first track point image sequence 2 may include: [first track point 1P11, front image 1 of the second wind turbine blade 102; first track point 1P12, front image 2 of the second wind turbine blade 101; ...; first track point MP1 M, front image M of the second wind turbine blade 102; overall front image of the second wind turbine blade 102];

[0074] The first track point image sequence three may include: [first track point 1P11, front image 1 of the third wind turbine blade 103; first track point 1P12, front image 2 of the third wind turbine blade 101; ...; first track point 1P1 M, front image M of the third wind turbine blade 103; overall front image of the third wind turbine blade 103].

[0075] On this basis, it can at least be understood that: the first track point image sequence one is used to represent the front image (segmented local) of the first wind turbine blade and the overall front image of the first wind turbine blade at different first track points after the drone inspection of the front of the first wind turbine blade, and the front images of the first wind turbine blade at different first track points can be spliced and combined to form the overall front image of the first wind turbine blade; the first track point image sequence two is used to represent the front image (segmented local) of the second wind turbine blade and the overall front image of the second wind turbine blade at different first track points after the drone inspection of the front of the second wind turbine blade, and the front images of the second wind turbine blade at different first track points can be spliced and combined to form the overall front image of the second wind turbine blade; the first track point image sequence three is used to represent the front image (segmented local) of the third wind turbine blade and the overall front image of the third wind turbine blade at different first track points after the drone inspection of the front of the third wind turbine blade, and the front image (segmented local) of the third wind turbine blade at different first track points can be spliced and combined to form the overall front image of the third wind turbine blade.

[0076] Optionally, the second inspection information includes a second track point image sequence 1, a second track point image sequence 2, and a second track point image sequence 3;

[0077] The second track point image sequence 1 at least includes: a second track point back image sequence SNPB11, a second track point back image sequence SNPB12; ...; a second track point front image sequence SNPB1 j; ...; a second track point back image sequence SNPB1 N; the second track point back image sequence SNPB1 j includes: [second track point jP2j, first wind turbine blade 101 back image j];

[0078] The second track point image sequence 2 at least includes: a second track point back image sequence SNPB21, a second track point back image sequence SNPB22; ...; a second track point front image sequence SNPB2j; ...; a second track point back image sequence SNPB2N; the second track point back image sequence SNPB2j includes: [second track point jP2j, second wind turbine blade 102 back image j];

[0079] The second track point image sequence three includes at least: the second track point back image sequence SNPB31, the second track point back image sequence SNPB32; ...; the second track point front image sequence SNPB3j; ...; the second track point back image sequence SNPB3N; the second track point back image sequence SNPB3j includes: [second track point jP2j, third wind turbine blade 103 back image j], j = 1, 2, 3, ..., N.

[0080] In the present application, the second inspection information includes a second track point image sequence one, a second track point image sequence two, and a second track point image sequence three;

[0081] The second track point image sequence 1 may include: [second track point 1P21, image 1 of the back side of the first wind turbine blade 101; second track point 2P22, image 2 of the back side of the first wind turbine blade 101; ...; second track point NP2N, image N of the back side of the first wind turbine blade 101; and an overall back side image of the first wind turbine blade 101];

[0082] The second track point image sequence 2 may include: [second track point 1P21, image 1 of the back side of the second wind turbine blade 102; second track point 2P22, image 2 of the back side of the second wind turbine blade 102; ...; second track point NP2N, image N of the back side of the second wind turbine blade 102; and an overall back side image of the second wind turbine blade 102];

[0083] The second track point image sequence three may include: [second track point 1P21, image 1 of the back side of the third wind turbine blade 103; second track point 2P22, image 2 of the back side of the third wind turbine blade 103; ...; second track point NP2N, image N of the back side of the third wind turbine blade 103; and an overall back side image of the third wind turbine blade 103].

[0084] On this basis, it can at least be understood that: the second track point image sequence one is used to represent the back image of the first wind turbine blade (segmented local) and the overall back image of the first wind turbine blade at different second track points after the drone inspection of the back of the first wind turbine blade, and the back images of the first wind turbine blade at different first track points can be spliced and combined to form the overall back image of the first wind turbine blade; the second track point image sequence two is used to represent the back image of the second wind turbine blade (segmented local) and the overall back image of the second wind turbine blade at different second track points after the drone inspection of the back of the second wind turbine blade, and the back images of the second wind turbine blade at different second track points can be spliced and combined to form the overall back image of the second wind turbine blade; the second track point image sequence three is used to represent the back image of the third wind turbine blade (segmented local) and the overall back image of the third wind turbine blade at different second track points after the drone inspection of the back of the third wind turbine blade, and the back images of the third wind turbine blade at different second track points can be spliced and combined to form the overall back image of the third wind turbine blade.

[0085] Alternatively, as Figure 3 As shown, starting the drone re-inspection mode includes the following steps:

[0086] Step S401: obtaining a sequence of defect track points corresponding to defective wind turbine blades according to the inspection results of the wind turbine blade drone;

[0087] In the present application, the defect track point image sequence includes one or more of the first track point front image sequence of the wind turbine blade corresponding to the defect type, and / or one or more of the second track point back image sequence; the result of the wind turbine blade having a defect and the wind turbine blade defect type, or the wind turbine blade having no defect and the wind turbine blade defect type being none,

[0088] Accordingly, based on the results of the drone inspection of wind turbine blades, that is, based on the results of the defects of the wind turbine blades and the types of defects of the wind turbine blades, it is possible to reversely determine which one or several wind turbine blades with the defective type come from the front images and / or back images of multiple wind turbine blades, and based on which one or several of the front images and / or back images of the multiple wind turbine blades, it is possible to correspondingly determine which one or several of the front images and / or back images of the multiple wind turbine blades come from one or more of the first track point image sequence one, the first track point image sequence two and the first track point image sequence three, and / or one or more of the second track point image sequence one, the second track point image sequence two and the second track point image sequence three, that is, more specifically, the defective track point image sequence comes from one or more of the first track point front image sequence SNPF1i, SNPF2i, SNPF3i of the first inspection information, and / or one or more of the second track point back image sequence SNPB1j, SNPB2j, SNPB3j in the second inspection information. In this way, one or more first track point front image sequences and / or one or more second track point back image sequences corresponding to the defective type wind turbine blade can be determined, and the defective track point image sequence corresponding to the defective type wind turbine blade can be obtained, that is, the defective track point image sequence includes one or more first track point front image sequences and / or one or more second track point back image sequences of the wind turbine blade corresponding to the defective type.

[0089] Step S402: acquiring defect track points corresponding to the defective wind turbine blades according to the defect track point image sequence, wherein the defect track points form a first re-inspection track point set and / or a second re-inspection track point set;

[0090] In the present application, the defect track point image sequence comes from one or more of the first track point front image sequence SNPF1 i, SNPF2 i, SNPF3i corresponding to the defect type wind turbine blade, and / or comes from one or more of the second track point back image sequence SNPB1 i, SNPB2 i, SNPB3i. Based on the defect track point image sequence, the track point corresponding to the defect type wind turbine blade (also called defect track point) and the corresponding front and / or back of the wind turbine blade can be determined.

[0091] In the present application, the defective track points are selected from one or more of the first track points iP1i, ..., the first track point MP1M, and / or from one or more of the second track points, the second track point 1P21, the second track point 2P22, ..., the second track point jP2j, ..., the second track point NP2N. Accordingly, the defective track points include defective first track point 1, defective first track point 2, ..., defective first track point r, ..., defective first track point P, and / or defective second track point 1, defective second track point 2, ..., defective second track point s, ..., defective second track point Q, where r = 1, 2, 3, ..., P; s = 1, 2, 3, ..., Q. On this basis, the first re-inspection track point set includes defective first track point 1, defective first track point 2,..., defective first track point r,..., defective first track point P, the second re-inspection track point set includes defective second track point 1, defective second track point 2,..., defective second track point s,..., defective second track point Q, the corresponding defective track points include the first track point and / or the second track point, the first track point can be one or more, and / or, the second track point can be one or more, the defective track points can be used to form the first re-inspection track point set and / or the second re-inspection track point set.

[0092] Step S403: planning a drone re-inspection route according to the first re-inspection track point set and / or the second re-inspection track point set, and starting the drone re-inspection of the wind turbine blades.

[0093] In some embodiments, the first re-inspection track point set and / or the second re-inspection track point set, that is, based on the defective first track point 1, the defective first track point 2,..., the defective first track point r,..., the defective first track point P, and / or the defective second track point 1, the defective second track point 2,..., the defective second track point s,..., the defective second track point Q, can plan the drone re-inspection route and start the drone re-inspection of the wind turbine blades.

[0094] In the application of the present invention, based on the defect first track point 1, the defect first track point 2, ..., the defect first track point r, ..., the defect first track point P, and / or the defect second track point 1, the defect second track point 2, ..., the defect second track point s, ..., the defect second track point Q, the re-inspection route of the third drone unit 203 can be planned, and the third drone unit 203 starts the drone re-inspection of the wind turbine blade, returns to one or more first track point positions and / or second track point positions corresponding to the defect type wind turbine blade given in the defect track point image sequence, and performs drone inspection on the wind turbine blade again.

[0095] like Figure 4 、 Figure 5As shown, optionally, the UAV re-inspection route includes a first re-inspection route, a re-inspection switching route T and a second re-inspection route, the re-inspection switching route T connects the first re-inspection route and the second re-inspection route, the first re-inspection track point set is used to form the first re-inspection route, the first re-inspection route is used to perform a front UAV inspection on the wind turbine blade at the first defect track point to form the first re-inspection information, the second re-inspection track point set is used to form the second re-inspection route, the second re-inspection route is used to perform a back UAV inspection on the wind turbine blade at the second defect track point to form the second re-inspection information.

[0096] It should be noted that, at the defect track points, the drone inspection of the front of the wind turbine blade and the drone inspection of the back of the wind turbine blade can preferably be only targeted at the front of the wind turbine blade and / or the back of the wind turbine blade corresponding to the defect type, which can improve the efficiency of the re-inspection; for the defect track points, the front of the wind turbine blade and / or the back of the wind turbine blade corresponding to the non-defect type can also be inspected (the camera takes pictures of the front of the wind turbine blade and / or the back of the wind turbine blade corresponding to the non-defect type) or not inspected (that is, the camera does not take pictures of the front of the wind turbine blade and / or the back of the wind turbine blade corresponding to the non-defect type). Those skilled in the art can make a reasonable choice based on the actual situation based on the image processing needs.

[0097] Obviously, in the application of the present invention, the first re-inspection route is used to perform a front-side drone inspection of the wind turbine blade at the defective track point to form the first re-inspection information, and the second re-inspection route is used to perform a back-side drone inspection of the wind turbine blade at the defective track point to form the second re-inspection information. Accordingly, there will be a technical problem of how to switch the drone to the second re-inspection route for flight inspection after the first re-inspection route is completed, or how to switch the drone to the first re-inspection route for flight inspection after the second re-inspection route is completed.

[0098] Therefore, the applicant conducted technical research on this and proposed a technical solution for the re-inspection switching route T. The UAV can switch to the re-inspection switching route after the flight inspection of the first re-inspection route is completed, and after the flight of the re-inspection switching route is completed, it reaches the starting position of the second re-inspection route, and the UAV performs flight inspection on the second re-inspection route. Alternatively, the UAV can switch to the re-inspection switching route after the flight inspection of the second re-inspection route is completed, and after the flight of the re-inspection switching route is completed, it reaches the starting position of the first re-inspection route, and the UAV performs flight inspection on the first re-inspection route.

[0099] like Figure 4 、 Figure 5As shown, optionally, the defective first track point 1CF1 at the start and the defective first track point PCFP at the end of the first re-inspection route, the defective second track point 1CB1 at the start and the defective second track point QCBQ at the end of the second re-inspection route are obtained, and the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, and the fourth distance D4 between the defective first track point P and the defective second track point Q are calculated; according to the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, the minimum distance Dmin among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4 is obtained, and according to the minimum distance Dmin, the UAV re-inspection route is planned.

[0100] Therefore, in the application of the present invention, the drone re-inspection route is planned according to the minimum distance Dmin. With the help of the drone's advanced navigation and positioning technology, the drone can fly accurately according to the planned re-inspection route to obtain high-resolution images and data, and by setting appropriate shooting angles and segmented shooting images, the integrity and consistency of the data are ensured, which facilitates comparative analysis and timely detection of subtle changes. The planned re-inspection route enables the drone to cover the target area in the shortest time, greatly improving the inspection efficiency. There is no need for a large amount of manpower to conduct long-term inspections on site, which reduces labor costs and labor intensity, and reduces equipment loss. Reasonable route planning can reduce unnecessary flights and operations of drones, extend their service life, and reduce equipment maintenance and replacement costs. According to the planned re-inspection route, the consistency and comparability of the data are guaranteed, which helps to obtain more accurate and comprehensive data, improve the accuracy and reliability of data analysis, and provide strong support for subsequent decision-making on wind turbine blade defect detection results.

[0101] In some embodiments, the spatial coordinates of the first defect track point 1CF1 are (x1, y1, z1), the spatial coordinates of the first defect track point PCFP are (xp, yp, zp), the spatial coordinates of the second defect track point 1CB1 are (x2, y2, z2), and the spatial coordinates of the second defect track point QCBQ are (xq, yq, zq). Accordingly, the first distance D1, the second distance D2, the third distance D3, and the fourth distance D4 are:

[0102]

[0103] Dmin=min[D1, D2, D3, D4].

[0104] Based on the minimum distance Dmin, the planned drone re-inspection route includes:

[0105] (1) When the minimum distance Dmin=D1, the planning of the drone re-inspection route includes: the drone patrols the first re-inspection route in the reverse direction, switches to the re-inspection switching route T, and the drone patrols the second re-inspection route in the forward direction; When the minimum distance Dmin=D2, the planning of the drone re-inspection route includes: the drone patrols the first re-inspection route in the reverse direction, switches to the re-inspection switching route T, and the drone patrols the second re-inspection route in the reverse direction;

[0106] (2) When the minimum distance Dmin=D3, the planned UAV re-inspection route includes: the UAV forward inspects the first re-inspection route, switches to the re-inspection switching route T, and the UAV forward inspects the second re-inspection route; when the minimum distance Dmin=D4, the planned UAV re-inspection route includes: the UAV forward inspects the first re-inspection route, switches to the re-inspection switching route T, and the UAV reverse inspects the second re-inspection route.

[0107] It should be noted that the above-mentioned UAV can use the third UAV unit 203. For the first re-inspection route, the forward inspection means: the UAV hovers at the first defect track point 1CF1 from takeoff, and hovers at the first defect track point PCFP at the end after the flight inspection is completed; the reverse inspection means: the UAV hovers at the first defect track point PCFP from takeoff, and hovers at the first defect track point 1CF1 at the end after the flight inspection is completed; for the second re-inspection route, the forward inspection means: the UAV hovers at the second defect track point 1CB1 from takeoff, and hovers at the second defect track point QCBQ at the end after the flight inspection is completed; the reverse inspection means: the UAV hovers at the second defect track point QCBQ from takeoff, and hovers at the second defect track point 1CB1 at the end after the flight inspection is completed.

[0108] Optionally, updating the wind turbine blade drone inspection results includes the following steps:

[0109] Step S501: Performing a UAV re-inspection of a wind turbine blade according to the UAV re-inspection route to obtain re-inspection information of the wind turbine blade, the re-inspection information including a blade front re-inspection image set and a blade back re-inspection image set; the blade front re-inspection image set includes a plurality of wind turbine blade front re-inspection images; and the blade front re-inspection image set includes a plurality of wind turbine blade back re-inspection images.

[0110] In the present application, at least one of the first wind turbine blade front side re-inspection image CF11, the second wind turbine blade front side re-inspection image CF21 and the third wind turbine blade front side re-inspection image CF31 of the defective first track point 1CF1, at least one of the first wind turbine blade front side re-inspection image CF12, the second wind turbine blade front side re-inspection image CF22 and the third wind turbine blade front side re-inspection image CF32 of the defective first track point 2CF2, ..., at least one of the first wind turbine blade front side re-inspection image CF1r, the second wind turbine blade front side re-inspection image CF2r and the third wind turbine blade front side re-inspection image CF3r of the defective first track point r, ..., at least one of the first wind turbine blade front side re-inspection image CF1p, the second wind turbine blade front side re-inspection image CF2P and the third wind turbine blade front side re-inspection image CF3P of the defective first track point PCFP is obtained. Correspondingly, one or more of the first wind turbine blade front side re-inspection image CF11, the first wind turbine blade front side re-inspection image CF12,..., the first wind turbine blade front side re-inspection image CF1r,..., and the first wind turbine blade front side re-inspection image CF1P are combined to form the first wind turbine blade front side re-inspection image; one or more of the second wind turbine blade front side re-inspection image CF21, the second wind turbine blade front side re-inspection image CF22,..., the second wind turbine blade front side re-inspection image CF2r,..., and the second wind turbine blade front side re-inspection image CF2P are combined to form the second wind turbine blade front side re-inspection image; one or more of the third wind turbine blade front side re-inspection image CF31, the third wind turbine blade front side re-inspection image CF32,..., the third wind turbine blade front side re-inspection image CF3r,..., and the third wind turbine blade front side re-inspection image CF3P form the second wind turbine blade front side re-inspection image. On this basis, the blade front side re-inspection image set includes: one or more of the first wind turbine blade front side re-inspection image, the second wind turbine blade front side re-inspection image, and the third front side re-inspection image, that is, the blade front side re-inspection image set includes: multiple wind turbine blade front side re-inspection images.

[0111] In the present application, at least one of the first wind turbine blade back side re-inspection image CB11, the second wind turbine blade back side re-inspection image CB21 and the third wind turbine blade back side re-inspection image CB31 of the defective second track point 1CB1, at least one of the first wind turbine blade back side re-inspection image CB12, the second wind turbine blade back side re-inspection image CB22 and the third wind turbine blade back side re-inspection image CB32 of the defective second track point 2CB2, ..., at least one of the first wind turbine blade back side re-inspection image CB1s, the second wind turbine blade back side re-inspection image CB2s and the third wind turbine blade back side re-inspection image CB3s of the defective second track point s, ..., at least one of the first wind turbine blade back side re-inspection image CB1Q, the second wind turbine blade back side re-inspection image CB2Q and the third wind turbine blade back side re-inspection image C3Q of the defective second track point QCBQ is obtained. Correspondingly, one or more of the first wind turbine blade back side re-inspection image CB11, the first wind turbine blade back side re-inspection image CB12,..., the first wind turbine blade back side re-inspection image CB1s,..., and the first wind turbine blade back side re-inspection image CB1Q are combined to form the first wind turbine blade back side re-inspection image; one or more of the second wind turbine blade back side re-inspection image CB21, the second wind turbine blade back side re-inspection image CB22,..., the second wind turbine blade back side re-inspection image CB2s,..., and the second wind turbine blade back side re-inspection image CB2Q are combined to form the second wind turbine blade back side re-inspection image; one or more of the third wind turbine blade back side re-inspection image CB31, the third wind turbine blade back side re-inspection image CB32,..., the third wind turbine blade back side re-inspection image CB3s,..., and the third wind turbine blade back side re-inspection image CB3Q form the second wind turbine blade back side re-inspection image. On this basis, the blade back side re-inspection image set includes: one or more of the first wind turbine blade back side re-inspection image, the second wind turbine blade back side re-inspection image, and the third back side re-inspection image, and the blade back side re-inspection image set includes: multiple wind turbine blade back side re-inspection images.

[0112] In some embodiments, the third drone unit 203 may also take off based on the structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm. The camera installed on the third drone unit 203 may only take pictures of wind turbine blades with defects detected in the drone inspection results of wind turbine blades at the defect track point (i.e., only the front and / or back of the wind turbine blades corresponding to the defect type), and may take pictures to obtain images of the front and / or back of the wind turbine blades with defects detected in the drone inspection results of wind turbine blades at the defect track point. This can improve the efficiency of the third drone unit 203 in re-inspecting the wind turbine blades.

[0113] Step S502: Obtain the wind turbine blade drone re-inspection result according to the re-inspection information, and update the wind turbine blade drone inspection result according to the wind turbine blade drone re-inspection result and the wind turbine blade drone inspection result.

[0114] In some embodiments, the results of a UAV re-inspection of wind turbine blades are obtained based on the re-inspection information. Similarly, based on the wind turbine blade defect judgment conditions, the wind turbine blade defect judgment includes: determining whether the wind turbine blade has fatigue defects, impact defects, erosion defects, and shedding defects. Specifically, based on the blade front re-inspection image set (i.e., multiple wind turbine blade front re-inspection images) and the blade back re-inspection image set (i.e., multiple wind turbine blade back re-inspection images), the wind turbine blade is judged to have one or more of fatigue defects, impact defects, erosion defects, and shedding defects. Furthermore, the wind turbine blade defect is judged to have one or more of the following defects: fatigue defects, impact defects, erosion defects, and shedding defects. More specifically, based on one or more wind turbine blade front images with defects detected in the UAV inspection results at the first defect track point, and / or one or more wind turbine blade back images with defects detected in the UAV inspection results at the second defect track point, the wind turbine blade defect judgment determines whether at least one of the first wind turbine blade, the second wind blade, and the third wind blade still has one or more of the following defects: fatigue defects, impact defects, erosion defects, and shedding defects.

[0115] In some embodiments, the wind turbine blade drone re-inspection results include: a result that the wind turbine blade has defects and the wind turbine blade re-inspection defect type, or a result that the wind turbine blade has no defects and the wind turbine blade defect type is none. On this basis, the wind turbine blade drone inspection results can be updated based on the wind turbine blade drone re-inspection results and the wind turbine blade drone inspection results. Specifically, step S502 includes:

[0116] Step S5021: Determine whether the wind turbine blade drone re-inspection result is consistent with the wind turbine blade drone inspection result. If so, update the wind turbine blade drone re-inspection result as the wind turbine blade drone inspection result. If not, go to step S5022.

[0117] Step S5022: at least based on the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the corresponding multiple back re-inspection images of the wind turbine blades, use a first classifier, a second classifier, and a third classifier to respectively classify the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the multiple back re-inspection images of the wind turbine blades, and obtain results of a first defect type set C1 and a first probability set P1 of the first classifier, a second defect type set C2 and a second probability set P2 of the second classifier, and a third defect type set C3 and a third probability set P3 of the third classifier;

[0118] Step S5023: Based on the first defect type set C1, the second defect type set C2 and the third defect type set C3, the first probability set P1, the second probability set P2 and the third probability set P3, calculate the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, evaluate the results of the defect types of the wind turbine blades based on the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, and update the results of evaluating the defect types of the wind turbine blades as the wind turbine blade drone inspection results.

[0119] Specifically, step S5023 includes:

[0120] Step S50231: Calculating the conflict coefficient CT according to the first defect type set C1, the second defect type set C2 and the third defect type set C3 includes:

[0121] CT=m / (3*k)

[0122] Among them, the total number of different defect types is m, and the average number of defect types is k.

[0123] In the present application, the first defect type set C1 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types of the first classifier is k1; the second defect type set C2 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types of the second classifier is k2; the third defect type set C3 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types of the third classifier is k3; the number of different types between the first defect type set C1 and the second defect type set C2 is m1, the number of different types between the second defect type set C2 and the third defect type set C3 is m2, and the number of different types between the third defect type set C3 and the third defect type set C1 is m3. Accordingly, the total number m of different defect types is: m = m1 + m2 + m3, and the average number of defect types is k = (k1 + k2 + k3) / 3.

[0124] In the present application, the first probability set P1 corresponding to the defect types in the first defect type set C1 is: P1 = [P11, P12, ..., P1 k1], the second probability set P2 corresponding to the defect types in the second defect type set C2 is: P2 = [P21, P22, ..., P2k2], and the third probability set P3 corresponding to the defect types in the third defect type set C3 is: P3 = [P31, P32, ..., P3k3]. On this basis, the confidence expectations E(C-1), E(C-2), E(C-3), and E(C-4) of the defect types C-1, C-2, C-3, and C-4 can be calculated respectively. The general calculation formula is:

[0125] E(CT)=w1P1T+w2P2T+w3P3T

[0126] Where T = 1, 2, 3, 4, w1 is the confidence of the first classifier, w2 is the confidence of the second classifier, w3 is the confidence of the third classifier, w1+w2+w3=1, P1T is the corresponding probability of the corresponding defect type in the first probability set P1, P2T is the corresponding probability of the corresponding defect type in the second probability set P2, and P3T is the corresponding probability of the corresponding defect type in the third probability set P3.

[0127] The average confidence expectation of different defect types C-1, C-2, C-3, and C-4 is: EC = (E(C-1) + E(C-2) + E(C-3) + E(C-4)) / 4.

[0128] In the present application, based on the first defect type set C1, the second defect type set C2, and the third defect type set, the number of votes R1 for defect type C-1, the number of votes R2 for defect type C-2, the number of votes R3 for defect type C-3, and the number of votes R4 for defect type C-4 can be calculated. The number of votes of different types is normalized. Accordingly, the true voting expectation V1 for defect type C-1, the true voting expectation V2 for defect type C-2, the true voting expectation V3 for defect type C-3, the true voting expectation V4 for defect type C-4, and the average true voting expectation V are:

[0129] V1=G1*R1 / (R1+R2+R3+R4), V2=G2*R2 / (R1+R2+R3+R4);

[0130] V3=G3*R3 / (R1+R2+R3+R4), V4=G4*R4 / (R1+R2+R3+R4);

[0131] V = (V1 + V2 + V3 + V4) / 4;

[0132] Among them, G1 is the first voting adjustment coefficient, G2 is the second voting adjustment coefficient, G3 is the third voting adjustment coefficient, and G4 is the fourth voting adjustment coefficient. G1, G2, G3, and G4 are constants. G1, G2, G3, and G4 are constants and can take values between 0.1-20 (including values 0.1, 1.0, 1.5, 2.0, 20, etc.), which can be reasonably set according to actual conditions.

[0133] Step S50232: evaluating the defect type of the wind turbine blade according to the conflict coefficient CT, the confidence expectations of different defect types, and the true voting expectations of different defect types.

[0134] Specifically, step S50232 includes:

[0135] Step S502321 includes: calculating the expectations of different defect types based on the conflict coefficient CT, the confidence expectations of different defect types, and the voting expectations of different defect types. Accordingly, the expectations and average expectations of different defect types C-1, C-2, C-3, and C-4 are E1, E2, E3, E4, and E are:

[0136] E1=E(C-1)*V1, E2=E(C-2)*V2;

[0137] E3=E(C-3)*V3, E4=E(C-4)*V4;

[0138] E=(E1+E2+E3+E4) / 4;

[0139] Step S502322 includes:

[0140] When CT≥0.8, the defect type that satisfies the expectation [E1, E2, E3, E4]≤E is taken as the result of evaluating the defect type of the wind turbine blade; when CT<0.8, the defect type that satisfies the true voting expectation [V1, V2, V3, V4]≥V is taken as the result of evaluating the defect type of the wind turbine blade.

[0141] In the present invention application, when CT≥0.8, it indicates that there are large differences between the first classifier, the second classifier, and the third classifier. In this case, it is necessary to comprehensively consider the confidence expectation and the voting expectation to determine the defect type. By introducing a comparison mechanism of the product and the average expectation, when the product of the confidence expectation and the true voting expectation of different defect types is still less than the average expectation, the defect type corresponding to these expectations is used as the result of evaluating the defect type of the wind turbine blade. In this way, reliable defect types can be screened out more rigorously, and the classification errors caused by classifier conflicts can be reduced, making the evaluation of wind turbine blade defect types more accurate and stable. It also helps to improve the accuracy of wind turbine blade defect identification in complex classification situations, reduce the misleading of subsequent maintenance decisions by misjudgment, reduce unnecessary maintenance costs and resource waste caused by the classifier's incorrect judgment of the defect type, ensure the safe operation of wind turbine blades, and ensure that the wind farm does not stop, thereby improving the operating efficiency of the entire wind farm.

[0142] In the present application, when CT<0.8, it means that the differences between the first classifier, the second classifier, and the third classifier are not large, and the confidence of the first classifier, the second classifier, and the third classifier is high. In this way, the relatively consistent judgments between the classifiers can be efficiently utilized, focusing on the true voting expectations. It is only necessary to comprehensively consider the true voting expectations to determine the defect type. When the true voting expectations of different defect types are still greater than the average true voting expectations, the defect types corresponding to these true voting expectations are used as the results of evaluating the defect types of wind turbine blades. In this way, the defect type of the wind turbine blade can be quickly identified, the complex multi-factor calculation process is avoided, and the efficiency and timeliness of the wind turbine blade defect assessment are improved. This is especially important for the inspection of large-scale wind turbine blades. A large amount of data can be processed in a short time and reliable results can be obtained. At the same time, when the classifier consistency is high, the advantages of multiple classifiers can be fully utilized to reduce the misclassification caused by individual factors or accidental errors, avoid unnecessary repairs or miss the best repair time due to misjudgment, reduce maintenance costs and wind turbine blade downtime, and improve the availability and power generation efficiency of wind power equipment. Moreover, with the accumulation of data and the improvement of the deep learning performance of the classifier, the accuracy and reliability are expected to be further improved, which can promote the development of intelligent drone inspections in the wind power industry towards a more refined and precise direction.

[0143] like Figure 6As shown, the present invention provides a wind turbine blade drone inspection device, which implements any wind turbine blade drone inspection method described in the first aspect above, including a parameter acquisition unit 100, an inspection unit 200, a display unit 300, a re-inspection unit 400 and a server 500:

[0144] The parameter acquisition unit 100 acquires structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm and sends them to the inspection unit. The structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades. The motion parameters include at least the wind turbine speed. The environmental parameters include at least the altitude and wind speed.

[0145] The inspection unit 200 performs a drone inspection on the front of a wind turbine blade based on structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, generates first inspection information, and transmits it to a server. Simultaneously, it performs a drone inspection on the back of the wind turbine blade, generates second inspection information, and transmits it to the server. The first inspection information includes a first track point dataset and a blade front image set, and the second inspection information includes a second track point dataset and a blade back image set. The blade front image set includes a plurality of complete front images of the entire wind turbine blade; and the blade back image set includes a plurality of complete back images of the entire wind turbine blade.

[0146] The server 500 obtains the wind turbine blade drone inspection result based on the first inspection information and the second inspection information and the wind turbine blade defect judgment condition, and sends it to the re-inspection unit 400 and the display unit 300, and the display unit displays the wind turbine blade drone inspection result;

[0147] The re-inspection unit 400 displays an interactive interface through the display unit 300 to indicate whether to start the drone re-inspection mode based on the drone inspection results of the wind turbine blades; if yes, the drone re-inspection mode is started and the drone inspection results of the wind turbine blades are updated; if not, the re-inspection mode is not started.

[0148] It should be noted that the wind turbine blade drone inspection device applied for by the present invention is used for any of the wind turbine blade drone inspection methods described in the first aspect above, and accordingly also includes: all the technical problems, technical solutions and technical effects recorded in any of the wind turbine blade drone inspection methods in the first aspect, and the present invention application will not be repeated here.

[0149] like Figure 7As shown, optionally, the re-inspection unit includes a defect image search unit 401, a defect track point forming unit 402, a re-inspection route planning unit 403 and a starting unit 404. The defect image search unit 401 obtains a defect track point image sequence corresponding to the defect type of the wind turbine blade based on the UAV inspection result of the wind turbine blade; the defect track point forming unit 402 obtains the defect track points corresponding to the defect type of the wind turbine blade based on the defect track point image sequence, and forms the defect track points into a first re-inspection track point set and / or a second re-inspection track point set; the re-inspection route planning unit 403 plans the UAV re-inspection route based on the first re-inspection track point set and / or the second re-inspection track point set; the starting unit 404 starts the UAV re-inspection of the wind turbine blade.

[0150] In the present invention application, the defect track point image sequence includes one or more first track point front image sequences of the wind turbine blade corresponding to the defect type, and / or one or more second track point back image sequences. All technical contents such as the first track point front image sequence and the second track point back image sequence can be found in the first aspect of the present invention application and will not be repeated.

[0151] In the present application, the defect image search unit can obtain a sequence of defect track point images corresponding to the defect type of the wind turbine blade based on the results of the drone inspection of the wind turbine blade; the defect track point formation unit can obtain defect track points corresponding to the defect type of the wind turbine blade based on the defect track point image sequence, and form the defect track points into a first re-inspection track point set and / or a second re-inspection track point set; the re-inspection route planning unit can plan a drone re-inspection route based on the first re-inspection track point set and / or the second re-inspection track point set; and the startup unit can start the drone re-inspection of the wind turbine blade. For example, the drone re-inspection route and the startup unit can also be displayed on the interactive interface of the display unit 300 or another human-computer interaction interface.

[0152] Optionally, the drone re-inspection route includes a first re-inspection route, a re-inspection switching route T and a second re-inspection route, the re-inspection switching route connects the first re-inspection route and the second re-inspection route, the first re-inspection track point set is used to form a first re-inspection route, the first re-inspection route is used to perform a frontal drone inspection on the defect type of the wind turbine blade to form a first re-inspection information, the second re-inspection track point set is used to form a second re-inspection route, the second re-inspection route is used to perform a back drone inspection on the defect type of the wind turbine blade to form a second re-inspection information; the re-inspection route planning unit 403 includes a distance calculation subunit, the distance calculation subunit obtains the first defect track point 1 starting from the first re-inspection route and the defective first track point P at the end, the defective second track point 1 at the starting point of the second re-inspection route and the defective second track point Q at the end, calculate the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, the fourth distance D4 between the defective first track point P and the defective second track point Q, obtain the minimum distance Dmin = min[D1, D2, D3, D4] among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, and send the minimum distance Dmin to the re-inspection route planning unit 403.

[0153] In some embodiments, the spatial coordinates of the first defect track point 1CF1 are (x1, y1, z1), the spatial coordinates of the first defect track point PCFP are (xp, yp, zp), the spatial coordinates of the second defect track point 1CB1 are (x2, y2, z2), and the spatial coordinates of the second defect track point QCBQ are (xq, yq, zq). Accordingly, the first distance D1, the second distance D2, the third distance D3, and the fourth distance D4 are:

[0154]

[0155] Dmin=min[D1, D2, D3, D4].

[0156] Based on the minimum distance Dmin, the planned drone re-inspection route includes:

[0157] (1) When the minimum distance Dmin=D1, the planning of the drone re-inspection route includes: the drone patrols the first re-inspection route in the reverse direction, switches to the re-inspection switching route T, and the drone patrols the second re-inspection route in the forward direction; When the minimum distance Dmin=D2, the planning of the drone re-inspection route includes: the drone patrols the first re-inspection route in the reverse direction, switches to the re-inspection switching route T, and the drone patrols the second re-inspection route in the reverse direction;

[0158] (2) When the minimum distance Dmin=D3, the planned UAV re-inspection route includes: the UAV forward inspects the first re-inspection route, switches to the re-inspection switching route T, and the UAV forward inspects the second re-inspection route; when the minimum distance Dmin=D4, the planned UAV re-inspection route includes: the UAV forward inspects the first re-inspection route, switches to the re-inspection switching route T, and the UAV reverse inspects the second re-inspection route.

[0159] Optionally, the re-inspection unit 400 also includes a re-inspection information acquisition unit 405, a first evaluation unit 406 and a second evaluation unit 407. The re-inspection information acquisition unit 405 re-inspects the wind turbine blades by drone according to the drone re-inspection route to obtain re-inspection information of the wind turbine blades; the first evaluation unit 406 obtains the wind turbine blade drone re-inspection results based on the re-inspection information; the second evaluation unit 407 updates the wind turbine blade drone inspection results based on the wind turbine blade drone re-inspection results and the wind turbine blade drone inspection results.

[0160] In the present application, the re-inspection information acquisition unit 405 re-inspects the wind turbine blades by drone according to the drone re-inspection route to obtain the re-inspection information of the wind turbine blades; the first evaluation unit 406 obtains the wind turbine blade drone re-inspection results based on the re-inspection information; the second evaluation unit 407 updates the wind turbine blade drone inspection results based on the wind turbine blade drone re-inspection results and the wind turbine blade drone inspection results.

[0161] The second evaluation unit 407 includes a judgment sub-unit 4071, a processing sub-unit 4072 and an evaluation update sub-unit 4073; the judgment sub-unit 4071 judges whether the wind turbine blade UAV re-inspection result and the wind turbine blade UAV inspection result are consistent. If so, the evaluation update sub-unit 4073 updates the wind turbine blade UAV re-inspection result as the wind turbine blade UAV inspection result; if not, the processing sub-unit 4072 uses the first classifier, the second classifier and the third classifier to respectively classify the overall front images of the multiple wind turbine blades corresponding to the defect type and / or the overall back images of the multiple wind turbine blades corresponding to the defect type, and the corresponding multiple front re-inspection images and / or the multiple back re-inspection images of the wind turbine blades. Classify the front inspection images of the wind turbine blades and / or the back inspection images of the wind turbine blades to obtain the results of the first defect type set C1 and the first probability set P1 of the first classifier, the second defect type set C2 and the second probability set P2 of the second classifier, and the third defect type set C3 and the third probability set P3 of the third classifier; calculate the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types based on the first defect type set C1, the second defect type set C2 and the third defect type set C3, the first probability set P1, the second probability set P2 and the third probability set P3, and evaluate the results of the defect types of the wind turbine blades based on the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types. The evaluation update sub-unit 4073 updates the results of evaluating the defect types of the wind turbine blades as the wind turbine blade drone inspection results.

[0162] In some embodiments, calculating the conflict coefficient CT according to the first defect type set C1, the second defect type set C2, and the third defect type set C3 includes:

[0163] CT=m / (3*k)

[0164] Among them, the total number of different defect types is m, and the average number of defect types is k.

[0165] In some embodiments, the first defect type set C1 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types in the first classifier is k1; the second defect type set C2 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types in the second classifier is k2; the third defect type set C3 is at least one of [C-1, C-2, C-3, C-4], and the total number of defect types in the third classifier is k3; the number of different types between the first defect type set C1 and the second defect type set C2 is m1, the number of different types between the second defect type set C2 and the third defect type set C3 is m2, and the number of different types between the third defect type set C3 and the third defect type set C1 is m3. Accordingly, the total number m of different defect types is: m=m1+m2+m3, and the average number of defect types is k=(k1+k2+k3) / 3.

[0166] In some embodiments, the first probability set P1 corresponding one-to-one to the defect types in the first defect type set C1 is: P1 = [P11, P12, ..., P1 k1], the second probability set P2 corresponding one-to-one to the defect types in the second defect type set C2 is: P2 = [P21, P22, ..., P2k2], and the third probability set P3 corresponding one-to-one to the defect types in the third defect type set C3 is: P3 = [P31, P32, ..., P3k3]. Based on this, the confidence expectations E(C-1), E(C-2), E(C-3), and E(C-4) of the defect types C-1, C-2, C-3, and C-4 can be calculated respectively, and the general calculation formula is:

[0167] E(CT)=w1P1T+w2P2T+w3P3T

[0168] Where T = 1, 2, 3, 4, w1 is the confidence of the first classifier, w2 is the confidence of the second classifier, w3 is the confidence of the third classifier, w1+w2+w3=1, P1T is the corresponding probability of the corresponding defect type in the first probability set P1, P2T is the corresponding probability of the corresponding defect type in the second probability set P2, and P3T is the corresponding probability of the corresponding defect type in the third probability set P3.

[0169] The average confidence expectation of different defect types C-1, C-2, C-3, and C-4 is: EC = (E(C-1) + E(C-2) + E(C-3) + E(C-4)) / 4.

[0170] In some embodiments, based on the first defect type set C1, the second defect type set C2, and the third defect type set, the number of votes R1 for defect type C-1, the number of votes R2 for defect type C-2, the number of votes R3 for defect type C-3, and the number of votes R4 for defect type C-4 can be calculated. The number of votes of different types can be normalized. Accordingly, the true voting expectation V1 for defect type C-1, the true voting expectation V2 for defect type C-2, the true voting expectation V3 for defect type C-3, the true voting expectation V4 for defect type C-4, and the average true voting expectation V are:

[0171] V1=G1*R1 / (R1+R2+R3+R4), V2=G2*R2 / (R1+R2+R3+R4);

[0172] V3=G3*R3 / (R1+R2+R3+R4), V4=G4*R4 / (R1+R2+R3+R4);

[0173] V = (V1 + V2 + V3 + V4) / 4;

[0174] Among them, G1 is the first voting adjustment coefficient, G2 is the second voting adjustment coefficient, G3 is the third voting adjustment coefficient, and G4 is the fourth voting adjustment coefficient. G1, G2, G3, and G4 are constants. G1, G2, G3, and G4 are constants and can take values between 0.1-20 (including values 0.1, 1.0, 1.5, 2.0, 20, etc.), which can be reasonably set according to actual conditions.

[0175] In some embodiments, the results of evaluating the defect type of the wind turbine blade according to the conflict coefficient CT, the confidence expectations of different defect types, and the true voting expectations of different defect types include:

[0176] According to the conflict coefficient CT, the confidence expectation of different defect types and the voting expectation of different defect types, the expectations of different defect types are calculated. Accordingly, the expectations and average expectations of different defect types C-1, C-2, C-3, and C-4 are E1, E2, E3, E4, and E are:

[0177] E1=E(C-1)*V1, E2=E(C-2)*V2;

[0178] E3=E(C-3)*V3, E4=E(C-4)*V4;

[0179] E=(E1+E2+E3+E4) / 4;

[0180] When CT≥0.8, the defect type that satisfies the expectation [E1, E2, E3, E4]≤E is taken as the result of evaluating the defect type of the wind turbine blade; when CT<0.8, the defect type that satisfies the true voting expectation [V1, V2, V3, V4]≥V is taken as the result of evaluating the defect type of the wind turbine blade.

[0181] In the present invention application, when CT≥0.8, it indicates that there are large differences between the first classifier, the second classifier, and the third classifier. In this case, it is necessary to comprehensively consider the confidence expectation and the voting expectation to determine the defect type. By introducing a comparison mechanism of the product and the average expectation, when the product of the confidence expectation and the true voting expectation of different defect types is still less than the average expectation, the defect type corresponding to these expectations is used as the result of evaluating the defect type of the wind turbine blade. In this way, reliable defect types can be screened out more rigorously, and the classification errors caused by classifier conflicts can be reduced, making the evaluation of wind turbine blade defect types more accurate and stable. It also helps to improve the accuracy of wind turbine blade defect identification in complex classification situations, reduce the misleading of subsequent maintenance decisions by misjudgment, reduce unnecessary maintenance costs and resource waste caused by the classifier's incorrect judgment of the defect type, ensure the safe operation of wind turbine blades, and ensure that the wind farm does not stop, thereby improving the operating efficiency of the entire wind farm.

[0182] In the present application, when CT<0.8, it means that the differences between the first classifier, the second classifier, and the third classifier are not large, and the confidence of the first classifier, the second classifier, and the third classifier is high. In this way, the relatively consistent judgments between the classifiers can be efficiently utilized, focusing on the true voting expectations. It is only necessary to comprehensively consider the true voting expectations to determine the defect type. When the true voting expectations of different defect types are still greater than the average true voting expectations, the defect types corresponding to these true voting expectations are used as the results of evaluating the defect types of wind turbine blades. In this way, the defect type of the wind turbine blade can be quickly identified, the complex multi-factor calculation process is avoided, and the efficiency and timeliness of the wind turbine blade defect assessment are improved. This is especially important for the inspection of large-scale wind turbine blades. A large amount of data can be processed in a short time and reliable results can be obtained. At the same time, when the classifier consistency is high, the advantages of multiple classifiers can be fully utilized to reduce the misclassification caused by individual factors or accidental errors, avoid unnecessary repairs or miss the best repair time due to misjudgment, reduce maintenance costs and wind turbine blade downtime, and improve the availability and power generation efficiency of wind power equipment. Moreover, with the accumulation of data and the improvement of the deep learning performance of the classifier, the accuracy and reliability are expected to be further improved, which can promote the development of intelligent drone inspections in the wind power industry towards a more refined and precise direction.

[0183] In the third aspect, the present invention application provides a wind turbine blade drone inspection device, which includes a processor, a memory, and a computing program stored on the memory and executable by the processor, wherein when the computing program is executed by the processor, the steps of the wind turbine blade drone inspection method as described in any one of the first aspects are implemented.

[0184] In a fourth aspect, the present invention application provides a computer-readable storage medium having a computing program stored thereon, wherein when the computing program is executed by a processor, the steps of the wind turbine blade drone inspection method as described in any one of the first aspects are implemented.

[0185] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. Those skilled in the art can understand that the various operations, methods, steps, measures, and schemes in the process discussed in the present application can be alternated, changed, combined or deleted; further, other steps, measures, and schemes in the various operations, methods, and processes discussed in the present application can also be alternated, changed, rearranged, decomposed, combined or deleted; further, the steps, measures, and schemes in the prior art that are the same as those disclosed in the present application can also be alternated, changed, rearranged, decomposed, combined or deleted.

[0186] Those skilled in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention. Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0187] In the several embodiments provided by the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (hereinafter referred to as: ROM), random access memory (hereinafter referred to as: RAM), disk or optical disk, and other media that can store program code.

[0188] The above-described embodiments only express several implementation methods of the embodiments of the present disclosure, and their descriptions are relatively specific and detailed, but they cannot be understood as limiting the patent scope of the embodiments of the present disclosure. It should be pointed out that for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the embodiments of the present disclosure, and these all fall within the scope of protection of the embodiments of the present disclosure. Therefore, the scope of protection of the embodiments of the present disclosure should be based on the attached claims.

Claims

1. A wind turbine blade drone inspection method, characterized in that: The following steps are involved: Step S100, obtaining structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, wherein the structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the altitude and wind speed; Step S200: Based on the structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, a drone inspection is performed on the front of the wind turbine blades to generate first inspection information, and a drone inspection is performed on the back of the wind turbine blades to generate second inspection information, wherein the first inspection information includes a first track point dataset and a blade front image set, and the second inspection information includes a second track point dataset and a blade back image set. The blade front image set includes a plurality of wind turbine blade overall front images; The blade back image set includes multiple wind turbine blade back images; Step S300: Obtaining wind turbine blade drone inspection results based on the first inspection information and the second inspection information in combination with wind turbine blade defect judgment conditions; According to the inspection results of the wind turbine blade drone, determine whether to start the drone re-inspection mode. If so, start the drone re-inspection mode and update the wind turbine blade drone inspection results; If not, do not start the re-inspection mode; Starting the drone re-inspection mode includes the following steps: Step S401: obtaining a sequence of defect track points corresponding to defective wind turbine blades according to the inspection results of the wind turbine blade drone; Step S402: acquiring defect track points corresponding to the defective wind turbine blades according to the defect track point image sequence, wherein the defect track points form a first re-inspection track point set and / or a second re-inspection track point set; Step S403: planning a drone re-inspection route according to the first re-inspection track point set and / or the second re-inspection track point set, and starting the drone re-inspection of the wind turbine blades.

2. A wind turbine blade drone inspection method according to claim 1, characterized in that: Updating wind turbine blade drone inspection results includes the following steps: Step S501: Performing a UAV re-inspection of a wind turbine blade according to the UAV re-inspection route to obtain re-inspection information of the wind turbine blade, the re-inspection information including a blade front re-inspection image set and a blade back re-inspection image set; the blade front re-inspection image set includes a plurality of wind turbine blade front re-inspection images; and the blade front re-inspection image set includes a plurality of wind turbine blade back re-inspection images. Step S502: Obtain the wind turbine blade drone re-inspection result according to the re-inspection information, and update the wind turbine blade drone inspection result according to the wind turbine blade drone re-inspection result and the wind turbine blade drone inspection result.

3. The wind turbine blade inspection method using a drone according to claim 2, characterized in that: Step S502 includes: Step S5021: determine whether the wind turbine blade drone re-inspection result is consistent with the wind turbine blade drone inspection result. If so, update the wind turbine blade drone re-inspection result as the wind turbine blade drone inspection result. If not, proceed to step S5022. Step S5022: at least based on the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the corresponding multiple back re-inspection images of the wind turbine blades, use a first classifier, a second classifier, and a third classifier to respectively classify the overall front images of the plurality of wind turbine blades corresponding to the defect types and / or the overall back images of the plurality of wind turbine blades corresponding to the defect types, and the corresponding multiple front re-inspection images of the wind turbine blades and / or the multiple back re-inspection images of the wind turbine blades, and obtain results of a first defect type set C1 and a first probability set P1 of the first classifier, a second defect type set C2 and a second probability set P2 of the second classifier, and a third defect type set C3 and a third probability set P3 of the third classifier; Step S5023: Based on the first defect type set C1, the second defect type set C2 and the third defect type set C3, the first probability set P1, the second probability set P2 and the third probability set P3, calculate the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, evaluate the results of the defect types of the wind turbine blades based on the conflict coefficient CT, the confidence expectations of different defect types and the true voting expectations of different defect types, and update the results of evaluating the defect types of the wind turbine blades as the wind turbine blade drone inspection results.

4. The wind turbine blade inspection method using a drone according to claim 3, characterized in that: The defect track point includes a first track point and a second track point, the UAV re-inspection route includes a first re-inspection route, a re-inspection switching route and a second re-inspection route, the re-inspection switching route connects the first re-inspection route and the second re-inspection route, the first re-inspection track point set is used to form a first re-inspection route, the first re-inspection route is used to perform a front UAV inspection on the wind turbine blade at the defect first track point to form first re-inspection information, the second re-inspection track point set is used to form a second re-inspection route, the second re-inspection route is used to perform a back UAV inspection on the wind turbine blade at the defect second track point to form second re-inspection information; Obtain the defective first track point 1 at the start and the defective first track point P at the end of the first re-inspection route, the defective second track point 1 at the start and the defective second track point Q at the end of the second re-inspection route, calculate the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, and the fourth distance D4 between the defective first track point P and the defective second track point Q; according to the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, obtain the minimum distance Dmin = min[D1, D2, D3, D4] among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, and plan the UAV re-inspection route according to the minimum distance Dmin.

5. A wind turbine blade drone inspection device, which implements the wind turbine blade drone inspection method according to any one of claims 1 to 4, comprising a parameter acquisition unit, an inspection unit, a display unit, a re-inspection unit, and a server: a parameter acquisition unit, which acquires structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm and sends them to the inspection unit, wherein the structural parameters include at least the impeller center height, blade profile curve, blade length, maximum chord length, impeller diameter, and number of blades; the motion parameters include at least the wind turbine speed; and the environmental parameters include at least the altitude and wind speed; The inspection unit performs a drone inspection on the front of a wind turbine blade based on structural parameters, motion parameters, and environmental parameters of at least one wind turbine in the wind farm, generates first inspection information, and transmits it to a server. Simultaneously, it performs a drone inspection on the back of the wind turbine blade, generates second inspection information, and transmits it to the server. The first inspection information includes a first track point data set and a blade front image set, and the second inspection information includes a second track point data set and a blade back image set. The blade front image set includes a plurality of entire front images of the wind turbine blade. The blade back image set includes multiple wind turbine blade back images; The server obtains the wind turbine blade drone inspection result based on the first inspection information and the second inspection information and the wind turbine blade defect judgment condition, and sends the result to the re-inspection unit and the display unit, and the display unit displays the wind turbine blade drone inspection result; The re-inspection unit displays an interactive interface for starting the drone re-inspection mode based on the inspection results of the wind turbine blades; If yes, start the drone re-inspection mode and update the wind turbine blade drone inspection results; If not, the re-inspection mode is not started.

6. The wind turbine blade inspection device using a drone according to claim 5, characterized in that: The re-inspection unit includes a defect image search unit, a defect track point formation unit, a re-inspection route planning unit and a starting unit. The defect image search unit obtains a defect track point image sequence corresponding to the defect type of the wind turbine blade based on the inspection result of the wind turbine blade drone; the defect track point formation unit obtains the defect track points corresponding to the defect type of the wind turbine blade based on the defect track point image sequence, and forms the defect track points into a first re-inspection track point set and / or a second re-inspection track point set; The re-inspection route planning department plans the UAV re-inspection route based on the first re-inspection track point set and / or the second re-inspection track point set; The startup department initiated the drone re-inspection of wind turbine blades.

7. The wind turbine blade inspection device using a drone according to claim 6, characterized in that: The drone re-inspection route includes the first re-inspection route, the re-inspection switching route T and the second re-inspection route. The re-inspection switching route connects the first re-inspection route and the second re-inspection route. The first re-inspection track point set is used to form the first re-inspection route. The first re-inspection route is used to perform a frontal drone inspection of the defect type of the wind turbine blade to form the first re-inspection information. The second re-inspection track point set is used to form the second re-inspection route. The second re-inspection route is used to perform a back drone inspection of the defect type of the wind turbine blade to form the second re-inspection information.

8. The wind turbine blade inspection device using a drone according to claim 7, characterized in that: The re-inspection route planning unit includes a distance calculation sub-unit, which obtains the defective first track point 1 at the start and the defective first track point P at the end of the first re-inspection route, the defective second track point 1 at the start and the defective second track point Q at the end of the second re-inspection route, calculates the first distance D1 between the defective first track point 1 and the defective second track point 1, the second distance D2 between the defective first track point 1 and the defective second track point Q, the third distance D3 between the defective first track point P and the defective second track point 1, and the fourth distance D4 between the defective first track point P and the defective second track point Q, obtains the minimum distance Dmin = min[D1, D2, D3, D4] among the first distance D1, the second distance D2, the third distance D3 and the fourth distance D4, and sends the minimum distance Dmin to the re-inspection route planning unit.

9. The wind turbine blade drone inspection device according to claim 8, characterized in that: The re-inspection unit further includes a re-inspection information acquisition unit, a first evaluation unit, and a second evaluation unit. The re-inspection information acquisition unit re-inspects the wind turbine blades by the drone according to the re-inspection route of the drone to obtain re-inspection information of the wind turbine blades; the first evaluation unit obtains the re-inspection result of the wind turbine blades by the drone according to the re-inspection information; The second assessment department updates the wind turbine blade drone inspection results based on the wind turbine blade drone re-inspection results and wind turbine blade drone inspection results.

Images