Thermal spraying control method

Abstract

The invention provides a thermal spraying control method, and relates to the technical field of non-traditional machining, and the method comprises the steps: obtaining an initial point cloud model of a to-be-sprayed device; identifying point cloud of a leading edge area in the initial point cloud model, and generating an initial spraying path based on point cloud fitting of the leading edge area; obtaining the thickness of a point cloud slice, and generating a plurality of curved surface point cloud slices distributed along the initial spraying path according to the thickness of the point cloud slice, each point cloud in the initial point cloud model, each point cloud in the initial spraying path and the normal vector of each point cloud, generating a key spraying path point set covering the surface of the initial point cloud model based on the intersection of each curved surface point cloud slice and the initial point cloud model; generating an optimal spraying path based on a connecting line of the key spraying path point set; and the spraying tool is controlled to spray the to-be-sprayed device according to the optimal spraying path. The method is used for rapidly planning the spraying path and ensuring the completeness of the device to be sprayed according to the spraying path when the device to be sprayed is sprayed.

Description

Technical Field

[0001] The present invention relates to the field of special processing technology, and in particular to a thermal spraying control method. Background Art

[0002] Thermal barrier coatings (TBCs), one of the three key manufacturing technologies for advanced aero-engine turbine blades, feature high-temperature resistance and excellent thermal insulation properties, which can improve the service life and reliability of turbine blades. They have become an essential thermal protection technology for advanced aero-engines and gas turbines and are of extremely important strategic significance. Typically, atmospheric plasma spraying (APS) is used to prepare thermal barrier coatings on the surfaces of aero-engine and gas turbine turbine blades. Compared to other thermal spraying technologies, APS offers the advantages of low cost and high efficiency, and is capable of coating larger and more complex turbine blades.

[0003] The thickness uniformity of thermal barrier coating affects the flow area and other aerodynamic characteristics of turbine blades, thus affecting the performance of the engine. At the same time, the spraying quality of thermal barrier coating affects its special microstructural characteristics (including the content and distribution of pores, unmelted particles, microcracks, etc.), and the microstructural characteristics of thermal barrier coating determine the special properties of thermal barrier coating (including thermal properties, mechanical properties and service life). At present, the conventional methods for preparing thermal barrier coating on turbine blades and ensuring that the prepared thermal barrier coating has good thickness uniformity and spraying quality include: pre-spraying path planning, controlling the robot to clamp the spraying equipment, controlling the variable speed drive, and so on. The positioning machine clamps the turbine blades and controls the mutual movement between the spraying equipment and the turbine blades according to the planned spraying path (including spraying angle, swing rate, spraying distance and coupling collision safety), ensuring the mutual movement relationship between the two (including spraying angle, swing rate, spraying distance and coupling collision safety). At the same time, the power, temperature and powder feeding of the spraying equipment are regulated during the movement to finally realize the preparation of the thermal barrier coating. Therefore, in order to ensure the spraying quality of the thermal barrier coating and the uniformity of the thickness of the thermal barrier coating, spray path planning is an indispensable step before spraying the thermal barrier coating on the turbine blades.

[0004] Typically, personnel use commercial offline programming software to implement spray path planning before automatically spraying thermal barrier coatings. Through multiple process verifications and spray path adjustments, the thickness of the thermal barrier coating prepared according to the spray path is ensured to be relatively evenly distributed, and the quality of the weak locations of the thermal barrier coating is guaranteed. Since there is currently no professional commercial offline programming software developed for APS for spray path planning, when using commercial offline programming software for path planning, it is necessary to first use a general programming development platform and then perform secondary development on the general programming development platform. This method is time-consuming and labor-intensive, with a long development cycle, and cannot perform rapid spray path planning. Moreover, when spraying turbine blades according to the spray path ultimately generated by this method, the integrity of the spraying cannot be guaranteed. Summary of the Invention

[0005] In view of this, the object of the present invention is to provide a thermal spraying control method that can solve the technical problems that the existing technology cannot quickly plan the spraying path and cannot ensure the spraying integrity when spraying according to the spraying path generated by the existing technology.

[0006] In order to achieve the above objectives, the technical solutions adopted in the embodiments of the present invention are as follows:

[0007] In a first aspect, an embodiment of the present invention provides a thermal spraying control method, comprising:

[0008] Acquire an initial point cloud model of the device to be sprayed; wherein the initial point cloud model is located in a point cloud coordinate system;

[0009] Identifying a point cloud of a leading edge region in the initial point cloud model, and generating an initial spraying path based on point cloud fitting of the leading edge region;

[0010] Obtaining a point cloud slice thickness, generating a plurality of surface point cloud slices distributed along the initial spray path according to the point cloud slice thickness, each point cloud in the initial point cloud model, each point cloud in the initial spray path, and a normal vector of each point cloud, and generating a key spray path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model;

[0011] Generating an optimal spraying path based on the connection of the key spraying path point set;

[0012] The spraying tool is controlled to spray the device to be sprayed according to the optimal spraying path.

[0013] Furthermore, an embodiment of the present invention provides a first possible implementation of the first aspect, wherein the step of identifying the point cloud of the leading edge area in the initial point cloud model and generating the initial spraying path based on the point cloud fitting of the leading edge area includes:

[0014] Searching the neighborhood of each point cloud in the initial point cloud model using a neighborhood search algorithm to obtain a neighborhood point cloud corresponding to each point cloud;

[0015] The curvature of each point cloud is determined based on the normal vector of each point cloud and the normal vector of each neighborhood point cloud, recorded as a first curvature, and curve fitting processing is performed on the point cloud whose first curvature is greater than a first curvature threshold to obtain the initial spraying path.

[0016] Furthermore, an embodiment of the present invention provides a second possible implementation of the first aspect, wherein the step of generating a plurality of surface point cloud slices distributed along the initial spraying path based on the thickness of the point cloud slice, each point cloud in the initial spraying path, and the normal vector of each point cloud, and generating a key spraying path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model includes:

[0017] The initial point cloud slice thickness is obtained, and the initial point cloud slice thickness is optimized and calculated based on a preset iterative algorithm to obtain the optimal point cloud slice thickness that can cover the entire surface of the device to be sprayed, and each of the surface point cloud slices is determined based on the optimal point cloud slice thickness, and the intersection of the surface point cloud slice and the initial point cloud model is used as the key spray path point set.

[0018] Furthermore, an embodiment of the present invention provides a third possible implementation of the first aspect, wherein the step of generating an optimal spray path based on the connection line of the key spray path point set includes:

[0019] Connecting points in the key spraying path point set that belong to the same surface point cloud slice to obtain multiple key spraying paths;

[0020] Each of the key spray paths is extended to both ends by a preset distance, and the extended key spray paths are connected in series to form the optimal spray path in the shape of a bow.

[0021] Furthermore, an embodiment of the present invention provides a fourth possible implementation of the first aspect, wherein the step of obtaining an initial point cloud model of the device to be sprayed includes:

[0022] Acquire a point cloud model of the device to be sprayed based on data collected by a sensor; wherein the point cloud model of the device to be sprayed is located in a sensor coordinate system;

[0023] Performing feature analysis on the point cloud model of the device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing the point cloud coordinate system based on the direction of the feature vector;

[0024] The point cloud model of the device to be sprayed is converted from the sensor coordinate system to the point cloud coordinate system to obtain an initial point cloud model.

[0025] Furthermore, an embodiment of the present invention provides a fifth possible implementation of the first aspect, wherein the step of performing feature analysis on the point cloud model of the device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing the point cloud coordinate system based on the direction of the feature vector includes:

[0026] Preprocessing each point cloud in the point cloud model of the device to be sprayed; wherein the preprocessing includes: statistical filtering processing, downsampling processing and point cloud smoothing processing;

[0027] Performing principal component analysis on the neighborhood point clouds corresponding to each of the preprocessed point clouds to obtain eigenvalues corresponding to each of the point clouds, determining each corner point in each of the point clouds based on the eigenvalues corresponding to each of the point clouds, and performing corner point processing on each of the corner points; wherein the corner point processing includes: clustering processing and corner point intensity comparison processing;

[0028] Dividing the point cloud model of the device to be sprayed into different modules based on the processed corner points;

[0029] Obtain the feature vectors of the point cloud model of the device to be sprayed after being divided into different modules, and establish a minimum bounding box and the point cloud coordinate system based on the direction represented by the feature vector; wherein the feature vector corresponds to an eigenvalue, and the direction represented by the eigenvector corresponding to the maximum eigenvalue is adjusted to align with the x-axis of the sensor coordinate system.

[0030] Furthermore, an embodiment of the present invention provides a sixth possible implementation of the first aspect, wherein the step of performing principal component analysis on the neighborhood point clouds corresponding to each of the preprocessed point clouds to obtain eigenvalues corresponding to each of the point clouds includes:

[0031] A covariance matrix corresponding to each point cloud is obtained based on the coordinates of the neighborhood point cloud corresponding to each point cloud after preprocessing, and a feature analysis is performed on the covariance matrix corresponding to each point cloud to obtain an eigenvalue.

[0032] Furthermore, an embodiment of the present invention provides a seventh possible implementation of the first aspect, wherein the step of determining each corner point in each point cloud based on the eigenvalue corresponding to each point cloud includes:

[0033] The curvature corresponding to each point cloud is calculated by using a corner detection algorithm on the eigenvalue corresponding to each point cloud to obtain the curvature corresponding to each point cloud, which is recorded as a second curvature, and each point cloud whose second curvature is greater than a second curvature threshold is regarded as a corner point.

[0034] Furthermore, an embodiment of the present invention provides an eighth possible implementation of the first aspect, wherein the step of dividing the point cloud model of the device to be sprayed into different modules based on the processed corner points includes:

[0035] Calculating the local density corresponding to each corner point based on the processed coordinates of each corner point and the coordinates of the neighborhood point cloud corresponding to each corner point;

[0036] The area where the corner points with a local density greater than a local density threshold are located is used as a first area in the point cloud model of the device to be sprayed, and the area where the corner points with a local density less than or equal to the local density threshold are located is used as a second area in the point cloud model of the device to be sprayed;

[0037] A corner point set is established by using each corner point after the area is divided as a seed point, and a point cloud in the neighborhood point cloud corresponding to each seed point that meets a preset growth condition is added to the corresponding corner point set; wherein the preset growth condition is: the distance to the corresponding seed point is less than a preset distance threshold and the angle between the normal vector of the corresponding seed point and the normal vector is less than a preset angle threshold;

[0038] The first area and the second area are divided into a plurality of modules based on the corner point set and module processing is performed; wherein the module processing includes: geometric processing, dilation processing, erosion processing, boundary smoothing processing and area screening processing.

[0039] Furthermore, an embodiment of the present invention provides a ninth possible implementation of the first aspect, wherein the step of controlling the spraying tool to spray the device to be sprayed according to the optimal spraying path includes:

[0040] Converting the optimal spraying path from the point cloud coordinate system to the world coordinate system where the device to be sprayed is located to obtain a final spraying path covering the device to be sprayed;

[0041] Determine the starting point of the final spraying path, and control the spraying tool to spray the device to be sprayed along the final spraying path from the starting point; wherein the spraying tool is clamped by a first robotic arm, and the device to be sprayed is clamped by a second robotic arm.

[0042] An embodiment of the present invention provides a thermal spraying control method, which includes: obtaining an initial point cloud model of a device to be sprayed; wherein the initial point cloud model is located in a point cloud coordinate system; identifying the point cloud of a leading edge area in the initial point cloud model, and generating an initial spraying path based on the point cloud fitting of the leading edge area; obtaining the point cloud slice thickness, generating a plurality of surface point cloud slices distributed along the initial spraying path according to the point cloud slice thickness, each point cloud in the initial spraying path, and the normal vector of each point cloud, generating a key spraying path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model; generating an optimal spraying path based on the connection of the key spraying path point set; and controlling the spraying tool to spray the device to be sprayed according to the optimal spraying path. The present invention obtains an initial point cloud model of the device to be sprayed, fits the points in the leading edge area of the initial point cloud model to obtain an initial spray path, obtains multiple surface point cloud slices based on the thickness of the point cloud slice and the point cloud information in the initial spray path, obtains the optimal spray path covering the entire initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model, and finally controls the spray tool to spray the device to be sprayed according to the optimal spray path. The thermal spray control method does not use a general programming development platform or secondary development of offline programming tools when planning the spray path. It only needs to obtain the device to be sprayed. The initial point cloud model can be used to plan the subsequent spraying path based on the initial point cloud model, which improves the planning speed of the spraying path. The optimal spraying path covering the entire surface of the initial point cloud model can be obtained through the intersection of multiple surface point cloud slices and the initial point cloud model, that is, the optimal spraying path covering the entire surface of the device to be sprayed is obtained. Finally, the spraying tool is controlled to spray the device to be sprayed according to the optimal spraying path to ensure the integrity of the spraying. The thermal spraying control method ensures that the spraying path of the device to be sprayed is quickly planned while ensuring the integrity of the spraying according to the spraying path.

[0043] Other features and advantages of the embodiments of the present invention will be described in the following description, or some features and advantages can be inferred or determined without doubt from the description, or can be learned by implementing the above-mentioned technologies of the embodiments of the present invention.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A flow chart of a thermal spraying control method provided by an embodiment of the present invention is shown;

[0047] Figure 2 A schematic diagram of a dual-manipulator model provided by an embodiment of the present invention is shown;

[0048] Figure 3 A schematic diagram of a point cloud model of a device to be sprayed after preprocessing provided by an embodiment of the present invention is shown.

[0049] icon:

[0050] 201-first robotic arm; 202-second robotic arm. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0052] At present, atmospheric plasma spraying technology is often used to spray turbine blades to prepare thermal barrier coatings. In order to ensure the uniformity and quality of the prepared thermal barrier coatings, spray path planning is usually required before spraying. Considering that it takes a long time to plan the spray path in the prior art, and the spraying integrity cannot be guaranteed when the turbine blades are sprayed according to the planned spray path, in order to improve this problem, an embodiment of the present invention provides a thermal spraying control method. This method can be applied to planning the spray path of turbine blades and ensuring the integrity of the turbine blades when spraying according to the planned spray path. The embodiment of the present invention is described in detail below.

[0053] This embodiment provides a thermal spraying control method, which can be applied to electronic equipment such as computers. Figure 1 The flowchart of a thermal spraying control method shown in FIG. 1 mainly includes the following steps:

[0054] Step 102: obtaining an initial point cloud model of the device to be sprayed; wherein the initial point cloud model is located in a point cloud coordinate system;

[0055] The above-mentioned device to be sprayed is a turbine blade, which is usually used in aircraft engines. At the same time, the turbine blade is usually multi-connected and has a complex bending and torsional structure. By obtaining the initial point cloud model of the turbine blade (the coordinates of each point cloud in the initial point cloud model are all located in the point cloud coordinate system), the initial point cloud model can better reflect the complex bending and torsional structure of the turbine blade, which is convenient for subsequent analysis.

[0056] Step 104 , identifying the point cloud of the leading edge region in the initial point cloud model, and generating an initial spraying path based on the point cloud fitting of the leading edge region;

[0057] Since the leading edge area of the turbine blade is usually directly subjected to the impact of high temperature, the point cloud of the leading edge area is fitted to generate the initial spray path, and subsequent path planning is performed while ensuring that the leading edge area is sprayed completely.

[0058] Step 106: Obtain the thickness of the point cloud slices, generate multiple surface point cloud slices distributed along the initial spray path based on the point cloud slice thickness, each point cloud in the initial point cloud model, each point cloud in the initial spray path, and the normal vector of each point cloud, and generate a key spray path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model;

[0059] In the above steps, multiple surface point cloud slices are obtained based on the surface point cloud slice solution formula:

[0060]

[0061] Among them, E is the point cloud slice thickness, P is the point cloud in the initial point cloud model, and P path (i) is the i-th point cloud on the initial spraying path, n path (i) is the normal vector of the i-th point cloud on the initial spraying path;

[0062] The obtained surface point cloud slices are parallel to each other, and the surface point cloud slices are controlled to intersect with the initial point cloud model in sequence along the first direction to obtain a key spraying path point set covering the entire surface of the initial point cloud model.

[0063] Step 108, generating an optimal spraying path based on the connection of the key spraying path point set;

[0064] By gradually connecting the key spray path point set covering the entire surface of the initial point cloud model (i.e., the turbine blade) in the order of the point cloud, the optimal spray path covering the entire surface of the turbine blade can be obtained, ensuring that even the complex structure of the turbine blade can be sprayed.

[0065] Step 110, controlling the spraying tool to spray the device to be sprayed according to the optimal spraying path;

[0066] The purpose of controlling the spraying tool to spray the spraying device is to prepare a thermal barrier coating on the surface of the turbine blade. Spraying according to the optimal spraying path ensures that the thermal barrier coating can be formed on the entire surface of the turbine blade (including complex structures).

[0067] The above-mentioned thermal spraying control method provided by the embodiment of the present invention obtains the initial point cloud model of the device to be sprayed (i.e., the turbine blade), fits the points in the leading edge area of the initial point cloud model to obtain the initial spraying path, obtains multiple surface point cloud slices based on the point cloud slice thickness and the point cloud information in the initial spraying path, obtains the optimal spraying path covering the entire initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model, and finally controls the spraying tool to spray the device to be sprayed according to the optimal spraying path. The thermal spraying control method does not use a general programming development platform when planning the spraying path, and no longer requires secondary development of offline programming tools. The thermal spraying control method can improve the planning speed of the spraying path, and the optimal spraying path covering the entire surface of the initial point cloud model can be obtained through the intersection of multiple surface point cloud slices and the initial point cloud model, that is, the optimal spraying path covering the entire surface of the device to be sprayed is obtained. Finally, the spraying tool is controlled to spray the device to be sprayed according to the optimal spraying path to ensure the integrity of the spraying. The thermal spraying control method ensures that the spraying path of the device to be sprayed is quickly planned while ensuring the integrity of the spraying according to the spraying path.

[0068] In one embodiment, this embodiment provides a method for identifying a point cloud of a leading edge region in an initial point cloud model, and generating an initial spraying path based on the point cloud fitting of the leading edge region, further comprising:

[0069] The neighborhood search algorithm is used to search the neighborhood of each point cloud in the initial point cloud model to obtain the neighborhood point cloud corresponding to each point cloud;

[0070] The above-mentioned neighborhood search algorithm includes the kd-tree algorithm, through which the neighborhood point clouds within the neighborhood radius of each point cloud in the initial point cloud model can be quickly found. The neighborhood point clouds of each point cloud can be used as the basis for judging the curvature of the point cloud; among them, the neighborhood radius range of each point cloud is 1 to 3 mm, and the preferred value is 2 mm.

[0071] The curvature of each point cloud is determined based on the normal vector of each point cloud and the normal vector of each corresponding neighboring point cloud, which is recorded as the first curvature. The point cloud whose first curvature is greater than the first curvature threshold is subjected to curve fitting processing to obtain the initial spraying path;

[0072] Since the leading edge of the turbine blade is usually an area with significant curvature changes, the curvature of each point cloud in the initial point cloud model is determined by local curvature analysis and recorded as the first curvature:

[0073]

[0074] Among them, Curvarure(p) is the first curvature of point cloud p, n p is the normal vector of point cloud p in the initial coordinate system, N is the number of neighboring point clouds corresponding to point cloud p, and p i is the neighborhood point cloud of point cloud p and i = 1, 2, 3, ..., N, is the neighborhood point cloud p i The normal vector of

[0075] The point clouds with a first curvature greater than the first curvature threshold are used as the point clouds of the leading edge area in the initial point cloud model. Normal vector analysis is performed on these point clouds to analyze their normal vector change trends. Normal vectors with consistent directions in the point clouds of the leading edge area are selected as the initial spraying path points.

[0076] The initial spraying path points are subjected to curve fitting processing based on the B-spline curve fitting method to generate the initial spraying path, and the direction of the initial spraying path is consistent with the tangent direction of each initial spraying path point, ensuring uniform coverage when spraying according to the path.

[0077] In one embodiment, this embodiment provides a method for generating multiple surface point cloud slices distributed along the initial spray path based on the thickness of the point cloud slices, each point cloud in the initial spray path, and the normal vector of each point cloud. The specific implementation method for generating a key spray path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model also includes:

[0078] Obtain the initial point cloud slice thickness, optimize and calculate the initial point cloud slice thickness based on a preset iterative algorithm, and obtain the optimal point cloud slice thickness that can cover the entire surface of the device to be sprayed. Determine the point cloud slices of each surface based on the optimal point cloud slice thickness, and use the intersection of the surface point cloud slices and the initial point cloud model as the key spray path point set;

[0079] The initial point cloud slice thickness in the above steps is usually obtained based on the diameter of the spray jet end of the spray tool, where the diameter of the spray jet end of the spray tool ranges from 10 to 25 mm, preferably 20 mm. Specifically, the initial point cloud slice thickness is set to 0.8*the diameter of the spray jet end, and multiple surface point cloud slices are obtained based on the initial point cloud slice thickness:

[0080]

[0081] Where D is the diameter of the spray jet end;

[0082] A key spraying path point set is obtained based on each surface point cloud slice obtained by the thickness of the initial point cloud slice;

[0083] Adjust the point cloud slice thickness based on the point cloud in the initial spray path:

[0084] Adjusted Thickness(i)=Initial Thickness×(1+κ1(i)),

[0085] Among them, Adjusted Thickness(i) is the adjusted point cloud slice thickness, Initial Thickness is the current point cloud slice thickness (the initial point cloud slice thickness during the first adjustment), and κ1(i) is the curvature of the i-th point cloud in the initial spraying path;

[0086] The above steps of adjusting the point cloud slice thickness are repeated through an adaptive iterative algorithm to finally obtain the optimal point cloud slice thickness to ensure uniform coverage of the key spray path points obtained subsequently. Based on the optimal point cloud slice thickness, the point cloud slices of each surface generated along the initial spray path are obtained;

[0087] Each surface point cloud slice obtained by controlling the optimal point cloud slice thickness intersects with the initial point cloud model in sequence along the first direction to obtain a key spraying path point set covering the entire surface of the initial point cloud model.

[0088] In one embodiment, the embodiment provides a specific implementation method for generating an optimal spray path based on the connection of a key spray path point set, further comprising:

[0089] Connect the points in the key spraying path point set that belong to the same surface point cloud slice to obtain multiple key spraying paths;

[0090] By connecting the key spray path points belonging to the same surface point cloud slice, multiple key spray paths are obtained. Based on multiple key spray paths, it is convenient to perform segmented analysis and subsequently integrate the key spray paths to obtain the optimal spray path.

[0091] Extend each key spray path to both ends by a preset distance, connect the extended key spray paths in series, and form an optimal bow-shaped spray path;

[0092] Obtain the unit tangent direction of the start point cloud and the end point cloud in each key spray path; the start point cloud is the first point cloud in each key spray path, and the end point cloud is the last point cloud in each key spray path;

[0093] Each key spray path is extended outward by a preset distance along the unit tangent direction of the start point cloud and the end point cloud to obtain the start extension point cloud and the end extension point cloud:

[0094] P e-start (k) = P start(k)-d·T start (k),

[0095] P e-end (k) = P end (k)-d·T end (k),

[0096] Where d is the preset distance, which can be set to twice the diameter of the spray gun jet end, P e-start (k) is the extended point cloud of the starting point of the kth key spraying path, P start (k) is the starting and ending point cloud of the kth critical spraying path, T start (k) is the unit tangent direction of the cloud of the starting and ending points of the kth critical spraying path, P e-end (k) is the end extension point cloud of the kth key spraying path, P end (k) is the end point cloud of the kth key spraying path, T end (k) is the unit tangent direction of the end point cloud of the kth critical spraying path;

[0097] An extended point cloud set is established based on the starting extended point cloud and the end extended point cloud corresponding to each key spray path. While ensuring that the robot arm moves the minimum distance during subsequent spraying, the extended point clouds at both ends of each key spray path after extending the preset distance are connected in a bow shape to form a bow-shaped optimal spray path, ensuring that the edges of the subsequent devices to be sprayed can also be fully sprayed.

[0098] In one embodiment, the specific implementation method of obtaining the initial point cloud model of the device to be sprayed provided in this embodiment also includes:

[0099] Acquire a point cloud model of the device to be sprayed based on data collected by the sensor; wherein the point cloud model of the device to be sprayed is located in the sensor coordinate system;

[0100] The above-mentioned sensor is a three-dimensional blue light scanner, which uses the three-dimensional blue light scanner to perform detailed scanning of the device to be sprayed (i.e., turbine blades) from multiple angles to obtain high-precision point cloud information of the device to be sprayed. Based on the high-precision point cloud information, a point cloud model that can accurately reflect the three-dimensional shape of the device to be sprayed is obtained, and the point cloud model of the device to be sprayed is saved in ply format; wherein, the point cloud model of the device to be sprayed is located in the sensor coordinate system, and the sensor coordinate system is a three-dimensional coordinate system established with the sensor (i.e., the three-dimensional blue light scanner) as the origin.

[0101] Performing feature analysis on the point cloud model of the device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing a point cloud coordinate system based on the direction of the feature vector;

[0102] The eigenvectors obtained in the above steps represent the main and secondary directions of the point cloud model of the device to be sprayed. Establishing the point cloud coordinate axes according to the directions represented by the eigenvectors can better reflect the characteristics and discreteness of the point cloud model of the device to be sprayed.

[0103] Convert the point cloud model of the device to be sprayed from the sensor coordinate system to the point cloud coordinate system to obtain an initial point cloud model;

[0104] Get the point cloud data of the point cloud model of the device to be sprayed in the point cloud coordinate system:

[0105]

[0106] Among them, P new is the coordinate of each point cloud in the point cloud model of the device to be sprayed in the point cloud coordinate system, P is the coordinate of each point cloud in the point cloud model of the device to be sprayed in the sensor coordinate system, is the mean value of each column of the point cloud data matrix of the point cloud model of the device to be sprayed in the sensor coordinate system, V T is the transposed matrix of the main eigenvector matrix of the point cloud model of the device to be sprayed in the sensor coordinate system;

[0107] An initial point cloud model is obtained based on the coordinates of each point cloud in the point cloud model of the device to be sprayed in the point cloud coordinate system, and the point cloud data in the initial point cloud model are subjected to translation, scaling and normalization processing. Specifically, the translation and scaling processing includes: obtaining the horizontal coordinate, vertical coordinate and vertical coordinate with the largest median value of the coordinates of each point cloud in the initial point cloud model, obtaining the horizontal coordinate, vertical coordinate and vertical coordinate with the smallest median value of the coordinates of each point cloud in the initial point cloud model, and converting each point cloud of the initial point cloud model into a preset cube range through translation and scaling operations to ensure that the range of the initial point cloud model is moderate and to avoid the coordinate values of each point cloud being too large or too small;

[0108] Normalization processing: Based on the initial point cloud model after translation and scaling, determine the horizontal coordinate scaling factor as 1 / (the maximum horizontal coordinate after translation and scaling - the minimum horizontal coordinate after translation and scaling), the vertical coordinate scaling factor as 1 / (the maximum vertical coordinate after translation and scaling - the minimum vertical coordinate after translation and scaling), and the vertical coordinate scaling factor as 1 / (the maximum vertical coordinate after translation and scaling - the minimum vertical coordinate after translation and scaling). First, subtract the minimum value of the horizontal, vertical, and vertical coordinates after translation and scaling from the horizontal, vertical, and vertical coordinates of each point cloud, and then multiply them by the corresponding horizontal, vertical, and vertical coordinate scaling factors, so that the coordinate values of each point cloud are based between 0 and 1, which is convenient for subsequent calculation and analysis.

[0109] In one embodiment, this embodiment provides a method for performing feature analysis on a point cloud model of a device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing a point cloud coordinate system based on the direction of the feature vector. The specific implementation method also includes:

[0110] Preprocessing each point cloud in the point cloud model of the spraying device; wherein the preprocessing includes: statistical filtering processing, downsampling processing and point cloud smoothing processing;

[0111] The point cloud model of the device to be sprayed is input into the Point Cloud Library (PCL) software and each point cloud in the point cloud model of the device to be sprayed is preprocessed. The preprocessing includes: statistical filtering processing, downsampling processing and point cloud smoothing processing. Specifically, the statistical filtering processing: isolated points and noise points in each point cloud are eliminated by the statistical filtering method; wherein, the distance threshold of each point cloud in the statistical filtering method is set to the average distance between each point cloud and other point clouds plus twice the standard deviation of the distance between each point cloud and other point clouds;

[0112] Downsampling: The point cloud data in the point cloud model of the spraying device is downsampled by a voxel grid filter with a preset voxel size to reduce the amount of point cloud data; the preset voxel size is 1mm 3 , preferably a cube with a length, width and height of 1 mm;

[0113] Point cloud smoothing: The surface of the point cloud model of the device to be sprayed is smoothed by the Moving Least Squares (MLS) method to maintain the recognition of the geometric features of the point cloud model of the device to be sprayed.

[0114] Performing principal component analysis on the neighborhood point clouds corresponding to each preprocessed point cloud to obtain the eigenvalues corresponding to each point cloud, determining each corner point in each point cloud based on the eigenvalues corresponding to each point cloud, and performing corner point processing on each corner point; wherein, the corner point processing includes: clustering processing and corner point intensity comparison processing;

[0115] The corresponding covariance matrix is established from the neighborhood point clouds corresponding to each preprocessed point cloud through principal component analysis, and the covariance matrix corresponding to each neighborhood point cloud is subjected to eigendecomposition to obtain the eigenvalues corresponding to each covariance matrix;

[0116] Based on the eigenvalues corresponding to each covariance matrix, the corner points in each preprocessed point cloud are judged, and each corner point is processed. The corner point processing includes clustering processing and corner point strength comparison processing. Specifically, the clustering processing: obtains the initial values of the selection radius (eps) and the minimum number of points (minPts) of the corner points, and adjusts the values of the selection radius and the minimum number of points based on the density-based spatial clustering of applications with noise (DBSCAN) to remove isolated points in the corner points; wherein the initial value range of the selection radius is between 1 and 3 mm, preferably 2 mm; the initial value range of the minimum number of points is between 3 and 7, preferably 5.

[0117] Corner point intensity comparison processing: obtain the local area window of each corner point, compare the corner point intensity of each corner point with the point cloud in the corresponding local area window, if the corner point is the point cloud with the largest corner point intensity in the local area window, then retain the corner point; if the corner point is not the point cloud with the largest corner point intensity in the local area window, then remove the corner point; by using the non-maximum suppression method to remove redundant corner points, the accuracy of the corner points is further improved; wherein, the radius range of the local area window of the corner point is 1 to 5 mm, preferably 3 mm.

[0118] Divide the point cloud model of the device to be sprayed into different modules based on the processed corner points;

[0119] The density-based clustering algorithm calculates the local density corresponding to each corner point after corner point processing. Adjusting the eps and minPts of the density clustering algorithm can ensure that the subsequent divided modules can maintain the original set structure characteristics. The point cloud model of the device to be sprayed is divided into different areas based on the spatial connectivity of the corner points and the local density of the corner points.

[0120] The segmentation algorithm based on region growing is used to segment the point cloud model divided into different regions to obtain multiple modules.

[0121] Obtaining eigenvectors of the point cloud model of the device to be sprayed after being divided into different modules, and establishing a minimum bounding box and a point cloud coordinate system based on the direction represented by the eigenvectors; wherein the eigenvectors correspond to eigenvalues, and the direction represented by the eigenvector corresponding to the maximum eigenvalue is adjusted to align with the x-axis of the sensor coordinate system;

[0122] Based on the principal component analysis method, the point cloud model of the device to be sprayed is analyzed for features, and the point cloud data matrix of the point cloud model of the device to be sprayed is established:

[0123]

[0124] Wherein, n represents the number of point clouds in the point cloud model of the device to be sprayed, and x, y, and z are the horizontal coordinate, vertical coordinate, and vertical coordinate of each point cloud in the point cloud model of the device to be sprayed;

[0125] Calculate the covariance matrix of the point cloud model of the device to be sprayed based on the point cloud data matrix:

[0126]

[0127] in, is the mean vector in the point cloud data matrix;

[0128] Perform eigenvalue decomposition on the covariance matrix C to obtain the eigenvalue λ of the point cloud model of the device to be sprayed i and the eigenvector v i :

[0129] Cv i =λ i v i ,

[0130] Eigenvector v i The direction represented represents the main direction of the point cloud, based on the eigenvector v i Establish the smallest cube that completely surrounds the point cloud model of the device to be sprayed as the minimum bounding box, and set the eigenvalue λ i The corresponding eigenvector v is the maximum i The direction represented is aligned with the x-axis of the sensor coordinate system, and finally the point cloud coordinate system is established based on the feature vector.

[0131] In one embodiment, this embodiment provides a specific implementation method of performing principal component analysis on the neighborhood point clouds corresponding to each preprocessed point cloud to obtain the eigenvalues corresponding to each point cloud, further comprising:

[0132] Based on the coordinates of the neighborhood point clouds corresponding to each preprocessed point cloud, the covariance matrix corresponding to each point cloud is obtained, and the covariance matrix corresponding to each point cloud is subjected to feature analysis to obtain the eigenvalue;

[0133] The eigenvalues obtained from the covariance matrix corresponding to each point cloud are helpful for subsequent eigenvalue analysis of each point cloud to obtain the second curvature corresponding to each point cloud.

[0134] In one embodiment, this embodiment provides a method for determining corner points in each point cloud based on the eigenvalues corresponding to each point cloud. The specific implementation method further includes:

[0135] The curvature of each point cloud is calculated by the corner detection algorithm based on the eigenvalues corresponding to each point cloud, which is recorded as the second curvature. Each point cloud whose second curvature is greater than the second curvature threshold is regarded as a corner point.

[0136] The corner detection algorithm in the above steps is the improved Harris-3D corner detection algorithm, which is specially designed for 3D point cloud models. The second curvature of each point cloud is obtained based on the ratio of the local eigenvalue λ0(i) and the global eigenvalue λ0(i)+λ1(i)+λ2(i) corresponding to each point cloud:

[0137]

[0138] Where κ2(i) is the second curvature of the i-th point cloud, λ0(i), λ1(i) and λ2(i) are the three eigenvalues corresponding to the i-th point cloud, and for each point cloud, the three eigenvalues λ0(i), λ1(i) and λ2(i) have a size relationship of λ0(i)≤λ1(i)≤λ2(i);

[0139] Get the second curvature threshold, and set the second curvature greater than the second curvature threshold κ th Each point cloud of is taken as a corner point; wherein, in order to ensure the robustness and flexibility in the process of determining the corner point, the range of the second curvature threshold is set to 0.03 to 0.1, preferably 0.05;

[0140] The Harris-3D corner detection algorithm not only relies on analyzing the local eigenvalues of the local neighborhood point cloud corresponding to the point cloud in the process of identifying corner points in the point cloud, but also considers the relative relationship of the overall eigenvalues of the local neighborhood point cloud corresponding to the point cloud. It can effectively identify point clouds with significant corner characteristics. The algorithm identifies corners by evaluating the curvature changes within the field point cloud corresponding to the point cloud, and is suitable for complex three-dimensional point cloud model analysis.

[0141] In one embodiment, this embodiment provides a specific implementation method for dividing the point cloud model of the device to be sprayed into different modules based on the processed corner points, and further includes:

[0142] Based on the processed coordinates of each corner point and the coordinates of the neighborhood point cloud corresponding to each corner point, the local density corresponding to each corner point is calculated;

[0143] Typically, the local density of corner points in the point cloud model of the device to be sprayed (i.e., turbine blade) can be used to divide it into regions. Therefore, the local density corresponding to each corner point in the point cloud model of the device to be sprayed is calculated:

[0144]

[0145] Among them, ρ a is the local density of corner point a, R a is the coordinate of corner point a in the sensor coordinate system, N a Represents the neighborhood point set formed by the neighborhood point cloud corresponding to the corner point a, b∈N aIndicates that point cloud b belongs to the neighborhood point set N a A neighborhood point cloud in R b is the coordinate of point cloud b, and σ is the parameter that controls the range of density calculation.

[0146] The area where the corner points with a local density greater than the local density threshold are located is taken as the first area in the point cloud model of the device to be sprayed, and the area where the corner points with a local density less than or equal to the local density threshold are located is taken as the second area in the point cloud model of the device to be sprayed;

[0147] The areas where the corner points with local density greater than the local density threshold are located are aggregated into the first area (i.e., the blade area) in the point cloud model of the device to be sprayed (i.e., the turbine blade), and the areas where the corner points with local density less than or equal to the local density threshold are located are aggregated into the second area (i.e., the edge plate area) in the point cloud model of the device to be sprayed (i.e., the turbine blade); wherein the range of the local density threshold is 0.8~1.2, preferably 1.

[0148] Each corner point after the division of the area is used as a seed point to establish a corner point set, and the point clouds in the neighborhood point clouds corresponding to various sub-points that meet the preset growth conditions are added to the corresponding corner point set; wherein the preset growth conditions are: the distance to the corresponding seed point is less than a preset distance threshold and the angle between the normal vector of the corresponding seed point and the normal vector is less than a preset angle threshold;

[0149] The region growing-based segmentation algorithm processes each corner point after the region is divided to establish a corner point set. Specifically, each corner point in the first region (i.e., the blade region) and the second region (i.e., the edge plate region) is used as a seed point (i.e., cluster center), and an initial set S(l) = {l} of the corner point set S(l) corresponding to each corner point is established; where l is each corner point after the region is divided;

[0150] All point clouds in the neighborhood point cloud corresponding to each corner point whose distance to the corresponding seed point is less than a preset distance threshold and whose angle with the normal vector of the corresponding seed point is less than a preset angle threshold are added to the corresponding corner point set S(l). Specifically, the distance between each point cloud in the neighborhood point cloud and the seed point is:

[0151] d(l,q)=∥R l -R q ∥,

[0152] Among them, d(l,q) is the distance between the seed point l and the corresponding point cloud q in the neighborhood point cloud, R l is the coordinate of the seed point l in the sensor coordinate system, R q is the coordinate of the neighborhood point cloud q corresponding to the seed point l in the sensor coordinate system;

[0153] The angle between the normal vector of each point cloud and the seed point in the neighborhood point cloud is:

[0154]

[0155] Among them, θ(l,q) is the angle between the normal vector of the seed point l and the corresponding point cloud q in the neighborhood point cloud;

[0156] Only when d(l,q) is less than the preset distance threshold d th And θ(l,q) is less than the preset angle threshold θ th , the corresponding neighborhood point cloud q can be added to the corresponding corner point set S(l), ensuring that the neighborhood point cloud added to the corresponding corner point set S(l) is close to the seed point in spatial distance and has a similar normal vector direction.

[0157] Based on the corner point set, the first area and the second area are divided into a plurality of modules and module processing is performed; wherein the module processing includes: geometric processing, dilation processing, erosion processing, boundary smoothing processing and area screening processing;

[0158] Based on the areas where the point clouds of each corner point are located, the first area (i.e., the blade area) and the second area (i.e., the edge plate area) are divided into multiple modules, and module processing is performed. The module processing includes geometric processing, dilation processing, erosion processing, boundary smoothing processing, and area screening processing. Specifically, geometric processing: geometric processing is performed on each module through a geometric processing algorithm so that each module maintains its original structural characteristics;

[0159] Expansion processing: Connect adjacent modules to reduce and split the small holes in the initial point cloud model;

[0160] Erosion: Erosion is performed on each module to restore the original size and shape of the initial point cloud model (i.e., turbine blade) while maintaining the connectivity obtained in the expansion process;

[0161] Smoothing: A smoothing algorithm based on curve fitting is used to smooth the boundary between the first region (i.e., the blade region) and the second region (i.e., the edge plate region). Continuous iterative optimization is performed to eliminate jagged boundaries and gradually reduce boundary irregularities.

[0162] Area screening: Identify and mark all modules smaller than a preset area threshold, remove these modules, and ultimately retain the main blades and edge panels;

[0163] The first area (i.e., the blade area) and the second area (i.e., the edge plate area) of the initial point cloud model are segmented using each set of corner points, accurately describing the local geometric structure of the area near each corner point. This step not only relies on the local density of the corner points themselves, but also takes into account the geometric distribution of these corner points in the initial point cloud model, so that each module after segmentation has relatively independent and obvious geometric structure characteristics.

[0164] In one embodiment, the embodiment provides a specific implementation method of controlling the spraying tool to spray the spraying device according to the optimal spraying path, which also includes:

[0165] The optimal spraying path is converted from the point cloud coordinate system to the world coordinate system where the device to be sprayed is located, and the final spraying path covering the device to be sprayed is obtained;

[0166] Convert each point cloud in the optimal spraying path from the point cloud coordinate system to the world coordinate system of the device to be sprayed:

[0167] P world =Q·(P bestpath -L·n)+T,

[0168] Among them, P world is the coordinate of each point cloud in the optimal point cloud path in the world coordinate system, Q is the rotation matrix used to transform each point cloud in the optimal spraying from the point cloud coordinate system to the world coordinate system where the device to be sprayed is located, P bestpath The coordinates of each point cloud in the optimal point cloud path in the point cloud coordinate system, T is the translation vector, n is the unit normal vector corresponding to each point cloud in the optimal point cloud path, and L is the preset spraying distance;

[0169] The final spraying path is obtained based on the coordinates of each point cloud in the optimal point cloud path in the world coordinate system.

[0170] Determine the starting point of the final spraying path, and control the spraying tool to spray the device to be sprayed along the final spraying path from the starting point; wherein the spraying tool is clamped by the first robotic arm, and the device to be sprayed is clamped by the second robotic arm;

[0171] See for example Figure 2 The schematic diagram of the dual-manipulator model shown determines the starting point of the final spraying path, controls the first manipulator 201 to clamp the spraying tool, controls the second manipulator 202 to clamp the device to be sprayed (i.e., the turbine blade), and controls the spraying tool through the first manipulator 201 to spray the device to be sprayed from the starting point along the final spraying path. Specifically, the first manipulator 201 and the second manipulator 202 both have six degrees of freedom, and the dual manipulator has twelve degrees of freedom. The speed of each joint of the dual manipulator changes smoothly, which effectively solves the problem of the dual spraying manipulator arm pausing when spraying different areas of the turbine blade, and further solves the problem of uneven spraying when preparing thermal barrier coatings on the surface of multi-joint turbine blades with complex bending and torsion structures.

[0172] The thermal spraying control method provided by the embodiment of the present invention obtains a point cloud model of the device to be sprayed, pre-processes the point cloud model of the device to be sprayed to reduce the number of point clouds, improves the efficiency of subsequent analysis, and enables the point cloud model of the device to be sprayed to maintain the recognition of the geometric structure. The neighborhood point cloud of each point cloud in the point cloud model of the device to be sprayed is determined by the kd-tree algorithm, and the feature analysis of each point cloud is performed by the principal component analysis method to obtain the eigenvalue corresponding to each point cloud. The second curvature of each point cloud is determined based on the eigenvalue, and the corner points in each point cloud are judged according to the second curvature, and the corner points in each point cloud are judged according to the second curvature. Corner point processing: remove isolated and redundant points in the corner points to further improve the accuracy of the corner points. Based on the density clustering algorithm, the point cloud model of the device to be sprayed is divided into different areas. A corner point set is established based on the corner points. Different areas are further divided into different modules through the corner point set to facilitate the subsequent analysis and application of the point cloud model. The modules are processed modularly to accurately determine the local geometric structure of each corner point, ensuring that each module after segmentation has relatively independent and obvious geometric structure characteristics. The point cloud model of the device to be sprayed is converted into the point cloud coordinate system to obtain the initial point cloud model.

[0173] Based on the three-dimensional features of the initial point cloud model, the point cloud of the leading edge area is identified, the initial spraying path is obtained, and uniform coverage of the spraying is ensured. The optimal slice thickness is obtained based on the adaptive iterative algorithm. The surface point cloud slice is obtained based on the optimal slice thickness. The key spraying path points covering the surface of the initial point cloud model are obtained by intersecting the surface point cloud slice with the initial point cloud model. The key spraying path points belonging to the same surface point cloud slice are connected to obtain multiple key spraying paths, and the key spraying paths are extended to ensure that the edges of the device to be sprayed can also be sprayed. The extended key spraying paths are connected to obtain the optimal spraying path in the shape of a bow. The optimal spraying path is converted to the world coordinate system where the device to be sprayed exists to obtain the final spraying path, and the spraying tool is controlled to spray the device to be sprayed along the final spraying path. Through the dual-manipulator spraying strategy, the point cloud model of the device to be sprayed is used to represent the By using the three-dimensional model as input, the accuracy and efficiency of spray path planning are improved, and through subsequent processing, when spraying the devices to be sprayed (i.e., multi-unit turbine blades) with different surface features according to the final spray path, the thermal barrier coating prepared by spraying can still be maintained with high precision, thereby improving the uniformity of the thickness and microstructure of the thermal barrier coating of the multi-unit turbine blades with complex free-form surface structures. In addition, the thermal spray control method does not use a general programming development platform and no longer requires secondary development of offline programming tools when planning the spray path, thereby reducing costs. It only needs to obtain the initial point cloud model of the device to be sprayed, and perform subsequent spray path planning based on the initial point cloud model, thereby improving the planning speed of the spray path, being able to quickly respond to the spraying requirements of the device to be sprayed, generating a spray path covering the entire device to be sprayed, and ensuring the integrity of the spraying of the device to be sprayed.

[0174] Based on the above embodiment, this embodiment provides an example of applying the above thermal spraying control method to spray a spraying device, which can be specifically performed with reference to the following steps:

[0175] Step S201: Control the sensor to scan the device to be sprayed, obtain a point cloud model of the device to be sprayed, and save the point cloud model of the device to be sprayed in ply format; wherein the device to be sprayed is a twin turbine blade with a free-form surface structure, and the sensor is a three-dimensional blue light scanner;

[0176] Step S203: Input the point cloud model of the twin turbine blades into the PCL software and pre-process the point cloud model of the twin turbine blades; wherein the pre-processing includes: statistical filtering, downsampling and point cloud smoothing; see Figure 3 The schematic diagram of the point cloud model of the twin turbine blades after preprocessing is shown. Preprocessing the point cloud model of the twin turbine blades will make the model smoother as a whole, increase the number of point clouds, and remove the influence of isolated points and noise points.

[0177] The kd-tree algorithm is used to perform neighborhood search on each point cloud in the pre-processed point cloud model of the twin turbine blades to obtain the neighborhood point cloud corresponding to each point cloud;

[0178] Based on the principal component analysis method, feature analysis is performed on each point cloud and its corresponding neighborhood point cloud to obtain the eigenvalues corresponding to each point cloud;

[0179] Based on the feature value corresponding to each point cloud, the curvature corresponding to each point cloud is obtained and recorded as the second curvature. Each point cloud with a second curvature greater than a second curvature threshold is regarded as a corner point, and corner point processing is performed on each corner point.

[0180] Step S205: further processing each corner point after corner point processing using a density-based clustering algorithm, and calculating the local density corresponding to each corner point based on the coordinates of each corner point and the coordinates of the neighboring point cloud corresponding to the corner point;

[0181] The area where the corner points with a local density greater than the local density threshold are located is regarded as the blade area (i.e., the first area) in the point cloud model of the twin turbine blades, and the area where the corner points with a local density less than or equal to the local density threshold are located is regarded as the edge plate area (i.e., the second area) in the point cloud model of the twin turbine blades;

[0182] Each of the corner points is used as a seed point, and a segmentation algorithm based on region growing is used to obtain a set of corner points corresponding to various sub-points. Based on each set of corner points, the point cloud model of the double turbine blade after the region is divided is divided into different modules, and each module is subjected to module processing.

[0183] Step S207, further dividing the point cloud model of the double turbine blade divided into different modules to obtain a point cloud model of a first single turbine blade and a point cloud model of a second single turbine blade;

[0184] Performing feature analysis on the point cloud model of the first single-joint turbine blade and the point cloud model of the second single-joint turbine blade based on the principal component analysis method to obtain a feature vector corresponding to the point cloud model of the first single-joint turbine blade and a feature vector corresponding to the point cloud model of the second single-joint turbine blade;

[0185] Establishing a minimum bounding box and a first point cloud coordinate system based on the direction represented by the eigenvector corresponding to the point cloud model of the first single-joint turbine blade, and establishing a minimum bounding box and a second point cloud coordinate system based on the direction represented by the eigenvector corresponding to the point cloud model of the second single-joint turbine blade;

[0186] Converting the point cloud model of the first single-joint turbine blade into a first point cloud coordinate system to obtain a first initial point cloud model, and converting the point cloud model of the second single-joint turbine blade into a second point cloud coordinate system to obtain a second initial point cloud model;

[0187] The first initial point cloud model and the second initial point cloud model are translated, scaled, and normalized.

[0188] Step S209, determining the point clouds of the leading edge region in the first initial point cloud model and the second initial point cloud model and performing curve fitting processing using a B-spline curve fitting method to obtain a first initial spraying path and a second initial spraying path;

[0189] Determine the midpoints of the first initial spray path and the second initial spray path:

[0190]

[0191] Wherein, L is the length of the first initial spraying path or the second initial spraying path, P mid is the midpoint point cloud of the first initial spraying path or the second initial spraying path, AccDist(j) is the integral cumulative distance of the first initial spraying path or the second initial spraying path from the starting point to each point, and the point cloud P of the first initial spraying path or the second initial spraying path is obtained when the integral cumulative distance of the first initial spraying path or the second initial spraying path is ≥ half of the length of the corresponding first initial spraying path or the second initial spraying path. j is the midpoint cloud.

[0192] Step S211, obtaining an initial point cloud slice thickness, obtaining a first optimal slice thickness based on the initial point cloud slice thickness and the curvature of each point cloud in the first initial spraying path through an adaptive iterative algorithm, and obtaining a second optimal slice thickness based on the initial point cloud slice thickness and the curvature of each point cloud in the second initial spraying path through an adaptive iterative algorithm;

[0193] Based on the first optimal slice thickness and each point cloud in the first initial spraying path, a plurality of curved surface point cloud slices distributed along the first initial spraying path are obtained; based on the second optimal slice thickness and each point cloud in the second initial spraying path, a plurality of curved surface point cloud slices distributed along the second initial spraying path are obtained;

[0194] Based on the intersection of multiple surface point cloud slices distributed along the first initial spray path and the first initial point cloud model, a first key spray path point set covering the surface of the first initial point cloud model is obtained. Based on the intersection of multiple surface point cloud slices distributed along the second initial spray path and the second initial point cloud model, a second key spray path point set covering the surface of the second initial point cloud model is obtained.

[0195] Step S213: connecting the points in the first key spray path point set that belong to the same surface point cloud slice to obtain multiple key spray paths covering the first initial point cloud model; connecting the points in the second key spray path point set that belong to the same surface point cloud slice to obtain multiple key spray paths covering the second initial point cloud model;

[0196] The key spray paths covering the first initial point cloud model are extended to both ends by a preset distance and connected in series to obtain the first optimal spray path in the shape of a bow. The key spray paths covering the second initial point cloud model are extended to both ends by a preset distance and connected in series to obtain the second optimal spray path in the shape of a bow.

[0197] Step S215, converting the first optimal spraying path into the world coordinate system to obtain a first final spraying path, and converting the second optimal spraying path into the world coordinate system to obtain a second final spraying path;

[0198] Control the first robotic arm 201 to hold the spray tool (i.e., spray gun), and the second robotic arm 202 to hold the double turbine blade. Due to the complex bending and twisting structure of the turbine blade, a spray shielding area will be generated on the surface of the turbine blade. Adaptively adjust the path point located in the spray shielding area of the first single turbine blade in the first final spray path, and adaptively adjust the path point located in the spray shielding area of the second single turbine blade in the second final spray path to ensure that the entire turbine blade can be covered during the spraying process;

[0199] According to the geometric characteristics of different positions of the turbine blade, multiple pairs of starting points and end points are marked in the first final spray path and the second final spray path through the spraying strategy, and the first final spray path and the second final spray path are decomposed into multiple trajectories respectively:

[0200] Segment i ={Point j |Point j ∈[Start i , End i ]},

[0201] Among them, Segment i is the i-th track in the first final spraying path or the second final spraying path, Point j Start is the trajectory point in the i-th trajectory in the first final spraying path or the second final spraying path. i and End i are the starting point and end point of the i-th segment of the first final spraying path or the second final spraying path, respectively; wherein, when spraying the turbine blade along the first final spraying path and the second final spraying path after the multi-segmented trajectory is divided, the first robotic arm 201 can be controlled to move to complete the spraying, the second robotic arm 202 can be controlled to move to complete the spraying, and the dual robotic arms can also be controlled to move to complete the spraying;

[0202] Based on the principle of minimum movement change of the first manipulator 201 of the spraying tool, the first final spraying path and the second final spraying path after dividing the multi-segment trajectory are modified to obtain the movement path of the first manipulator 201, and the movement paths of the first manipulator 201 and the first final spraying path and the second final spraying path after dividing the multi-segment trajectory are matrix transformed to obtain the movement path of the second manipulator 202. The first manipulator 201 is controlled to clamp the spraying tool to reach the starting point of the final spraying path of the turbine blade. The spraying of the turbine blade is completed based on the final spraying path. During the spraying process, the first manipulator 201 moves along the movement path of the first manipulator 201, and the second manipulator 202 moves along the movement path of the second manipulator 202, thereby realizing the collaborative spraying of the twelve degrees of freedom of the dual manipulators. At the same time, it is necessary to add normal vector information to the first optimal spraying path and the second optimal spraying path:

[0203] PathPoint=(x,y,z,n x ,n y ,n z )

[0204] Among them, (x, y, z) is the coordinate of the path point in the first optimal spraying path or the second optimal spraying path, (n x ,n y,n z ) is the normal vector information of the coordinates of the path points in the first optimal spraying path or the second optimal spraying path, which represents the end posture of the spraying tool; by determining the end posture of the spraying tool, it is ensured that the spraying direction of the spraying tool can be perpendicular to the surface to be sprayed of the turbine blade during the spraying process, thereby ensuring the integrity of the spraying.

[0205] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0206] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A thermal spraying control method, characterized in that: include: Acquire an initial point cloud model of the device to be sprayed; wherein the initial point cloud model is located in a point cloud coordinate system; Identifying a point cloud of a leading edge region in the initial point cloud model, and generating an initial spraying path based on point cloud fitting of the leading edge region; Obtaining a point cloud slice thickness, generating a plurality of surface point cloud slices distributed along the initial spray path according to the point cloud slice thickness, each point cloud in the initial point cloud model, each point cloud in the initial spray path, and a normal vector of each point cloud, and generating a key spray path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model; Generating an optimal spraying path based on the connection of the key spraying path point set; The spraying tool is controlled to spray the device to be sprayed according to the optimal spraying path.

2. The thermal spraying control method according to claim 1, characterized in that: The step of identifying the point cloud of the leading edge area in the initial point cloud model and generating an initial spraying path based on the point cloud fitting of the leading edge area includes: Searching the neighborhood of each point cloud in the initial point cloud model using a neighborhood search algorithm to obtain a neighborhood point cloud corresponding to each point cloud; The curvature of each point cloud is determined based on the normal vector of each point cloud and the normal vector of each corresponding neighborhood point cloud, recorded as a first curvature, and curve fitting processing is performed on the point cloud whose first curvature is greater than a first curvature threshold to obtain the initial spraying path.

3. The thermal spraying control method according to claim 1, characterized in that: The step of generating a plurality of surface point cloud slices distributed along the initial spraying path according to the thickness of the point cloud slice, each point cloud in the initial spraying path, and the normal vector of each point cloud, and generating a key spraying path point set covering the surface of the initial point cloud model based on the intersection of each surface point cloud slice and the initial point cloud model, comprises: The initial point cloud slice thickness is obtained, and the initial point cloud slice thickness is optimized and calculated based on a preset iterative algorithm to obtain the optimal point cloud slice thickness that can cover the entire surface of the device to be sprayed, and each of the surface point cloud slices is determined based on the optimal point cloud slice thickness, and the intersection of the surface point cloud slice and the initial point cloud model is used as the key spray path point set.

4. The thermal spraying control method according to claim 1, characterized in that: The step of generating an optimal spraying path based on the connection line of the key spraying path point set includes: Connecting points in the key spraying path point set that belong to the same surface point cloud slice to obtain multiple key spraying paths; Each of the key spray paths is extended to both ends by a preset distance, and the extended key spray paths are connected in series to form the optimal spray path in the shape of a bow.

5. The thermal spraying control method according to claim 2, characterized in that: The step of obtaining an initial point cloud model of the device to be sprayed includes: Acquire a point cloud model of the device to be sprayed based on data collected by a sensor; wherein the point cloud model of the device to be sprayed is located in a sensor coordinate system; Performing feature analysis on the point cloud model of the device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing the point cloud coordinate system based on the direction of the feature vector; The point cloud model of the device to be sprayed is converted from the sensor coordinate system to the point cloud coordinate system to obtain an initial point cloud model.

6. The thermal spraying control method according to claim 5, characterized in that: The step of performing feature analysis on the point cloud model of the device to be sprayed to obtain a feature vector of the point cloud model of the device to be sprayed, and establishing the point cloud coordinate system based on the direction of the feature vector includes: Preprocessing each point cloud in the point cloud model of the device to be sprayed; wherein the preprocessing includes: statistical filtering processing, downsampling processing and point cloud smoothing processing; Performing principal component analysis on the neighborhood point clouds corresponding to each of the preprocessed point clouds to obtain eigenvalues corresponding to each of the point clouds, determining each corner point in each of the point clouds based on the eigenvalues corresponding to each of the point clouds, and performing corner point processing on each of the corner points; wherein the corner point processing includes: clustering processing and corner point intensity comparison processing; Dividing the point cloud model of the device to be sprayed into different modules based on the processed corner points; Obtain the feature vectors of the point cloud model of the device to be sprayed after being divided into different modules, and establish a minimum bounding box and the point cloud coordinate system based on the direction represented by the feature vector; wherein the feature vector corresponds to an eigenvalue, and the direction represented by the eigenvector corresponding to the maximum eigenvalue is adjusted to align with the x-axis of the sensor coordinate system.

7. The thermal spraying control method according to claim 6, characterized in that: The step of performing principal component analysis on the neighborhood point clouds corresponding to the preprocessed point clouds to obtain eigenvalues corresponding to the point clouds comprises: A covariance matrix corresponding to each point cloud is obtained based on the coordinates of the neighborhood point cloud corresponding to each point cloud after preprocessing, and a feature analysis is performed on the covariance matrix corresponding to each point cloud to obtain an eigenvalue.

8. The thermal spraying control method according to claim 6, characterized in that: The step of determining each corner point in each point cloud based on the eigenvalue corresponding to each point cloud comprises: The curvature corresponding to each point cloud is calculated by using a corner detection algorithm on the eigenvalue corresponding to each point cloud to obtain the curvature corresponding to each point cloud, which is recorded as a second curvature, and each point cloud whose second curvature is greater than a second curvature threshold is regarded as a corner point.

9. The thermal spraying control method according to claim 6, characterized in that: The step of dividing the point cloud model of the device to be sprayed into different modules based on the processed corner points includes: Calculating the local density corresponding to each corner point based on the processed coordinates of each corner point and the coordinates of the neighborhood point cloud corresponding to each corner point; The area where the corner points with a local density greater than a local density threshold are located is used as a first area in the point cloud model of the device to be sprayed, and the area where the corner points with a local density less than or equal to the local density threshold are located is used as a second area in the point cloud model of the device to be sprayed; A corner point set is established by using each corner point after the area is divided as a seed point, and a point cloud in the neighborhood point cloud corresponding to each seed point that meets a preset growth condition is added to the corresponding corner point set; wherein the preset growth condition is: the distance to the corresponding seed point is less than a preset distance threshold and the angle between the normal vector of the corresponding seed point and the normal vector is less than a preset angle threshold; The first area and the second area are divided into a plurality of modules based on the corner point set and module processing is performed; wherein the module processing includes: geometric processing, dilation processing, erosion processing, boundary smoothing processing and area screening processing.

10. The thermal spraying control method according to any one of claims 1 to 9, wherein the step of controlling the spraying tool to spray the device to be sprayed according to the optimal spraying path comprises: Converting the optimal spraying path from the point cloud coordinate system to the world coordinate system where the device to be sprayed is located to obtain a final spraying path covering the device to be sprayed; Determine the starting point of the final spraying path, and control the spraying tool to spray the device to be sprayed along the final spraying path from the starting point; wherein the spraying tool is clamped by a first robotic arm, and the device to be sprayed is clamped by a second robotic arm.

Images