Power plant blade viewpoint planning method and device based on three-dimensional functional partition information entropy

Abstract

The application discloses a kind of based on three-dimensional function partition information entropy power device blade viewpoint planning method and device, the method includes: to blade original point cloud preprocessing is purified point cloud, based on principal component analysis bounding box algorithm extraction target point cloud and main shaft, based on main shaft division function area constructs regional feature set, constructs fitness function iterative solution and obtains optimal viewpoint parameter;Based on optimal viewpoint parameter and normal vector are carried out blind area detection and coverage verification, dynamically update blind area until meeting termination condition, output viewpoint scanning planning parameter and color point cloud containing coverage verification information.It is extracted by principal component analysis to target point cloud and divided function area by the application, and regional feature is used to drive particle swarm optimization to solve viewpoint, combined with blind area dynamic updating to realize collaborative coverage, effectively solve the blind area problem of complex curved surface blade scanning, can be under the condition of guaranteeing high coverage, with least viewpoint, complete efficient, reliable three-dimensional scanning.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional measurement and robot vision guidance technology, and in particular to a method and apparatus for planning the viewpoint of a power unit blade based on the information entropy of three-dimensional functional partitions. Background Technology

[0002] As core aerodynamic components, the accurate measurement of the three-dimensional geometry of power plant blades, such as aero-engine blades, is crucial for performance evaluation, quality inspection, and reverse engineering. Currently, non-contact three-dimensional measurement technology based on optical scanning is widely used. However, for complex freeform surfaces like blades with thin walls, twisting characteristics, and high reflectivity, automatically planning the optimal scanning viewpoint to obtain complete and high-precision point cloud data still faces many challenges.

[0003] In existing technologies, viewpoint planning methods often rely on manual setting by operators based on their experience or selection based on discrete preset viewpoint sets. This approach is inefficient and struggles to guarantee global optimality, easily creating scanning blind spots and resulting in the loss of some surface data. Furthermore, some automated planning methods often fail to achieve efficient full-coverage scanning when dealing with objects with weak textures and sparse features, such as leaves, due to inaccurate initial region segmentation or optimization algorithms getting stuck in local optima. Summary of the Invention

[0004] The main objective of this invention is to provide a viewpoint planning method and apparatus for power plant blades based on three-dimensional functional partition information entropy, aiming to solve the technical problems of low viewpoint planning efficiency, easy generation of scanning blind spots, resulting in loss of power plant blade surface data, and inability to achieve efficient full-coverage scanning in the prior art.

[0005] To achieve the above objectives, the present invention provides a method for planning the viewpoint of a power plant blade based on three-dimensional functional partition information entropy, the method comprising the following steps: The raw point cloud data of the power unit blades is acquired, and the raw point cloud data is preprocessed to obtain the purification blade point cloud data. The point cloud data of the purification blades is processed based on the principal component analysis bounding box algorithm to extract the target point cloud data of the blades and three principal axes; Calculate the normal vector of the target point cloud data of the blade, and divide the blade functional regions based on the three principal axes to construct a set of regional features; A fitness function is constructed based on the particle swarm optimization algorithm and the set of regional features. The fitness function is then iteratively solved to obtain the optimal viewpoint parameters. Blind zone update detection is performed based on the optimal viewpoint parameters and the normal vector to obtain blind zone point cloud data, and coverage verification is performed based on the blind zone point cloud data. The blind spot point cloud is dynamically updated until the preset termination condition is met. The viewpoint scanning planning parameters are then output, and color point cloud data containing coverage verification information is generated.

[0006] Optionally, the step of acquiring the raw point cloud data of the power unit blades and preprocessing the raw point cloud data to obtain the purification blade point cloud data includes: The original point cloud data of the power unit blades is obtained, each point in the original point cloud data is traversed, and illegal points are filtered to obtain filtered point cloud data. Calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data, and perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; The number of points in the normalized point cloud data is counted. When the number of points is greater than a preset point cloud number threshold, the normalized point cloud data is randomly sampled to obtain sampled point cloud data. When the number of points is less than or equal to the preset point cloud number threshold, the normalized point cloud data is used as the sampled point cloud data. The sampled point cloud data was identified as the point cloud data of the purification blade.

[0007] Optionally, the principal component analysis bounding box algorithm is used to process the point cloud data of the purification blade to extract the target point cloud data of the blade and three principal axes, including: Calculate the covariance matrix of the point cloud data of the purification blade, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix. Based on the principal component analysis rotation matrix, the purification blade point cloud data is transformed to the principal component analysis coordinate system, and the axis alignment bounding box boundary parameters of the purification blade point cloud data are calculated in the principal component analysis coordinate system. Based on a preset expansion ratio value, the axis-aligned bounding box boundary parameters are expanded along each principal axis direction to obtain the expanded bounding box boundary parameters. Points in the purification blade point cloud data that are within the range of the extended bounding box boundary parameters are selected, and the selected points are used to construct the blade target point cloud data. The three principal axes are then associated with the blade target point cloud data and output.

[0008] Optionally, the step of calculating the normal vector of the blade target point cloud data and dividing the blade functional regions based on the three principal axes to construct a region feature set includes: The normal vector of each point in the target point cloud data of the blade is calculated based on the neighborhood search algorithm; The eigenvector corresponding to the largest eigenvalue among the three principal axes is selected as the longest principal axis direction. The projection value of each point in the blade target point cloud data is calculated on the longest principal axis direction. The length of the longest chord is determined based on the minimum and maximum values ​​among all projection values. Based on a preset truncation ratio threshold and the longest chord length, the projected value is divided into intervals, and the blade target point cloud data is divided into leading edge region point cloud, trailing edge region point cloud and blade body region point cloud. Calculate the first average normal vector of the leading edge region point cloud, the second average normal vector of the trailing edge region point cloud, and the third average normal vector of the blade region point cloud, respectively. Calculate the spatial boundary parameters of each region point cloud. Assign a first initial weight value to the leading edge region point cloud and the trailing edge region point cloud, and assign a second initial weight value to the blade region point cloud. The first average normal vector, the second average normal vector, the third average normal vector, the spatial boundary parameter, the first initial weight value, and the second initial weight value are combined to construct a regional feature set, wherein the first initial weight value is greater than the second initial weight value.

[0009] Optionally, the step of constructing a fitness function based on the particle swarm optimization algorithm and the region feature set, and iteratively solving the fitness function to obtain the optimal viewpoint parameters, includes: Initialize particle swarm parameters and randomly generate multiple particles within a preset physical constraint range. Each particle is characterized by candidate viewpoint parameters through spherical coordinate parameters, including scanning distance, azimuth angle, and zenith angle. Based on the set of regional features, the visibility of the candidate viewpoint parameters is determined, and particles with a number of visible functional regions greater than or equal to a preset threshold for the number of visible regions are selected as effective particles. The normalized coverage area and regional information entropy of each effective particle are calculated based on the set of regional features, and a fitness function is constructed based on the normalized coverage area and the regional information entropy. The fitness value of each effective particle is calculated based on the fitness function. The individual optimal solution and the global optimal solution of the swarm are updated based on the fitness value. The particle velocity and position are adjusted according to the particle swarm update rule. When the preset iteration termination condition is met, the viewpoint parameters corresponding to the current global optimal solution of the population are output as the optimal viewpoint parameters.

[0010] Optionally, the step of performing blind zone update detection based on the optimal viewpoint parameters and the normal vector to obtain blind zone point cloud data, and performing coverage verification based on the blind zone point cloud data, includes: Convert the optimal viewpoint parameters into Cartesian coordinates and calculate the line-of-sight direction vector from the viewpoint position to each point in the target point cloud data of the blade; Calculate the angle between the line-of-sight direction vector and the normal vector of each point in the blade target point cloud data to obtain a set of angle values; Based on the set of included angle values ​​and the preset coverage determination angle threshold, the target point cloud data of the blade is used to determine the blind zone, and the blind zone point cloud data is obtained. The percentage of points in the blind zone point cloud data is calculated. If the percentage of points is greater than a preset blind zone percentage threshold, an uncovered verification result is generated. If the percentage of points is less than or equal to the preset blind zone percentage threshold, a covered verification result is generated.

[0011] Optionally, the dynamic updating of the blind spot point cloud continues until a preset termination condition is met, outputting viewpoint scanning planning parameters and generating color point cloud data containing coverage verification information, including: The blind spot point cloud is dynamically updated, and the current condition is monitored to see if the preset termination condition is met. The preset termination condition is that the current blind spot ratio is not higher than the preset ratio threshold, or the current number of viewpoints reaches the preset number of viewpoints threshold. If the preset termination condition is not met, the blind spot point cloud data will be used as the coverage detection target for the next round of viewpoint planning. The regional feature set will remain unchanged, and the viewpoint planning and blind spot detection process will be repeated to generate supplementary viewpoint parameters. If the preset termination condition is met, the iteration stops, and all generated viewpoint parameters are determined as viewpoint scanning planning parameters. The blade target point cloud data is colored based on the viewpoint scanning planning parameters and coverage verification results to generate colored point cloud data containing coverage verification information.

[0012] Furthermore, to achieve the above objectives, the present invention also proposes a power plant blade viewpoint planning device based on three-dimensional functional partition information entropy, which applies the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described above. The device includes: The data processing module is used to acquire the raw point cloud data of the power unit blades and preprocess the raw point cloud data to obtain the purification blade point cloud data. The data extraction module is used to process the point cloud data of the purification leaf based on the principal component analysis bounding box algorithm, and extract the target point cloud data of the leaf and three principal axes; The region construction module is used to calculate the normal vector of the target point cloud data of the blade, and divide the functional regions of the blade based on the three principal axes to construct a set of regional features. The viewpoint optimization module is used to construct a fitness function based on the particle swarm optimization algorithm and the region feature set, and to iteratively solve the fitness function to obtain the optimal viewpoint parameters. The blind spot detection module is used to perform blind spot update detection based on the optimal viewpoint parameters and the normal vector, obtain blind spot point cloud data, and perform coverage verification based on the blind spot point cloud data. The viewpoint planning module is used to dynamically update the blind spot point cloud until the preset termination condition is met, output viewpoint scanning planning parameters, and generate color point cloud data containing coverage verification information.

[0013] Optionally, the data processing module is further configured to: acquire raw point cloud data of the power unit blades; traverse each point in the raw point cloud data and filter illegal points to obtain filtered point cloud data; calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data; perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; count the number of points in the normalized point cloud data; when the number of points is greater than a preset point cloud number threshold, perform random sampling processing on the normalized point cloud data to obtain sampled point cloud data; when the number of points is less than or equal to the preset point cloud number threshold, use the normalized point cloud data as sampled point cloud data; and determine the sampled point cloud data as the purification blade point cloud data.

[0014] Optionally, the data extraction module is further configured to calculate the covariance matrix of the purification blade point cloud data, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix; transform the purification blade point cloud data to a principal component analysis coordinate system based on the principal component analysis rotation matrix, calculate the axis-aligned bounding box boundary parameters of the purification blade point cloud data in the principal component analysis coordinate system; expand the axis-aligned bounding box boundary parameters along each principal axis direction based on a preset expansion ratio to obtain expanded bounding box boundary parameters; filter the points in the purification blade point cloud data that are located within the range of the expanded bounding box boundary parameters, construct the blade target point cloud data based on the filtered points, and associate the three principal axes with the blade target point cloud data for output.

[0015] This invention employs a principal component analysis bounding box algorithm to effectively filter out noise and background interference, accurately extracting the target point cloud of the blade. It then divides the leading edge, trailing edge, and blade functional regions based on the longest principal axis projection ratio. Combined with a differentiated weight allocation strategy, this significantly improves the segmentation accuracy and planning targeting of key regions on complex curved blade surfaces. A particle swarm optimization algorithm is driven by a weighted fitness function that integrates regional information entropy and normalized coverage area. Information entropy characterizes the uniformity of point cloud distribution to reflect information richness, while coverage area weight prioritizes the scanning range. This achieves global viewpoint optimization while meeting equipment physical constraints, effectively overcoming local optima problems and ensuring high coverage and engineering adaptability of the viewpoint scheme. Through a blind zone detection and dynamic update mechanism based on the angle between the normal vector and the line of sight, a minimum viewpoint sequence is iteratively generated. This completely eliminates scanning blind zones, ensures data integrity, and minimizes the number of scans, achieving an optimal balance between measurement efficiency and accuracy. Coverage verification and color point cloud visualization output enable quantifiable verification of the planning results, enhancing the reliability and engineering practical value of the method. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of a power unit blade viewpoint planning device based on three-dimensional functional partition information entropy in the hardware operating environment of the embodiment of the present invention. Figure 2 This is a flowchart illustrating the first embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 3 This is a flowchart illustrating the second embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 4 This is a flowchart illustrating the third embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 5 This is a flowchart illustrating the fourth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 6 This is a flowchart illustrating the fifth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 7 This is a flowchart illustrating the sixth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention. Figure 8 This is a schematic diagram of the viewpoint optimization iteration process based on the particle swarm optimization algorithm in one embodiment of the present invention; Figure 9 This is a structural block diagram of the first embodiment of the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0020] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a power plant blade viewpoint planning device based on three-dimensional functional partition information entropy, which is part of the hardware operating environment of the embodiment of the present invention.

[0021] like Figure 1 As shown, the power unit blade viewpoint planning device based on three-dimensional functional partition information entropy may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen, an input unit such as a keyboard, and may also include standard wired or wireless interfaces. The network interface 1004 may optionally include standard wired or wireless interfaces (such as Wireless-Fidelity (Wi-Fi) interfaces). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0022] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0023] like Figure 1As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a power unit blade viewpoint planning program.

[0024] exist Figure 1 In the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy can be set in the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy. The power plant blade viewpoint planning device based on three-dimensional functional partition information entropy calls the power plant blade viewpoint planning program stored in the memory 1005 through the processor 1001 and executes the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy provided in the embodiment of the present invention.

[0025] This invention provides a method for planning the viewpoint of a power plant blade based on three-dimensional functional partition information entropy, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy according to the present invention.

[0026] In this embodiment, the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy includes the following steps: Step S10: Obtain the raw point cloud data of the power unit blades and preprocess the raw point cloud data to obtain the purified blade point cloud data.

[0027] It should be understood that the executing entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a terminal electronic device capable of performing the above functions. The following description uses a power unit blade viewpoint planning device based on three-dimensional functional partition information entropy (hereinafter referred to as the planning device) as an example to illustrate this embodiment and the following embodiments.

[0028] It should be noted that raw point cloud data can be the initial set of 3D points of the blade directly acquired by a 3D scanning device without any processing, and may contain invalid data such as noise and illegal values. Cleaned blade point cloud data can be preprocessed, containing no illegal values, with uniform scale, and with a data volume suitable for computational needs.

[0029] In practical implementation, the planning equipment can collect the original point cloud data of the power unit blades through a three-dimensional optical scanning device and load a standard point cloud format file; traverse the original point cloud data and remove illegal numerical points; calculate the geometric center and maximum range of the point cloud in three-dimensional space, and normalize all point cloud coordinates to unify the scale; when the amount of point cloud data exceeds a preset threshold, perform random sampling operation to reduce the amount of data while retaining the geometric features of the blade body, and finally obtain the purified blade point cloud data.

[0030] Furthermore, in order to effectively eliminate data interference and improve data processing stability while preserving the core geometric features of the blade, step S10 may include: Step S101: Obtain the original point cloud data of the power unit blades, traverse each point in the original point cloud data, and filter out illegal points to obtain filtered point cloud data. Step S102: Calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data, and perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; Step S103: Count the number of points in the normalized point cloud data. When the number of points is greater than a preset point cloud number threshold, perform random sampling on the normalized point cloud data to obtain sampled point cloud data. Step S104: When the number of points is less than or equal to the preset point cloud number threshold, the normalized point cloud data is used as the sampled point cloud data. Step S105: The sampled point cloud data is identified as the point cloud data of the purification blade.

[0031] It should be noted that an illegal point can be an invalid coordinate point in point cloud data whose coordinate value is not numerical or infinite and has no actual geometric meaning.

[0032] In practical implementation, the data acquisition and processing process of the planning equipment is as follows: The original point cloud data of the power unit blades is obtained by reading the original point cloud file (in .pcd or .ply format). Illegal point filtering: Traverse every point in the point cloud .like If any coordinate is NaN (Not a Number) or Inf (Infinity), the point is removed from the point cloud. After implementation, the number of points in the point cloud is denoted as . .

[0033] Normalization: Calculate the axially aligned bounding box (AABB) of the filtered point cloud to obtain the minimum coordinates. and maximum coordinates .

[0034] Computational geometry center O: Calculate the maximum range : For each point Translate and scale to obtain normalized coordinates. : At this point, the point cloud is normalized into a cube approximately located at the origin and with a size of approximately 1 unit.

[0035] Random sampling: Set a threshold for the number of point clouds. .like Then random sampling is performed. A uniform random distribution is used to select points from the normalized point cloud without repetition. These points constitute the preprocessed point cloud.

[0036] Step S20: Process the point cloud data of the purification blade based on the principal component analysis bounding box algorithm to extract the target point cloud data of the blade and the three principal axes.

[0037] It should be noted that the principal component analysis bounding box algorithm in this embodiment can be a point cloud processing algorithm that extracts the principal direction of the point cloud based on principal component analysis, constructs and expands the bounding box, and accurately separates the target object from the background.

[0038] It should be noted that the three principal axes refer to the three orthogonal principal direction eigenvectors of the point cloud obtained from principal component analysis decomposition, corresponding to the core geometric directions of the blade's length, width, and thickness. The blade target point cloud data refers to the effective three-dimensional point cloud data retaining only the main structure of the blade after removing background noise.

[0039] In practical implementation, the planning equipment can calculate the centroid and covariance matrix of the purification blade point cloud data, perform eigenvalue decomposition on the covariance matrix to obtain the three orthogonal principal axes of the point cloud, transform the purification point cloud to the coordinate system corresponding to the principal axes, construct an axis-aligned bounding box and expand the boundary according to a preset ratio, filter the point cloud data located within the expanded bounding box, remove background noise and irrelevant points, obtain the target point cloud data of the blade, and simultaneously output the three principal axis parameters.

[0040] Step S30: Calculate the normal vector of the target point cloud data of the blade, and divide the blade functional region based on the three principal axes to construct a set of regional features.

[0041] It should be noted that the normal vector refers to the direction vector perpendicular to the local curved surface of the blade point cloud, used to characterize the orientation characteristics of the blade surface. The blade functional regions refer to the three core functional areas—leading edge, trailing edge, and blade body—divided according to the blade's aerodynamic structure and geometric characteristics. The region feature set refers to a structured feature data set that integrates the average normal vector, spatial boundary, and weight information of each functional region.

[0042] In some embodiments, the planning device may employ a neighborhood radius search normal vector estimation algorithm to calculate the local normal vector of each point in the blade target point cloud data; select the longest of the three principal axes as the longest chord, project the point cloud onto this direction and cut it out according to a preset ratio to divide it into three major functional regions: leading edge, trailing edge, and blade body; calculate the average normal vector, spatial boundary, and weight parameters of each region respectively, and integrate them to form a set of regional features.

[0043] Step S40: Construct a fitness function based on the particle swarm optimization algorithm and the region feature set, and iteratively solve the fitness function to obtain the optimal viewpoint parameters.

[0044] It should be noted that the fitness function is the core function used to quantitatively evaluate the quality of candidate viewpoints, integrating both regional information richness and coverage as indicators. The optimal viewpoint parameters can be spatial coordinate parameters representing the best scanning pose, and are the core output of viewpoint planning.

[0045] In some embodiments, the running parameters of the particle swarm optimization algorithm are initialized, and the set of regional features is used as the algorithm input; the regional information entropy and the normalized coverage area are fused to construct a weighted fitness function; candidate viewpoint particles are randomly generated within the preset physical constraints, and effective particles that meet the visibility requirements are selected; the particle fitness value is iteratively calculated, and the individual optimal and swarm optimal parameters are updated; after the algorithm converges, the optimal viewpoint parameters are output.

[0046] Step S50: Perform blind zone update detection based on the optimal viewpoint parameters and the normal vector to obtain blind zone point cloud data, and perform coverage verification based on the blind zone point cloud data.

[0047] It should be noted that blind spot point cloud data refers to the set of uncovered points on the blades that cannot be effectively scanned by the current optimal viewpoint.

[0048] In some embodiments, the planning device can convert the optimal viewpoint parameters into three-dimensional Cartesian coordinates, calculate the line-of-sight vector pointing from the viewpoint to the blade surface, calculate the angle between the line-of-sight vector and the blade point cloud normal vector one by one, determine the uncovered points and summarize them to form blind spot point cloud data, count the proportion of blind spot point cloud, evaluate the coverage effect of the current viewpoint, and complete the coverage verification.

[0049] Step S60: Dynamically update the blind spot point cloud until the current preset termination condition is met, output the viewpoint scanning planning parameters, and generate color point cloud data containing coverage verification information.

[0050] It should be noted that viewpoint scanning planning parameters refer to the set of pose parameters for all optimal viewpoints, which can be directly used to guide the scanning equipment in execution. Colored point cloud data can be visualized point cloud data that is colored according to functional areas and coverage status, intuitively displaying the scanning effect.

[0051] In practical implementation, the planning equipment can use the current blind spot point cloud data as the target for a new round of optimization, repeatedly execute the viewpoint optimization and blind spot detection process; determine in real time whether the preset termination conditions are met; stop iteration after the conditions are met, organize the pose parameters of all optimal viewpoints and output them; assign colors to the point cloud according to the functional area and coverage status, and generate colored point cloud data with coverage verification information.

[0052] This embodiment achieves fully automated viewpoint planning for complex curved surfaces of power unit blades, significantly improving the execution efficiency of 3D scanning and reducing the time and labor costs of manual planning. Through functional partitioning and feature mining, it effectively extracts the core features of the blade surface and accurately matches scanning requirements. Through intelligent optimization algorithms and blind zone iterative detection, it completely solves the problem of blind zones in complex surface scanning and improves the integrity of data acquisition. The overall solution balances scanning efficiency and coverage accuracy, achieving global optimization of viewpoint planning, making 3D scanning more suitable for the engineering inspection needs of precision power units such as aero-engine blades.

[0053] refer to Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy according to the present invention.

[0054] Based on the first embodiment described above, in this embodiment, step S20 further includes: Step S201: Calculate the covariance matrix of the point cloud data of the purification blade, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix.

[0055] It should be noted that the covariance matrix can be a matrix that reflects the degree of dispersion and correlation of coordinates in each dimension of the point cloud of the purification blade, and it is the core calculation carrier of principal component analysis.

[0056] Eigenvalues ​​are numerical values ​​obtained from decomposition, representing the degree of dispersion of the point cloud distribution along the corresponding eigenvector direction.

[0057] Eigenvectors are orthogonal vectors obtained from decomposition, and are the specific mathematical expression of the three principal axes of a point cloud.

[0058] The three principal axes refer to the three core geometric directions of the point cloud determined by the feature vectors, which match the direction of the blade length, width, and thickness.

[0059] Principal component analysis rotation matrix is ​​a transformation matrix composed of three eigenvectors, used for point cloud coordinate system transformation.

[0060] In its implementation, principal component analysis principal axis calculation includes: Calculate the centroid and covariance matrix C of the point cloud obtained after preprocessing.

[0061] Principal component analysis is achieved through eigenvalue decomposition of the covariance matrix. The covariance matrix C is calculated as follows: in, For the first point cloud The coordinate vector of a point, For the center of mass of the point cloud, The number of points.

[0062] Perform eigenvalue decomposition on the covariance matrix C: in, For eigenvalues, For the corresponding eigenvectors, construct the PCA rotation matrix. Where V is the eigenvector matrix; It is a diagonal matrix of eigenvalues.

[0063] Eigenvalue decomposition of the covariance matrix C yields three eigenvalues ​​( ,in ), and its corresponding unit eigenvector (i.e., the principal axis of principal component analysis). Among them... The direction in which the point cloud distribution is most dispersed (the direction along the blade length). The direction in which the distribution is most concentrated (the direction of blade thickness).

[0064] Step S202: Based on the principal component analysis rotation matrix, transform the purification blade point cloud data to the principal component analysis coordinate system, and calculate the axis-aligned bounding box boundary parameters of the purification blade point cloud data in the principal component analysis coordinate system.

[0065] It should be noted that the principal component analysis coordinate system is a dedicated coordinate system with three principal axes as coordinate axes, which perfectly matches the actual geometry of the blade.

[0066] An axis-aligned bounding box can be the smallest bounding cube in the principal component analysis coordinate system, where all sides are parallel to the principal axes.

[0067] The axis-aligned bounding box boundary parameters refer to the minimum and maximum set of three-axis coordinates that define the spatial extent of the axis-aligned bounding box.

[0068] In some embodiments, the planning equipment uses a principal component analysis rotation matrix to rotate each coordinate point of the purification blade point cloud, switching the point cloud from the original coordinate system to the principal component analysis coordinate system; iterates through all the transformed points and calculates the minimum and maximum coordinates in the three principal axis directions respectively; it integrates the minimum and maximum coordinates of the three axes to generate axis-aligned bounding box boundary parameters.

[0069] Step S203: Based on a preset expansion ratio, the axis-aligned bounding box boundary parameters are expanded along each principal axis direction to obtain the expanded bounding box boundary parameters.

[0070] It should be noted that the preset expansion scale value can be a pre-defined bounding box magnification factor, used to retain valid data at the blade edges. The expanded bounding box boundary parameters refer to the minimum and maximum coordinate parameters of the bounding box along its three axes after scaling.

[0071] In the specific implementation, the planning equipment transforms the preprocessed point cloud into the PCA coordinate system; in the PCA coordinate system, the minimum values ​​of the point cloud on the three principal axes are calculated. and maximum value This forms an initial axis-aligned bounding box; along each principal axis direction, the bounding box size is expanded by a certain proportion β=15%. The expanded boundary is: Step S204: Filter the points in the purification blade point cloud data that are within the range of the extended bounding box boundary parameters, construct the blade target point cloud data based on the filtered points, and output the three principal axes associated with the blade target point cloud data.

[0072] In the specific implementation, the planning equipment traverses each point of the purification blade point cloud one by one, and determines whether its coordinates are within the bounding box boundary range; retains the valid points within the range and removes background noise and noise outside the range; combines the retained valid points into blade target point cloud data; binds the three main axis parameters with the blade target point cloud data and outputs them synchronously to the subsequent process.

[0073] This embodiment uses a step-by-step process of covariance matrix calculation, coordinate system transformation, bounding box expansion, and point cloud filtering to accurately extract the effective main body data of the blade, completely eliminate irrelevant noise, determine the core geometric principal axis of the blade, and establish a standardized geometric analysis benchmark. This ensures that the complete structure of the blade is not lost, reduces the complexity of subsequent calculations, and improves the geometric analysis accuracy and operational stability of the overall solution.

[0074] refer to Figure 4 , Figure 4This is a flowchart illustrating the third embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention.

[0075] Based on the above embodiments, in this embodiment, step S30 further includes: Step S301: Calculate the normal vector of each point in the target point cloud data of the blade based on the neighborhood search algorithm.

[0076] In some embodiments, the planning device constructs a kd-tree spatial index structure for the blade target point cloud data to improve the efficiency of neighbor point search; sets a fixed neighborhood search radius and traverses each three-dimensional coordinate point in the blade target point cloud data one by one; with the current calculation point as the center, it searches for and obtains a set of neighboring points within the set neighborhood radius; performs principal component analysis on the set of neighboring points to extract the eigenvector corresponding to the minimum eigenvalue; and performs direction unification processing on the eigenvector to determine it as the normal vector of the current point, thus completing the calculation of the normal vectors of all points.

[0077] In the specific implementation, the calculation process of the normal vector includes: constructing a kd-tree data structure for the target point cloud to accelerate neighborhood search; for each point Q in the point cloud, calculating the normal vector within its radius... Find the nearest neighbor set within a spherical neighborhood; perform principal component analysis on the neighborhood set, and the eigenvector corresponding to the smallest eigenvalue is the normal vector of that point. And unify the direction of the normal vector.

[0078] Step S302: Select the eigenvector corresponding to the largest eigenvalue from the three principal axes as the longest principal axis direction, calculate the projection value of each point in the blade target point cloud data on the longest principal axis direction, and determine the longest chord length based on the minimum and maximum values ​​among all projection values.

[0079] It should be noted that the longest principal axis direction refers to the eigenvector direction with the highest dispersion and largest eigenvalue among the three principal axes, corresponding to the blade length principal direction. The projection value refers to the projection value of the point cloud centroid vector onto the longest principal axis direction, used to locate the position of the point along the blade length direction. The longest chord length refers to the total projection span of the target point cloud on the longest principal axis direction, representing the overall length of the blade.

[0080] In some embodiments, the eigenvalues ​​corresponding to the three principal axes are compared, and the eigenvector with the largest eigenvalue is selected as the direction of the longest principal axis. The coordinate vector of each point in the blade target point cloud data relative to the centroid of the point cloud is calculated. The coordinate vector is projected onto the direction of the longest principal axis to obtain the projection value corresponding to each point. All projection values ​​are traversed, and the minimum and maximum projection values ​​are extracted. The difference between the maximum and minimum projection values ​​is the length of the longest chord.

[0081] Step S303: Divide the projected value into intervals based on the preset truncation ratio threshold and the longest chord length, and divide the blade target point cloud data into leading edge region point cloud, trailing edge region point cloud and blade body region point cloud.

[0082] It should be noted that the preset cutoff ratio threshold can be a pre-set length cutoff ratio used to define the boundaries of the leading and trailing edges of the blade.

[0083] In some embodiments, a preset truncation ratio threshold is invoked, and the truncation length is calculated as the product of the longest chord length and the preset truncation ratio threshold. The projection value intervals are divided: points whose projection values ​​are less than or equal to the sum of the minimum projection value and the truncation length are classified as leading edge region point clouds; points whose projection values ​​are greater than or equal to the difference between the maximum projection value and the truncation length are classified as trailing edge region point clouds; points whose projection values ​​are between the above two intervals are classified as leaf body region point clouds; the point cloud classification of the three major functional regions is completed.

[0084] Step S304: Calculate the first average normal vector of the leading edge region point cloud, the second average normal vector of the trailing edge region point cloud, and the third average normal vector of the blade region point cloud, respectively; calculate the spatial boundary parameters of each region point cloud; assign a first initial weight value to the leading edge region point cloud and the trailing edge region point cloud; and assign a second initial weight value to the blade region point cloud.

[0085] It should be noted that the first average normal vector is the arithmetic mean of all normal vectors in the point cloud of the leading edge region, representing the overall surface orientation of the leading edge region. The second average normal vector is the arithmetic mean of all normal vectors in the point cloud of the trailing edge region, representing the overall surface orientation of the trailing edge region. The third average normal vector is the arithmetic mean of all normal vectors in the point cloud of the leaf blade region, representing the overall surface orientation of the leaf blade region.

[0086] It should be noted that the first initial weight value is the weight value assigned to the critical aerodynamic regions of the leading and trailing edges. The second initial weight value is the basic weight value assigned to the main body region of the blade.

[0087] Understandably, the planning equipment calculates the arithmetic mean of all normal vectors in the point clouds of the leading edge, trailing edge, and blade region to obtain the first average normal vector, the second average normal vector, and the third average normal vector. It then iterates through the point clouds of each region one by one, statistically analyzing the minimum and maximum coordinates along the three-dimensional coordinate axes to obtain the spatial boundary parameters of each region. Based on the importance of the blade's aerodynamic function, it assigns a first initial weight value to the point clouds of the leading edge and trailing edge regions and a second initial weight value to the point clouds of the blade region, thereby quantifying the overall orientation and spatial range of each functional region. By assigning weights, it highlights the scanning priority of the key leading and trailing edge regions, adapting to engineering inspection needs.

[0088] Step S305: Combine the first average normal vector, the second average normal vector, the third average normal vector, the spatial boundary parameter, the first initial weight value, and the second initial weight value to construct a regional feature set, wherein the first initial weight value is greater than the second initial weight value.

[0089] In the specific implementation, the longest principal axis of the PCA is selected as the longest chord. Based on the three PCA principal axes obtained in S202, the eigenvector corresponding to the largest eigenvalue is selected, i.e. As the longest string.

[0090] For each point in the target point cloud of the blade, calculate its vector relative to the centroid O; and calculate the projection value of this vector in the direction of the longest chord; traverse all points, find the minimum and maximum values ​​of the projection values, and obtain the length L of the longest chord.

[0091] Based on the longest chord L, 15% of its length is cut off, resulting in three main functional regions: Leading edge region: All satisfying point; Trailing edge region: All satisfy point; Leaf blade area: All satisfied point.

[0092] Simultaneously, a set of regional features is constructed, including point clouds, average normal vectors, spatial boundaries, and weights for each region.

[0093] This embodiment uses a step-by-step process, including normal vector calculation, longest principal axis determination, functional area division, feature parameter calculation, and feature set construction, to accurately extract the blade surface orientation features, achieve scientific division of functional areas, strengthen the scanning priority of key areas by setting weights, and construct structured regional features that are adapted for intelligent optimization. This provides accurate regional input for subsequent viewpoint planning and greatly improves the pertinence and effectiveness of viewpoint planning.

[0094] refer to Figure 5 , Figure 5 This is a flowchart illustrating the fourth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy according to the present invention.

[0095] Based on the above embodiments, in this embodiment, step S40 further includes: Step S401: Initialize particle swarm parameters. Randomly generate multiple particles within a preset physical constraint range. Each particle is characterized by candidate viewpoint parameters through spherical coordinate parameters, including scanning distance, azimuth angle, and zenith angle. Step S402: Based on the region feature set, perform visibility determination on the candidate viewpoint parameters, and filter out particles whose number of visible functional regions is greater than or equal to a preset visible region number threshold as effective particles; Step S403: Calculate the normalized coverage area and regional information entropy of each effective particle based on the region feature set, and construct a fitness function based on the normalized coverage area and the regional information entropy; Step S404: Calculate the fitness value of each effective particle based on the fitness function, update the individual optimal solution and the global optimal solution of the swarm based on the fitness value, and adjust the particle velocity and position according to the particle swarm update rule; Step S405: When the preset iteration termination condition is met, output the viewpoint parameters corresponding to the current global optimal solution of the population as the optimal viewpoint parameters.

[0096] In its specific implementation, the viewpoint optimization planning based on the Particle Swarm Optimization (PSO) algorithm includes the following steps: Step 1. Problem Modeling and Parameter Initialization: In a spherical coordinate system with the centroid of the target point cloud as the origin, a viewpoint Depend on definition. For scanning distance, It is the azimuth angle. Zenith angle. PSO parameters are set as follows: particle number. Maximum number of iterations Inertial weight A dynamic change strategy is adopted, with an initial value of 0.7, which decreases to 0.2 when the maximum number of iterations is reached; the learning factor... = Random numbers and The value range is [0.5, 1.0].

[0097] Step 2. First round of particle swarm optimization (planning viewpoint) ): The three functional region feature sets are used as input; for each particle (i.e., a candidate viewpoint parameter) Perform a coordinate transformation to convert the spherical coordinates to Cartesian coordinates. Simultaneously simulating from Observe the blades, calculate the angle between the line of sight and the average normal vector of each region, and determine visibility. Convert the spherical coordinate parameters of the particles to Cartesian coordinates, calculate the angle between the viewpoint and the average normal vector of each functional region. If the angle is less than 80°, the region is considered "visible"; only particles with ≥2 visible functional regions are retained as valid solutions.

[0098] Let the viewpoint position be Blade surface points The normal vector of this point is The line-of-sight vector is: Calculate the angle between the line of sight and the normal. : Count the number of visible areas and the total weight of these regions .

[0099] Normalized coverage area The calculation is based on the projected area of ​​each functional region, which is the ratio of the bounding box area of ​​the region on the XY plane to the maximum range of the point cloud.

[0100] Based on the uniformity of point cloud distribution along the height direction (Z-axis) within the visible area, the Z-axis range is divided into 256 levels, and the percentage of points in each level is statistically analyzed. Calculate the regional information entropy .

[0101] Where m is the number of partitions (e.g., 256 levels), This represents the percentage of points in the point cloud that are evenly spaced along the Z-axis.

[0102] Substitute the values ​​into the fitness function to calculate the fitness value of the particle.

[0103] The fitness function F is defined as: in, The weighting coefficient for the coverage area. This is the weighting coefficient for regional information entropy, ensuring coverage priority.

[0104] The velocities and positions of all particles are updated according to the PSO update formula.

[0105] The position (viewpoint parameter) and velocity of each particle are updated according to the following rules: in, This refers to the particle position (i.e., the viewpoint parameter). For particle velocity, For inertial weights, As a learning factor, It is a random number. This is the optimal position in the particle's history. This represents the group's historically optimal position.

[0106] Record the historical best for each particle ( ) and the global optimum of the entire population ( When it reaches or The iteration stops when the absolute value of the fitness value change is less than 0.005 for eight consecutive generations. The globally optimal solution is output as the first optimal viewpoint. .

[0107] Step 3. Blind Zone Calculation and Second Round of PSO Optimization (Planning Viewpoint) ): based on The precise parameters are used to perform detailed visibility calculations for each point in the target point cloud. This is based on all included angles. The points form a blind spot point cloud; the blind spot point cloud is used as the target point cloud for coverage detection, and the original leaf functional area division results are used as the input area feature set for PSO optimization; the above PSO optimization process is repeated to generate the optimal viewpoint until the termination condition is met.

[0108] This embodiment achieves automated global intelligent optimization of the scanning viewpoint through step-by-step processing, including parameter initialization, effective particle selection, fitness function construction, iterative optimization, and termination output. It balances coverage and information acquisition quality, improves the accuracy and efficiency of viewpoint planning, avoids the subjectivity of manual planning, and ensures that the optimal viewpoint adapts to the 3D scanning requirements of the complex curved surface of the blade.

[0109] refer to Figure 6 , Figure 6 This is a flowchart illustrating the fifth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention.

[0110] Based on the above embodiments, in this embodiment, step S50 further includes: Step S501: Convert the optimal viewpoint parameters into Cartesian coordinates and calculate the line-of-sight direction vector from the viewpoint position to each point in the target point cloud data of the blade; Step S502: Calculate the angle between the line-of-sight direction vector and the normal vector of each point in the blade target point cloud data to obtain a set of angle values; Step S503: Based on the set of included angle values ​​and the preset coverage determination angle threshold, perform blind zone determination on the blade target point cloud data to obtain blind zone point cloud data; Step S504: Calculate the percentage of points in the blind spot point cloud data. If the percentage of points is greater than a preset blind spot percentage threshold, generate an uncovered verification result. If the percentage of points is less than or equal to the preset blind spot percentage threshold, generate a covered verification result.

[0111] In its implementation, global blind spot detection and coverage verification include the following steps: First, comprehensive coverage determination: For each point in the target point cloud, calculate the angle between its line of sight and the generated viewpoint. If the normal vector of a point is an invalid value (including NaN), it is directly determined to be uncovered; otherwise, iterate through all viewpoints and calculate the angle between the line of sight direction vector and the normal vector. If there exists any viewpoint that satisfies the angle ≤ 85°, then the point is determined to be covered; points not covered by any viewpoint are determined to be the final blind zone.

[0112] Second, performance index calculation: Count the number of uncovered points And calculate the overall surface coverage C and the percentage of blind spots. .

[0113] in, This represents the total number of points in the target point cloud for the leaf.

[0114] This embodiment accurately identifies scanning blind spots and quantifies coverage effects through step-by-step processing of coordinate transformation, angle calculation, blind spot determination, and coverage verification. It achieves an objective assessment of scanning integrity, provides clear optimization targets for subsequent blind spot iteration updates, ensures that the coverage accuracy of viewpoint planning is quantifiable and verifiable, and effectively solves the problem of difficulty in determining scanning blind spots of complex curved blades.

[0115] refer to Figure 7 , Figure 7 This is a flowchart illustrating the sixth embodiment of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy of the present invention.

[0116] Based on the above embodiments, in this embodiment, step S60 further includes: Step S601: Dynamically update the blind spot point cloud and monitor whether the current condition is met. The preset termination condition is that the current blind spot ratio is not higher than the preset ratio threshold, or the current number of viewpoints reaches the preset number of viewpoints threshold.

[0117] In the specific implementation, based on the blind spot point cloud data obtained from the previous round of blind spot detection, the ratio of the number of blind spot point cloud points to the total number of target point cloud points on the blade is calculated in real time to obtain the current blind spot percentage; the total number of generated optimal viewpoints and supplementary viewpoints is calculated simultaneously to obtain the current number of viewpoints; preset percentage thresholds and preset viewpoint number thresholds are retrieved, and the current blind spot percentage and current number of viewpoints are compared with the corresponding thresholds; it is monitored whether either the "blind spot percentage is not higher than the preset percentage threshold" or the "number of viewpoints reaches the preset viewpoint number threshold" condition is met, and the preset termination condition is determined in real time.

[0118] Step S602: If the preset termination condition is not met, the blind spot point cloud data is used as the coverage detection target for the next round of viewpoint planning. The regional feature set remains unchanged, and the viewpoint planning and blind spot detection process is repeated to generate supplementary viewpoint parameters.

[0119] In the specific implementation, after determining that the preset termination condition is not met, the current blind spot point cloud data is locked as the core coverage detection target for the next round of viewpoint optimization; the previously constructed regional feature set is used without adjusting parameters such as functional partitions, average normal vectors, and weights; the particle swarm optimization viewpoint planning process is restarted, and candidate viewpoint generation, effective particle screening, fitness calculation, and optimal solution solving are completed in sequence to generate supplementary viewpoint parameters for the blind spot; based on the supplementary viewpoint parameters, blind spot detection is performed again to complete a new round of updates to the blind spot point cloud data.

[0120] Step S603: If the preset termination condition is met, stop the iteration and determine all generated viewpoint parameters as viewpoint scanning planning parameters.

[0121] In the specific implementation, once the preset termination condition is met, the entire iterative process of viewpoint planning and blind spot detection is immediately terminated; the optimal viewpoint parameters and supplementary viewpoint parameters generated in all iteration rounds are summarized, and the parameter format and data type are uniformly organized; all the integrated viewpoint parameters are formally determined as viewpoint scanning planning parameters that can directly drive the 3D scanning equipment to execute.

[0122] Step S604: Based on the viewpoint scanning planning parameters and coverage verification results, perform color processing on the blade target point cloud data to generate colored point cloud data containing coverage verification information.

[0123] In practical implementation, the planning device will generate spherical coordinates of the viewpoint. and Cartesian coordinates Write data to the text file `optimal_viewpoints.txt`; create a new color point cloud. Color each point according to its final coverage state and its original functional region.

[0124] Different colors are used to distinguish the areas: the leading edge is marked in red, the trailing edge in blue, the leaf blade in green, and uncovered points in black. Save the colored point cloud as blade_coverage.pcd; open blade_coverage.pcd using PCL Visualizer or a similar tool.

[0125] In 3D space, use clear markers (such as colored spheres) to indicate the location of each viewpoint. Print key metrics in the visualization window title bar or console.

[0126] Reference Figure 8 , Figure 8 This is a schematic diagram of the viewpoint optimization iteration process based on the particle swarm optimization algorithm in one embodiment. Taking the feature set of the three functional areas of the power unit blade as input, the initial viewpoint set is first determined, the viewpoint parameters are obtained, and the viewpoint fitness is calculated. Then, based on the visibility of the region, the effective solutions with a coverage area not less than a set threshold are selected. Next, the individual particle optimal (pbest) and the global swarm optimal (gbest) are calculated, the particle velocity and position (i.e., viewpoint parameters) are updated, and iterative optimization is carried out. After the iteration is completed, it is first determined whether the coverage requirement is met. If it is met, the optimal viewpoint is directly output to form a scanning scheme. If it is not met, it is further determined whether the maximum number of iterations or global optimal stability has been reached. If not, the process returns to redetermine the initial viewpoint set and repeats the above process. If the maximum number of iterations or global optimal stability has been reached, the optimal viewpoint is also output, and finally a complete and executable viewpoint scanning scheme is formed.

[0127] This embodiment achieves gradual reduction of scanning blind zones and global coverage optimization through step-by-step processing of blind zone dynamic iteration, termination condition monitoring, supplementary viewpoint generation, parameter integration, and visualization coloring. It achieves the highest coverage efficiency with the fewest viewpoints and outputs planning parameters that can directly drive the scanning equipment and intuitive and verifiable visualization results, taking into account the engineering practicality, coverage accuracy, and result verifiability of the solution.

[0128] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium storing a power plant blade viewpoint planning program. When the power plant blade viewpoint planning program is executed by a processor, it implements the steps of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described above.

[0129] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0130] The aforementioned computer-readable storage medium may be included in a power plant blade viewpoint planning device based on three-dimensional functional partition information entropy; or it may exist independently and not be assembled into a power plant blade viewpoint planning device based on three-dimensional functional partition information entropy.

[0131] Furthermore, this invention also proposes a computer program product, including a power plant blade viewpoint planning program, which, when executed by a processor, implements the steps of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described above.

[0132] The specific implementation of the computer program product of the present invention is basically the same as the embodiments of the above-mentioned power plant blade viewpoint planning method based on three-dimensional functional partition information entropy, and will not be repeated here.

[0133] Reference Figure 9 , Figure 9 This is a structural block diagram of the first embodiment of the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy of the present invention.

[0134] like Figure 9 As shown, the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy proposed in this embodiment of the invention includes: The data processing module 10 is used to acquire the raw point cloud data of the power unit blades and preprocess the raw point cloud data to obtain the purification blade point cloud data. Data extraction module 20 is used to process the point cloud data of the purification leaf based on the principal component analysis bounding box algorithm, and extract the target point cloud data of the leaf and three principal axes; The region construction module 30 is used to calculate the normal vector of the target point cloud data of the blade, and divide the blade functional region based on the three principal axes to construct a region feature set; The viewpoint optimization module 40 is used to construct a fitness function based on the particle swarm optimization algorithm and the region feature set, and to iteratively solve the fitness function to obtain the optimal viewpoint parameters. The blind spot detection module 50 is used to perform blind spot update detection based on the optimal viewpoint parameters and the normal vector, obtain blind spot point cloud data, and perform coverage verification based on the blind spot point cloud data. The viewpoint planning module 60 is used to dynamically update the blind spot point cloud until the current preset termination condition is met, output viewpoint scanning planning parameters, and generate color point cloud data containing coverage verification information.

[0135] Furthermore, the data processing module 10 is also used to acquire the original point cloud data of the power unit blades, traverse each point in the original point cloud data, and filter illegal points to obtain filtered point cloud data; calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data, and perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; count the number of points in the normalized point cloud data, and when the number of points is greater than a preset point cloud number threshold, perform random sampling processing on the normalized point cloud data to obtain sampled point cloud data; when the number of points is less than or equal to the preset point cloud number threshold, use the normalized point cloud data as sampled point cloud data; and determine the sampled point cloud data as the purification blade point cloud data.

[0136] Furthermore, the data extraction module 20 is also used to calculate the covariance matrix of the purification blade point cloud data, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix; based on the principal component analysis rotation matrix, transform the purification blade point cloud data to the principal component analysis coordinate system, calculate the axis-aligned bounding box boundary parameters of the purification blade point cloud data in the principal component analysis coordinate system; expand the axis-aligned bounding box boundary parameters along each principal axis direction based on a preset expansion ratio to obtain expanded bounding box boundary parameters; filter the points in the purification blade point cloud data that are located within the range of the expanded bounding box boundary parameters, construct the blade target point cloud data based on the filtered points, and output the three principal axes in association with the blade target point cloud data.

[0137] This embodiment achieves fully automated viewpoint planning for complex curved surfaces of power unit blades, significantly improving the execution efficiency of 3D scanning and reducing the time and labor costs of manual planning. Through functional partitioning and feature mining, it effectively extracts the core features of the blade surface and accurately matches scanning requirements. Through intelligent optimization algorithms and blind zone iterative detection, it completely solves the problem of blind zones in complex surface scanning and improves the integrity of data acquisition. The overall solution balances scanning efficiency and coverage accuracy, achieving global optimization of viewpoint planning, making 3D scanning more suitable for the engineering inspection needs of precision power units such as aero-engine blades.

[0138] The power plant blade viewpoint planning device based on three-dimensional functional partition information entropy provided in this application adopts the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy in the above embodiments, and can solve the technical problem of power plant blade viewpoint planning based on three-dimensional functional partition information entropy. Compared with the prior art, the beneficial effects of the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy provided in this application are the same as the beneficial effects of the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy provided in the above embodiments, and other technical features in the power plant blade viewpoint planning device based on three-dimensional functional partition information entropy are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0139] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0140] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0141] In addition, for technical details not described in detail in this embodiment, please refer to the power plant blade viewpoint planning method based on three-dimensional functional partition information entropy provided in any embodiment of the present invention, which will not be repeated here.

[0142] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0143] It should be noted that the user information (including but not limited to user device information, user personal information, user location information, user behavior information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0144] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0145] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0146] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for planning the viewpoint of a power plant blade based on the information entropy of three-dimensional functional partitioning, characterized in that, The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy includes: The raw point cloud data of the power unit blades is acquired, and the raw point cloud data is preprocessed to obtain the purification blade point cloud data. The point cloud data of the purification blades is processed based on the principal component analysis bounding box algorithm to extract the target point cloud data of the blades and three principal axes; Calculate the normal vector of the target point cloud data of the blade, and divide the blade functional regions based on the three principal axes to construct a set of regional features; A fitness function is constructed based on the particle swarm optimization algorithm and the set of regional features. The fitness function is then iteratively solved to obtain the optimal viewpoint parameters. Blind zone update detection is performed based on the optimal viewpoint parameters and the normal vector to obtain blind zone point cloud data, and coverage verification is performed based on the blind zone point cloud data. The blind spot point cloud is dynamically updated until the preset termination condition is met. The viewpoint scanning planning parameters are then output, and color point cloud data containing coverage verification information is generated.

2. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The process of acquiring raw point cloud data of the power unit blades and preprocessing the raw point cloud data to obtain purified blade point cloud data includes: The original point cloud data of the power unit blades is obtained, each point in the original point cloud data is traversed, and illegal points are filtered to obtain filtered point cloud data. Calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data, and perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; The number of points in the normalized point cloud data is counted. When the number of points is greater than a preset point cloud number threshold, the normalized point cloud data is randomly sampled to obtain sampled point cloud data. When the number of points is less than or equal to the preset point cloud number threshold, the normalized point cloud data is used as the sampled point cloud data. The sampled point cloud data was identified as the point cloud data of the purification blade.

3. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The principal component analysis bounding box algorithm is used to process the point cloud data of the purification blades, extracting the target point cloud data of the blades and three principal axes, including: Calculate the covariance matrix of the point cloud data of the purification blade, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix. Based on the principal component analysis rotation matrix, the purification blade point cloud data is transformed to the principal component analysis coordinate system, and the axis alignment bounding box boundary parameters of the purification blade point cloud data are calculated in the principal component analysis coordinate system. Based on a preset expansion ratio value, the axis-aligned bounding box boundary parameters are expanded along each principal axis direction to obtain the expanded bounding box boundary parameters. Points in the purification blade point cloud data that are within the range of the extended bounding box boundary parameters are selected, and the selected points are used to construct the blade target point cloud data. The three principal axes are then associated with the blade target point cloud data and output.

4. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The calculation of the normal vector of the blade target point cloud data, and the division of the blade functional regions based on the three principal axes to construct a region feature set, includes: The normal vector of each point in the target point cloud data of the blade is calculated based on the neighborhood search algorithm; The eigenvector corresponding to the largest eigenvalue among the three principal axes is selected as the longest principal axis direction. The projection value of each point in the blade target point cloud data is calculated on the longest principal axis direction. The length of the longest chord is determined based on the minimum and maximum values ​​among all projection values. Based on a preset truncation ratio threshold and the longest chord length, the projected value is divided into intervals, and the blade target point cloud data is divided into leading edge region point cloud, trailing edge region point cloud and blade body region point cloud. Calculate the first average normal vector of the leading edge region point cloud, the second average normal vector of the trailing edge region point cloud, and the third average normal vector of the blade region point cloud, respectively. Calculate the spatial boundary parameters of each region point cloud. Assign a first initial weight value to the leading edge region point cloud and the trailing edge region point cloud, and assign a second initial weight value to the blade region point cloud. The first average normal vector, the second average normal vector, the third average normal vector, the spatial boundary parameter, the first initial weight value, and the second initial weight value are combined to construct a regional feature set, wherein the first initial weight value is greater than the second initial weight value.

5. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The fitness function is constructed based on the particle swarm optimization algorithm and the region feature set. The fitness function is iteratively solved to obtain the optimal viewpoint parameters, including: Initialize particle swarm parameters and randomly generate multiple particles within a preset physical constraint range. Each particle is characterized by candidate viewpoint parameters through spherical coordinate parameters, including scanning distance, azimuth angle, and zenith angle. Based on the set of regional features, the visibility of the candidate viewpoint parameters is determined, and particles with a number of visible functional regions greater than or equal to a preset threshold for the number of visible regions are selected as effective particles. The normalized coverage area and regional information entropy of each effective particle are calculated based on the set of regional features, and a fitness function is constructed based on the normalized coverage area and the regional information entropy. The fitness value of each effective particle is calculated based on the fitness function. The individual optimal solution and the global optimal solution of the swarm are updated based on the fitness value. The particle velocity and position are adjusted according to the particle swarm update rule. When the preset iteration termination condition is met, the viewpoint parameters corresponding to the current global optimal solution of the population are output as the optimal viewpoint parameters.

6. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The process of performing blind zone update detection based on the optimal viewpoint parameters and the normal vector to obtain blind zone point cloud data, and performing coverage verification based on the blind zone point cloud data, includes: Convert the optimal viewpoint parameters into Cartesian coordinates and calculate the line-of-sight direction vector from the viewpoint position to each point in the blade target point cloud data; Calculate the angle between the line-of-sight direction vector and the normal vector of each point in the blade target point cloud data to obtain a set of angle values; Based on the set of included angle values ​​and the preset coverage determination angle threshold, the target point cloud data of the blade is used to determine the blind zone, and the blind zone point cloud data is obtained. The percentage of points in the blind zone point cloud data is calculated. If the percentage of points is greater than a preset blind zone percentage threshold, an uncovered verification result is generated. If the percentage of points is less than or equal to the preset blind zone percentage threshold, a covered verification result is generated.

7. The power plant blade viewpoint planning method based on three-dimensional functional partition information entropy as described in claim 1, characterized in that, The dynamic updating of the blind spot point cloud continues until a preset termination condition is met. Viewpoint scanning planning parameters are then output, generating color point cloud data containing coverage verification information, including: The blind spot point cloud is dynamically updated, and the current condition is monitored to see if the preset termination condition is met. The preset termination condition is that the current blind spot ratio is not higher than the preset ratio threshold, or the current number of viewpoints reaches the preset number of viewpoints threshold. If the preset termination condition is not met, the blind spot point cloud data will be used as the coverage detection target for the next round of viewpoint planning. The regional feature set will remain unchanged, and the viewpoint planning and blind spot detection process will be repeated to generate supplementary viewpoint parameters. If the preset termination condition is met, the iteration stops, and all generated viewpoint parameters are determined as viewpoint scanning planning parameters. The blade target point cloud data is colored based on the viewpoint scanning planning parameters and coverage verification results to generate colored point cloud data containing coverage verification information.

8. A power plant blade viewpoint planning device based on three-dimensional functional partition information entropy, characterized in that, The device includes: The data processing module is used to acquire the raw point cloud data of the power unit blades and preprocess the raw point cloud data to obtain the purification blade point cloud data. The data extraction module is used to process the point cloud data of the purification leaf based on the principal component analysis bounding box algorithm, and extract the target point cloud data of the leaf and three principal axes; The region construction module is used to calculate the normal vector of the target point cloud data of the blade, and divide the blade functional regions based on the three principal axes to construct a set of region features; The viewpoint optimization module is used to construct a fitness function based on the particle swarm optimization algorithm and the region feature set, and to iteratively solve the fitness function to obtain the optimal viewpoint parameters. The blind spot detection module is used to perform blind spot update detection based on the optimal viewpoint parameters and the normal vector, obtain blind spot point cloud data, and perform coverage verification based on the blind spot point cloud data. The viewpoint planning module is used to dynamically update the blind spot point cloud until the preset termination condition is met, output viewpoint scanning planning parameters, and generate color point cloud data containing coverage verification information.

9. The power plant blade viewpoint planning device based on three-dimensional functional partition information entropy as described in claim 8, characterized in that, The data processing module is also used to acquire the original point cloud data of the power unit blades, traverse each point in the original point cloud data, and filter illegal points to obtain filtered point cloud data. Calculate the geometric center coordinates and the maximum range value in three-dimensional space of the filtered point cloud data, and perform coordinate normalization processing on the filtered point cloud data based on the geometric center coordinates and the maximum range value in three-dimensional space to obtain normalized point cloud data; The number of points in the normalized point cloud data is counted. If the number of points is greater than a preset point cloud number threshold, the normalized point cloud data is randomly sampled to obtain sampled point cloud data. If the number of points is less than or equal to the preset point cloud number threshold, the normalized point cloud data is used as sampled point cloud data. The sampled point cloud data was identified as the point cloud data of the purification blade.

10. The power plant blade viewpoint planning device based on three-dimensional functional partition information entropy as described in claim 8, characterized in that, The data extraction module is further configured to calculate the covariance matrix of the purification leaf point cloud data, perform eigenvalue decomposition on the covariance matrix to obtain three eigenvalues ​​and three corresponding eigenvectors, use the three eigenvectors as three principal axes, and construct a principal component analysis rotation matrix; based on the principal component analysis rotation matrix, transform the purification leaf point cloud data to the principal component analysis coordinate system, and calculate the axis alignment bounding box boundary parameters of the purification leaf point cloud data in the principal component analysis coordinate system. The axis-aligned bounding box boundary parameters are expanded along each principal axis direction based on a preset expansion ratio value to obtain expanded bounding box boundary parameters; points in the purification blade point cloud data located within the range of the expanded bounding box boundary parameters are selected, and the selected points are used to construct the blade target point cloud data, and the three principal axes are associated with the blade target point cloud data for output.

Images