Blade position angle calculation method, device and equipment of fan and medium

Abstract

The invention relates to the technical field of unmanned aerial vehicle autonomous inspection, and discloses a fan blade position angle calculation method, device and equipment and a medium, and the method comprises the steps: obtaining a first point cloud model of a target fan; filtering the first point cloud model in the target coordinate system according to the height threshold to obtain a second point cloud model; deleting points which are in the target range after being projected to the horizontal plane of the target coordinate system in the second point cloud model to obtain a third point cloud model; deleting the center hub part feature points in the geometric center area in the third point cloud model to obtain a fourth point cloud set; and obtaining a bounding box of each point cluster in the fourth point cloud model, and determining the orientation angle of the bounding box of each point cluster as the blade position angle of the target fan. According to the scheme, the calculation efficiency is improved while the angle recognition accuracy is guaranteed, and the timeliness of calculation of the blade position angle of the fan is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle autonomous inspection, in particular to a blade position angle calculation method, device, equipment and medium of a fan. BACKGROUND

[0002] A wind turbine is installed with dozens of meters long blades, which rotate to drive the motor inside the wind turbine, and finally convert wind energy into electrical energy, so the normal operation of the blades is crucial to the smooth power output of the wind turbine.

[0003] In order to ensure the normal operation of the wind turbine, it is necessary to regularly inspect the fan blades, and timely find the abnormalities and defects for repair to ensure the normal operation of the generator set. The current mainstream inspection methods are divided into two kinds, one is to observe by the staff on the ground with a telescope, and the other is to make a close shot by the pilot on the ground by controlling the unmanned aerial vehicle carrying a camera to inspect the blades. The basis and key technology of unmanned aerial vehicle automatic inspection is the measurement and acquisition of random attitude parameters such as blade position angle when the fan is stopped. In the prior art, laser radar can be used to scan the 3D point cloud of the fan and its surrounding scene, so as to position and navigate the blades. Although this method has high accuracy, due to the large size of the fan, the time-consuming of the pre-data collection and processing by laser radar is long, which restricts the high timeliness requirement of this business. SUMMARY

[0004] Therefore, the present application provides a blade position angle calculation method, device, equipment and medium of a fan, which improves the timeliness of the blade position angle calculation of the fan. The technical solution is as follows.

[0005] In a first aspect, a blade position angle calculation method of a fan is provided, the method comprising:

[0006] obtaining a first point cloud model of a target fan; the first point cloud model comprising each feature point in a target coordinate system;

[0007] filtering the first point cloud model in the target coordinate system according to a height threshold to obtain a second point cloud model;

[0008] deleting points in the second point cloud model that are projected to the horizontal plane of the target coordinate system and within a target range to obtain a third point cloud model; the target range being determined according to elliptical fitting of the planar point set of the second point cloud model projected to the horizontal plane of the target coordinate system;

[0009] deleting the center hub part feature points in the third point cloud model located in the geometric center region to obtain a fourth point cloud set;

[0010] An orientation angle of the bounding box of each point cluster in the fourth point cloud model is determined as a blade position angle of the target wind turbine.

[0011] In an optional implementation, the first point cloud model of the target wind turbine is obtained by:

[0012] A wind turbine picture of the target wind turbine is obtained.

[0013] A three-dimensional reconstruction is performed based on the wind turbine picture of the target wind turbine, and the first point cloud model is obtained.

[0014] In an optional implementation, the wind turbine picture of the target wind turbine is obtained by:

[0015] Physical parameters of the target wind turbine are obtained.

[0016] Based on the physical parameters of the target wind turbine, an origin of a target coordinate system and a flight trajectory of the unmanned aerial vehicle are determined.

[0017] A specified number of wind turbine pictures are collected in a circling process of the unmanned aerial vehicle according to the flight trajectory and with the nose of the unmanned aerial vehicle kept facing the origin.

[0018] In an optional implementation, before the first point cloud model in the target coordinate system is filtered according to the height threshold, the method further includes:

[0019] A height difference of the first point cloud model in a vertical direction is obtained.

[0020] A specified proportion of the height difference is obtained as the height threshold.

[0021] In an optional implementation, the first point cloud model in the target coordinate system is filtered to obtain a second point cloud model, including:

[0022] Points of the first point cloud model with a height less than the height threshold in the target coordinate system are deleted to obtain a first candidate model.

[0023] Points in the first candidate model are subjected to Euclidean clustering to obtain the second point cloud model.

[0024] In an optional implementation, before the points in the second point cloud model that are within a target range after being projected onto a horizontal plane of the target coordinate system are deleted to obtain a third point cloud model, the method further includes:

[0025] A point cloud in a specified height range is selected with a Y-axis coordinate of a bottom end of the second point cloud model as a starting point, and the point cloud is projected onto the horizontal plane of the target coordinate system to obtain a plane point set.

[0026] An ellipse region is fitted in the plane point set to obtain a target ellipse region;

[0027] A center point of the target ellipse region is taken as a center, a radius is generated according to a major axis of the target ellipse region, and a circular target range is obtained.

[0028] In an optional implementation, the deleting of the center hub part feature points located in the geometric center region in the third point cloud model comprises:

[0029] A plane where each point cluster in the third point cloud model is located is fitted, and a plane normal vector is calculated;

[0030] The third point cloud model is rotated to be parallel to a specified vertical plane according to the plane normal vector;

[0031] After the third point cloud model is projected to the specified vertical plane, the center hub part feature points located in the geometric center region in the third point cloud model are deleted.

[0032] In a second aspect, a blade position angle calculation device of a fan is provided, and the device comprises:

[0033] A model acquisition module is configured to acquire a first point cloud model of a target fan; the first point cloud model comprises feature points in a target coordinate system;

[0034] A height filtering module is configured to filter the first point cloud model in the target coordinate system according to a height threshold to obtain a second point cloud model;

[0035] A plane projection module is configured to delete points in the second point cloud model that are within a target range after being projected to a horizontal plane of a target coordinate system to obtain a third point cloud model; the target range is determined according to ellipse fitting of a plane point set of the second point cloud model projected to the horizontal plane of the target coordinate system;

[0036] A hub deletion module is configured to delete center hub part feature points located in a geometric center region in the third point cloud model to obtain a fourth point cloud set;

[0037] An angle acquisition module is configured to acquire a bounding box of each point cluster in the fourth point cloud model and determine an orientation angle of the bounding box of each point cluster as a blade position angle of the target fan.

[0038] In a third aspect, an electronic device is provided, which comprises a processor and a storage medium; the storage medium stores program instructions executable by the processor; and the processor executes the program instructions to perform the blade position angle calculation method.

[0039] In a fourth aspect, a computer readable storage medium is provided, and the storage medium stores at least one instruction, which is loaded by a processor to execute the blade position angle calculation method of the fan.

[0040] In a fourth aspect, a computer readable storage medium is provided, and the storage medium stores at least one instruction, which is loaded by a processor to execute the blade position angle calculation method of the fan.

[0041] The technical scheme provided in the application can include the following beneficial effects:

[0042] After obtaining the first point cloud model of the target fan, the first point cloud model in the target coordinate system can be filtered according to the height threshold to obtain a second point cloud model. The points in the second point cloud model that are projected onto the horizontal plane of the target coordinate system and within the target range are deleted to obtain a third point cloud model. The target range is determined by elliptical fitting on the planar point set of the second point cloud model projected onto the horizontal plane of the target coordinate system. The center hub part feature points in the third point cloud model located in the geometric center region are deleted to obtain a fourth point cloud set. At this time, the fourth point cloud set has removed the feature points of the tower column and the hub part relative to the first point cloud model, and only the feature points of the blades remain. Therefore, the bounding box of each point cluster in the fourth point cloud model is obtained. At this time, the orientation angle of the bounding box of each point cluster can be directly used as the blade position angle of the target fan. The above scheme does not need to use a laser radar, but directly generates a three-dimensional model through the collected images, and filters the feature points other than the blades through each step. Finally, the orientation angle of the bounding box in which the feature points are located is directly used as the blade position angle, which improves the calculation efficiency and the timeliness of the blade position angle calculation of the fan while ensuring the accuracy of the angle recognition. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical scheme in the specific embodiments or prior art of the present application, the drawings needed in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creating any creative labor.

[0044] Figure 1 is a method flowchart of a blade position angle calculation method of a fan according to an exemplary embodiment.

[0045] Figure 2 is a method flowchart of a blade position angle calculation method of a fan according to an exemplary embodiment.

[0046] Figure 3An initial point cloud diagram obtained by a three-dimensional reconstruction method.

[0047] Figure 4 A remaining point cloud diagram obtained by initial filtering.

[0048] Figure 5 A fan part point cloud diagram that is reserved.

[0049] Figure 6 A point cloud sampling diagram below a tower column.

[0050] Figure 7 A diagram of removing a tower column part from a fan point cloud.

[0051] Figure 8 A blade part point cloud diagram.

[0052] Figure 9 A point cloud rotation diagram related to an embodiment of the present application is shown.

[0053] Figure 10 A point cloud projection diagram related to an embodiment of the present application is shown.

[0054] Figure 11 A diagram of removing a center hub part point cloud related to an embodiment of the present application is shown.

[0055] Figure 12 A diagram of calculating a blade point cloud OBB bounding box related to the present application.

[0056] Figure 13 A structure diagram of a blade position angle calculation device of a fan provided by an embodiment of the present application.

[0057] Figure 14 A structure diagram of an electronic device provided by an optional embodiment of the present application. DETAILED DESCRIPTION

[0058] Wind turbine is an important equipment in new energy industry. Its function is to convert wind energy into mechanical work, and the mechanical work drives the rotor to rotate, and finally outputs alternating current. Wind turbine is installed with dozens of meters long blades, which rotate to drive the motor inside the wind turbine to rotate, and finally convert wind energy into electrical energy. Therefore, the normal operation of the blades is crucial to the stable power output of the wind turbine. The blades work in high altitude and all-weather conditions, bear heavy load, and operate in harsh environment, such as wind, sun, rain, lightning, corrosion, etc., which are constantly eroded or affected by various media, and have a great impact on the service life of the blades. Therefore, it is necessary to regularly inspect the wind turbine blades to find abnormalities and defects for repair in time to ensure the normal operation of the generator set. At present, the mainstream inspection methods are divided into two types, one is to observe by the staff on the ground with a telescope, and the other is to make a close shot by a pilot on the ground by controlling a UAV carrying a camera to inspect. Compared with the two traditional methods, the emerging UAV automatic inspection is favored by the industry because of its convenience, speed and flexibility. When using UAV automatic inspection to inspect wind turbine blades, not only a lot of inspection time can be saved and clearer blade images can be obtained, but also the accuracy of inspection can be improved. The basic and key technology of UAV automatic inspection is the measurement and acquisition of random attitude parameters such as blade position angle when the wind turbine is stopped. For the image acquisition of the whole UAV wind turbine blade inspection, the attitude parameter measurement technology when the wind turbine is stopped ensures the accuracy of the flight path planning, which directly determines whether the UAV can fly accurately. At present, a commonly used technical route is to use laser radar to scan the 3D point cloud of the wind turbine and its surrounding scene, and then to locate and navigate the blades. Although this method has high accuracy, it takes a long time to collect and process the data in advance due to the large size of the wind turbine, which restricts the high timeliness requirement of this business.

[0059] To solve the above problems, an embodiment of the present application provides a wind turbine blade position angle calculation method. Figure 1 A method flow chart of a wind turbine blade position angle calculation method according to an example embodiment is shown. The method is applied to an electronic device, which can be a UAV or a control device remotely connected with the UAV, and the method comprises:

[0060] Step 101, acquiring a first point cloud model of a target wind turbine; the first point cloud model comprises a plurality of feature points in a target coordinate system.

[0061] In the embodiment, the first point cloud model is obtained by taking multi-view images around the target wind turbine dome / hub at equal angles by the unmanned aerial vehicle, and is reconstructed based on SFM (Structure From Motion) and MVS (Multi-View Stereo) algorithms, and the model includes various feature points (such as blade edge points, nacelle contour points, etc.) in the target coordinate system. Optionally, if the wind turbine structure parameters (tower height, blade length, etc.) are known, they can be used as prior information to assist sparse point cloud reconstruction, so as to improve the point cloud density and matching accuracy.

[0062] In step 102, the first point cloud model in the target coordinate system is filtered according to a height threshold, and a second point cloud model is obtained.

[0063] In the embodiment, based on the position relationship between the wind turbine tower and the ground, a height threshold (for example, 1 / 6 of the overall height of the wind turbine above the lowest end of the tower) is preset, and the ground and stray noise points below the threshold in the first point cloud model are removed, so as to obtain a second point cloud model containing only the remaining tower, nacelle and blade parts. Optionally, statistical filtering or conditional filtering can be performed on the original point cloud to further remove floating point noise during flight.

[0064] In step 103, points in the second point cloud model that are within a target range after being projected to a horizontal plane of the target coordinate system are deleted, and a third point cloud model is obtained; the target range is determined by elliptical fitting on a plane point set of the second point cloud model projected to the horizontal plane of the target coordinate system.

[0065] In the embodiment, in order to delete the tower part point cloud in the second point cloud model, the Y-axis coordinate of the bottom end of the second point cloud model can be selected as the starting point, and a specified height range (for example, 1 / 15 of the overall height) point cloud can be selected. At this time, the point cloud in the specified height range can be considered as a point cloud containing only the tower, and therefore the point cloud in the specified height range is projected to the XOZ plane (i.e., the horizontal plane), an ellipse is fitted to the projected point set by using the least square method, and 1.5 times the major axis of the ellipse is used as a radius threshold to remove all projected points within the tower range, and finally the third point cloud model containing only the nacelle and the blade is obtained by reverse mapping in the three-dimensional space.

[0066] In step 104, the center hub part feature points in the geometric center region of the third point cloud model are deleted, and a fourth point cloud set is obtained.

[0067] In this embodiment, first, the geometric center of the third point cloud model is calculated (such as the arithmetic mean coordinates of all points), and then a radius threshold corresponding to the hub area is set, and all points with a distance from the geometric center less than the threshold are removed to remove the hub and its attached structure point cloud, and finally a fourth point cloud set containing only three blade clusters is obtained.

[0068] In step 105, the bounding box of each point cluster in the fourth point cloud model is obtained, and the orientation angle of the bounding box of each point cluster is determined as the blade position angle of the target fan.

[0069] In this embodiment, the fourth point cloud set is respectively subjected to Euclidean clustering, and three blade clusters are identified; then OBB (Oriented Bounding Box) envelope fitting is performed on each cluster to calculate the included angle between the principal axis direction and the tower axis (vertical direction or Y-axis of the normalized world coordinate system in advance), and the angle is the shutdown position angle of the blade.

[0070] In summary, after obtaining the first point cloud model of the target fan, the first point cloud model in the target coordinate system can be filtered according to the height threshold to obtain the second point cloud model; the points in the second point cloud model that are projected onto the horizontal plane of the target coordinate system and are within the target range are deleted to obtain the third point cloud model; the target range is determined according to the ellipse fitting of the planar point set projected onto the horizontal plane of the target coordinate system; the feature points of the central hub part located in the geometric center region in the third point cloud model are deleted to obtain the fourth point cloud set; at this time, the fourth point cloud set has removed the feature points of the tower and the hub part relative to the first point cloud model, and only the feature points of the blades remain; therefore, the bounding box of each point cluster in the fourth point cloud model is obtained, and the orientation angle of the bounding box of each point cluster can be directly used as the blade position angle of the target fan. The above scheme does not need to use a laser radar, but directly generates a three-dimensional model through the collected images, and filters out the feature points other than the blades through each step, and finally directly uses the orientation angle of the bounding box where the feature points are located as the blade position angle, thereby ensuring the accuracy of angle recognition, improving the calculation efficiency, and improving the timeliness of the blade position angle calculation of the fan.

[0071] Figure 2 It is a method flowchart of a blade position angle calculation method of a fan according to an example embodiment. The method is applied to an electronic device, which can be a drone or a control device remotely connected to the drone, and the method comprises:

[0072] In step 201, a fan picture of the target fan is obtained.

[0073] Optionally, the physical parameters of the target wind turbine are acquired; based on the physical parameters of the target wind turbine, the origin of the target coordinate system and the flight trajectory of the UAV are determined; and the UAV is controlled to fly around the origin along the flight trajectory with the nose always pointing to the origin, and a specified number of wind turbine pictures are collected during the flying around.

[0074] In this embodiment, the electronic device first acquires the physical parameters of the target wind turbine (such as the tower height, hub diameter, blade length, etc.) from the pre-laid UAV flight parameter library or the ground monitoring system, and determines the origin of the target coordinate system and the optimal flight trajectory (for example, an equi-radius circular path) of the UAV based on the physical parameters. Subsequently, the UAV is controlled to fly around the origin along the trajectory, with the nose always pointing to the origin, and a specified number (for example, 18 or 24) of high-resolution wind turbine images are collected at equally divided angles.

[0075] Optionally, the shooting time and the gimbal pitch angle can be adjusted in combination with the wind turbine operating state (stopped / low-speed state) and the meteorological conditions (wind speed, visibility, etc.) to minimize image blur and obstruction.

[0076] Step 202, three-dimensional reconstruction is performed based on the wind turbine pictures of the target wind turbine to obtain the first point cloud model.

[0077] Figure 3 An initial point cloud diagram obtained by the three-dimensional reconstruction method is shown. In this embodiment, the wind turbine to be detected is stopped to an arbitrary state, the UAV is started and the program on the onboard computer is automatically run, and a three-dimensional reconstruction algorithm is used to obtain a point cloud model of the appearance of the wind turbine. First, a number of appearance structure parameter files of the known wind turbine to be detected reserved on the local storage space of the onboard computer are read, including the known fixed information such as the geographic position information of the tower column and the hub height. Based on the geographic position information of the tower column and the hub height, the UAV autonomously flies to above the wind turbine, the UAV sets an appropriate circumferential semi-diameter with the height position directly above the wind turbine as the origin, adjusts the gimbal to an appropriate angular position during the circling, and the nose always points to the origin. During the circling, the eighteen equally divided points of the circle are used as the shooting points, the UAV flies along the predetermined route to take pictures to obtain 18 wind turbine point cloud pictures at different angles with overlaps, and then a three-dimensional reconstruction method is used to model the wind turbine according to the wind turbine pictures to obtain a wind turbine point cloud model (i.e., the first point cloud model described above).

[0078] Step 203, the first point cloud model of the target wind turbine is acquired; the first point cloud model includes various feature points in the target coordinate system.

[0079] In the embodiment, the first point cloud model records all reconstructed feature points located in the target coordinate system, including the tower surface, the nacelle contour, and the blade surface, etc. The model can be directly used for subsequent spatial segmentation and geometric analysis. However, the initial generated dense point cloud of the fan appearance covers a very rich scene, but contains a lot of other scene point clouds that are not of interest, so the point cloud of the fan part is filtered based on the significant difference between the fan to be detected and the surrounding environment.

[0080] In step 204, the first point cloud model in the target coordinate system is filtered according to a height threshold to obtain a second point cloud model.

[0081] Optionally, in the embodiment of the application, the height difference of the first point cloud model in the vertical direction is obtained, and a specified proportion of the height difference is obtained as the height threshold.

[0082] In the embodiment of the application, the first point cloud model can first remove irrelevant noise points through preprocessing. Figure 3 The remaining point cloud obtained after initial filtering is shown in the figure. In addition, since the fan is the highest scene within the image range captured by the unmanned aerial vehicle, the surrounding is flat land and low plants, power lines, etc. According to this feature, the maximum and minimum values of the overall point cloud in the direction of the fan tower axis (for example, the vertical direction in the target coordinate system) are calculated, and the point cloud in the field_value region below the axis is filtered out. field_value can be set to 1 / 6 of the absolute value of the difference between the maximum value and the minimum value, that is, about 1 / 6 of the lower part of the overall height of the fan is first removed. The ground is basically removed, and only the fan and other small part of the point cloud are retained in the scene. Figure 4 The remaining point cloud obtained after initial filtering is shown in the figure.

[0083] Then, based on the characteristics that the fan point cloud part of the remaining point cloud has the largest volume and the most number of point clouds, the point clouds are Euclidean clustered, and the largest point cloud is retained. At this time, the fan point cloud is thus well retained, and other noise point clouds are completely filtered out, thereby obtaining the retained second point cloud model. Figure 5 The retained fan part point cloud is shown in the figure.

[0084] In step 205, points in the second point cloud model that are located in a target range after being projected to a horizontal plane of the target coordinate system are deleted to obtain a third point cloud model; the target range is determined according to elliptical fitting of a point set on the horizontal plane of the target coordinate system to which the second point cloud model is projected.

[0085] Optionally, taking the Y-axis coordinate at the bottom of the second point cloud model as the starting point, select a specified height range (for example, 1 / 15 of the overall height) of the point cloud and project it onto the horizontal plane of the target coordinate system to obtain a plane point set; fit the target elliptical area in the plane point set; take the center point of the target elliptical area as the center, generate a radius according to the major axis of the target elliptical area, and obtain the circular target range.

[0086] Based on the second point cloud model obtained through filtering in step 204, this application further removes the tower and nacelle, retaining only the blades. Since the wind turbine tower can be roughly viewed as a cylinder, the center of the cylinder can be found and its radius estimated, allowing it to be removed entirely.

[0087] Specifically, it is preferred to partially sample the point cloud below the tower for calculation, such as Figure 6 As shown, Figure 6 This is a schematic diagram of sampling the point cloud below the tower column. All points in this portion of the point cloud are then projected onto the (XOZ plane) to form a planar point set. An ellipse detection algorithm is then used to fit this point set, obtaining the center point position and the major and minor semi-axes of the ellipse. Furthermore, the major semi-axis of the ellipse is set to 1.5 times the radius of the cylinder, thus obtaining the end face location of the cylindrical tower column. Now that the cylinder radius, center, and axis direction are known, it can be removed from the overall wind turbine point cloud. Figure 7 Schematic diagram of removing the tower column from the wind turbine point cloud.

[0088] At this point, only the blade part and the cabin part are left in the point cloud, and there are some other small point cloud clusters that have not been removed cleanly. First calculate and retain the two parts of the point cloud with the largest number of spatial points (i.e., the blade and the cabin); record the relative spatial position relationship between the blade and the cabin. Here, the vector pointing from the geometric center of the cabin point cloud to the center of the blade point cloud can be calculated and recorded. This step will be used later to calculate the front orientation position of the blade wind wheel surface to prevent erroneous solutions caused by vectors in the positive and negative directions. In order to ensure that the tower column is removed cleanly and thoroughly, perform another Euclidean clustering of the point cloud at this time, retaining only the largest point cloud part, thereby obtaining the third point cloud model. Figure 8 This is a schematic diagram of the point cloud with only the blade part remaining.

[0089] Step 206 : Fit the plane where each point cluster in the third point cloud model is located, and calculate the plane normal vector.

[0090] Optionally, in the embodiment of the present application, firstly, all three-dimensional points in the point cluster are regarded as a group of integral data, and the middle position thereof is found, that is, the coordinates of all points are added and averaged to obtain a center point. The center point is equivalent to the balanced position after all points are converged together. Then all points are translated relative to the center point. For the translated points, the distribution of the points in three orthogonal directions (X, Y, Z) is observed, and the direction in which the points change the least is taken as the normal vector of the plane.

[0091] In step 207, the third point cloud model is rotated to be parallel to the specified vertical plane according to the plane normal vector.

[0092] In the embodiment of the present application, the plane in which the blade point cloud is located needs to be fitted, and the plane normal vector is calculated. Because the wind turbine blade part is in an arbitrary attitude position in space, it is inconvenient to process it, so the blade point cloud needs to be rotated to a certain plane position parallel to the YOZ or YOX plane (the vertical direction is generally set as the Y axis) according to the calculated plane normal vector, and finally the entire blade point cloud is projected onto the rotated plane. Figure 9 A point cloud rotation schematic diagram related to the embodiment of the present application is shown. Figure 10 A point cloud projection schematic diagram related to the embodiment of the present application is shown.

[0093] In step 208, after the third point cloud model is projected onto the specified vertical plane, the feature points of the center hub part located in the geometric center region in the third point cloud model are deleted to obtain a fourth point cloud model. Figure 11 A schematic diagram of removing part of the point cloud of the center hub part related to the embodiment of the present application is shown.

[0094] Optionally, each three-dimensional point in the third point cloud is projected onto a vertical plane (such as the YOZ plane) perpendicular to the axis of the fan tower column to obtain a group of two-dimensional projection points. The centroid of the projection points is calculated as the hub center, and then a radius threshold R on the projection plane is set. For each projection point, if the Euclidean distance thereof from the centroid C does not exceed R, it is determined that the original three-dimensional point corresponding to the projection point belongs to the hub region and is removed; the remaining points are restored to three-dimensional coordinates to constitute a fourth point cloud model containing only blade point clusters.

[0095] In step 209, after the fourth point cloud model is projected onto the specified vertical plane, the bounding box of each point cluster in the fourth point cloud model is obtained.

[0096] In the embodiment, the normalized fourth point cloud set is projected onto the target vertical plane (such as the YOZ plane), and the OBB (Oriented Bounding Box) fitting algorithm is applied to each blade cluster point set to obtain its minimum oriented bounding box, including the center, the long axis direction vector, and the box edge length. Figure 12A schematic diagram for calculating an OBB of a blade point cloud of a target wind turbine.

[0097] In step 210, the orientation angle of the OBB of each point cluster is determined as the blade position angle of the target wind turbine.

[0098] In this embodiment, the projection angle between the long axis direction vector of each oriented OBB and the tower axis (the Y axis of the world coordinate system or the aligned reference axis) is calculated, i.e., the blade shutdown position angle.

[0099] In summary, after obtaining the first point cloud model of the target wind turbine, the first point cloud model in the target coordinate system is filtered according to the height threshold to obtain a second point cloud model; the points in the second point cloud model that are projected onto the horizontal plane of the target coordinate system and within the target range are deleted to obtain a third point cloud model; the target range is determined by elliptical fitting on the planar point set projected onto the horizontal plane of the target coordinate system; the feature points of the center hub part located in the geometric center region in the third point cloud model are deleted to obtain a fourth point cloud set; at this time, the fourth point cloud set has removed the feature points of the tower and the hub part relative to the first point cloud model, and only the feature points of the blades remain; therefore, the OBB of each point cluster in the fourth point cloud model is obtained; at this time, the orientation angle of the OBB of each point cluster can be directly used as the blade position angle of the target wind turbine. The above scheme does not need to use a laser radar, but directly generates a three-dimensional model through the collected images, and filters the feature points other than the blades through each step, and finally directly uses the orientation angle of the OBB where the feature points are located as the blade position angle, thereby ensuring the accuracy of angle recognition, improving the calculation efficiency, and improving the timeliness of the blade position angle calculation of the wind turbine.

[0100] In the embodiments of the present application, a wind turbine blade position angle calculation device is also provided, which is used to implement the above embodiments and preferred embodiments, and will not be described again. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware or a combination of software and hardware is also possible and contemplated.

[0101] The embodiment of the present application provides a wind turbine blade position angle calculation device, Figure 13 is a structural schematic diagram of a wind turbine blade position angle calculation device provided by the embodiment of the present application, and the device comprises:

[0102] The model acquisition module 1301 is configured to acquire a first point cloud model of a target wind turbine; the first point cloud model comprises feature points in a target coordinate system;

[0103] a height filtering module 1302, configured to filter the first point cloud model in the target coordinate system according to a height threshold, to obtain a second point cloud model;

[0104] a plane projection module 1303, configured to delete points in the second point cloud model that are within a target range after being projected to a horizontal plane of the target coordinate system, to obtain a third point cloud model; the target range is determined according to ellipse fitting on a set of plane points of the second point cloud model projected to the horizontal plane of the target coordinate system;

[0105] a hub deleting module 1304, configured to delete feature points of a central hub part located in a geometric center region in the third point cloud model, to obtain a fourth point cloud set;

[0106] an angle obtaining module 1305, configured to obtain a bounding box of each point cluster in the fourth point cloud model, and determine an orientation angle of the bounding box of each point cluster as a blade position angle of the target wind turbine.

[0107] In summary, after obtaining the first point cloud model of the target wind turbine, the first point cloud model in the target coordinate system is filtered according to a height threshold, to obtain a second point cloud model; points in the second point cloud model that are within a target range after being projected to a horizontal plane of the target coordinate system are deleted, to obtain a third point cloud model; the target range is determined according to ellipse fitting on a set of plane points of the second point cloud model projected to the horizontal plane of the target coordinate system; feature points of a central hub part located in a geometric center region in the third point cloud model are deleted, to obtain a fourth point cloud set; at this time, the fourth point cloud set has removed feature points of a tower column and a hub part from the first point cloud model, and only feature points of blades are left; a bounding box of each point cluster in the fourth point cloud model is obtained; at this time, an orientation angle of the bounding box of each point cluster can be directly used as a blade position angle of the target wind turbine. The above scheme does not need to use a laser radar, but directly generates a three-dimensional model through collected images, and filters feature points other than blades through each step, so that the orientation angle of the bounding box in which the feature points are located is directly used as the blade position angle, thereby guaranteeing the accuracy of angle recognition, improving the calculation efficiency, and improving the timeliness of blade position angle calculation of the wind turbine.

[0108] Further function descriptions of the above modules and units are the same as those of the above corresponding embodiments, and will not be described here.

[0109] The above apparatus is presented in the form of functional units. Here, the units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories executing one or more software or fixed programs, and / or other devices that can provide the above functions.

[0110] See also Figure 14 , Figure 14 This is a schematic diagram of the structure of an electronic device provided by an optional embodiment of the present invention. The electronic device may be a control device for a drone, such as Figure 14 As shown, the electronic device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface).

[0111] The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0112] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0113] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of an electronic device for displaying a mini-program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state drive; the memory 20 may also include a combination of the above-mentioned types of memory.

[0114] The electronic device further includes a communication interface 30 for the electronic device to communicate with other devices or a communication network.

[0115] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or be implemented as computer codes stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded through a network and stored in a local storage medium, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer codes, when the software or computer codes are accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0116] Although the embodiments of the present application are described in conjunction with the accompanying drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A method for calculating the blade position angle of a fan, characterized in that: The method comprises: Acquire a first point cloud model of a target wind turbine; the first point cloud model includes each characteristic point in a target coordinate system; filtering the first point cloud model in the target coordinate system according to a height threshold to obtain a second point cloud model; Deleting points in the second point cloud model that are within a target range after being projected onto a horizontal plane of the target coordinate system, thereby obtaining a third point cloud model; the target range is determined by performing ellipse fitting on a plane point set projected onto the horizontal plane of the target coordinate system from the second point cloud model; Deleting feature points of the central hub located in the geometric center area of ​​the third point cloud model to obtain a fourth point cloud set; A bounding box of each point cluster in the fourth point cloud model is obtained, and an orientation angle of the bounding box of each point cluster is determined as a blade position angle of the target wind turbine.

2. The method according to claim 1, characterized in that The step of obtaining a first point cloud model of a target wind turbine includes: Obtaining a wind turbine image of the target wind turbine; Perform three-dimensional reconstruction based on the wind turbine image of the target wind turbine to obtain the first point cloud model.

3. The method according to claim 2, characterized in that The step of obtaining a wind turbine image of a target wind turbine includes: Obtain the physical parameters of the target wind turbine; Determining the origin of the target coordinate system and the flight trajectory of the UAV based on the physical parameters of the target wind turbine; Obtain a specified number of wind turbine images collected by the drone during its flight along the flight trajectory with the nose of the drone facing the origin.

4. The method according to any one of claims 1 to 3, characterized in that: Before filtering the first point cloud model in the target coordinate system according to the height threshold, the method further includes: Obtaining a height difference of the first point cloud model in a vertical direction; A specified ratio of the height difference is obtained as a height threshold.

5. The method according to claim 4, characterized in that Filtering the first point cloud model in the target coordinate system according to a height threshold to obtain a second point cloud model, including: Deleting points of the first point cloud model in the target coordinate system whose heights are less than the height threshold to obtain a first candidate model; Perform Euclidean clustering on the points in the first candidate model to obtain a second point cloud model.

6. The method according to any one of claims 1 to 3, characterized in that: Before deleting the points in the second point cloud model that are projected onto the horizontal plane of the target coordinate system and are within the target range to obtain the third point cloud model, the method further includes: Taking the bottom Y-axis coordinate of the second point cloud model as the starting point, selecting point clouds within a specified height range and projecting them onto the horizontal plane of the target coordinate system to obtain a plane point set; Fitting the target elliptical area in the plane point set; Taking the center point of the target elliptical area as the center, generating a radius according to the major axis of the target elliptical area, and obtaining the circular target range.

7. The method according to any one of claims 1 to 3, characterized in that: Deleting the feature points of the central hub located in the geometric center area of ​​the third point cloud model includes: Fitting the plane where each point cluster in the third point cloud model is located, and calculating the plane normal vector; Rotating the third point cloud model to be parallel to the specified vertical plane according to the plane normal vector; After projecting the third point cloud model onto the designated vertical plane, the feature points of the central hub portion located in the geometric center area of ​​the third point cloud model are deleted.

8. A device for calculating the blade position angle of a fan, characterized in that: The device comprises: A model acquisition module is used to acquire a first point cloud model of a target wind turbine; the first point cloud model includes each characteristic point in a target coordinate system; a height filtering module, configured to filter the first point cloud model in the target coordinate system according to a height threshold to obtain a second point cloud model; a plane projection module, configured to delete points in the second point cloud model that are within a target range after being projected onto a horizontal plane of a target coordinate system, thereby obtaining a third point cloud model; the target range being determined by performing ellipse fitting on a plane point set projected onto a horizontal plane of the target coordinate system from the second point cloud model; A hub deletion module is used to delete the feature points of the central hub located in the geometric center area of ​​the third point cloud model to obtain a fourth point cloud set; The angle acquisition module is used to acquire the angle of the bounding box of each point cluster in the fourth point cloud model, and determine the orientation angle of the bounding box of each point cluster as the blade position angle of the target wind turbine.

9. An electronic device, characterized in that: The electronic device includes a processor and a storage medium, wherein the storage medium stores program instructions executable by the processor, and the processor executes the program instructions to perform the method for calculating the blade position angle of a wind turbine according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The storage medium stores at least one instruction, and the at least one instruction is loaded by the processor to execute the method for calculating the blade position angle of a wind turbine according to any one of claims 1 to 7.

Images