A method, device and system for segmenting the profile of a multi-jointed aviation blade

Abstract

The present invention discloses a method, device and system for profile segmentation of aviation multi-joint blades, which belongs to the field of aviation blade detection. The method includes: S1: constructing a KD-tree corresponding to the measuring points on the aviation multi-joint blades, traversing the KD-tree to calculate the distance density of each measuring point; S2: dividing all measuring points on the aviation multi-joint blades into multiple clusters according to the distance density and connectivity, and determining the initial blade profile based on the chord length of the measuring points of each cluster; S3: deleting the noise points and overflow points on each initial blade profile to obtain the target blade profile; S4: grouping the target blade profile sorted according to the height average value into each joint section to obtain the target blade profile. The above method deletes the noise points and overflow points to obtain multiple groups of target blade profiles in a way of first stratification and then grouping. Each group of target blade profiles has a one-to-one correspondence with the theoretical blade profile section, which can improve the segmentation accuracy and prepare for the subsequent matching evaluation work.

Description

Technical Field

[0001] The present invention belongs to the field of aviation blade detection, and more specifically, relates to a method, device and system for segmenting the profile of aviation multi-blades. Background Art

[0002] As the world's most complex industrial product with the highest technical threshold, aircraft engines have always been hailed as the jewel in the crown of industry. As the core component of aircraft engines, aircraft blades directly affect the aerodynamic performance of aircraft engines, so it is particularly important to accurately control their surface quality. With the development of technology, existing aircraft blades are more often designed as multi-blades, but there is still a lack of suitable methods for the problem of multi-blade surface segmentation.

[0003] In the process of segmenting the multi-blade multi-profile measurement points, problems such as overflow points and the mismatch between the multi-blade profile measurement points and the theoretical blade profile often lead to low segmentation accuracy. Summary of the invention

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a method, device and system for surface segmentation of a multi-joint aviation blade, thereby solving the technical problem of low segmentation accuracy of multi-joint blade multi-surface measurement points.

[0005] To achieve the above object, according to one aspect of the present invention, a method for segmenting a profile of an aviation multi-blade is provided, comprising:

[0006] S1: constructing a KD-tree corresponding to the measuring points on the aviation multi-joint blade, traversing the KD-tree to calculate the distance density of each measuring point;

[0007] S2: dividing all the measuring points on the aviation multi-joint blade into a plurality of clusters according to the distance density and connectivity, and determining an initial blade profile based on the chord length of the measuring points of each cluster;

[0008] S3: deleting noise points and overflow points on each of the initial blade profiles to obtain a target blade profile;

[0009] S4: grouping the target blade profiles sorted according to the height average values ​​into groups per cross section, so that each group of the target blade profiles obtained corresponds one-to-one to a theoretical blade profile cross section.

[0010] In one embodiment, the S1 includes:

[0011] S1: selecting a splitting node from a measuring point set based on a splitting dimension, dividing the measuring point set into two sub-measuring point sets according to the splitting node, and then recursively calculating the two sub-measuring point sets to obtain the KD-tree; the measuring point set is a set of all measuring points on the aviation multi-joint blade;

[0012] S2: searching for the N closest points corresponding to each of the measuring points based on the KD-tree and obtaining the corresponding N distance values; determining the number of distance values ​​less than a distance threshold from the N distance values ​​to obtain the distance density of each of the measuring points.

[0013] In one embodiment, the S2 includes:

[0014] For any measuring point P, multiple measuring points {P0, P1, P2, ..., P m} and marked as P density reachable, the measurement points that are P density reachable are regarded as a cluster, so that all the measurement points are divided into multiple clusters;

[0015] The clusters whose number of measurement points exceeds the threshold M are regarded as the clusters of the undetermined profile, the chord length of the measurement points of each of the clusters of the undetermined profile is calculated, and the clusters of the undetermined profile whose measurement point chord length is greater than the chord length threshold are regarded as the initial blade profile.

[0016] In one embodiment, the S3 includes:

[0017] S31: using the RANSAC method to identify and delete noise points on each of the initial blade profiles;

[0018] S32: For each of the initial blade profiles, starting from a measuring point Q, when the distance between the next node Q1 and Q is greater than a distance threshold, the passed measuring point is regarded as a measurement trajectory; the initial blade profile is traversed to obtain several measurement trajectories, and the intersection part is clipped based on the first and last coordinates of each measurement trajectory to obtain the target blade profile; wherein, the overflow point exists in the intersection part.

[0019] In one embodiment, the S31 includes:

[0020] The initial blade profile including the number of measurement points less than the threshold M is regarded as an external noise point and deleted;

[0021] The envelope curve of the initial blade profile is solved by using the rolling ball method, and the measuring points not on the envelope curve are regarded as internal noise points and deleted;

[0022] The RANSAC method is used to randomly fit the initial blade profile to the blade theoretical profile, and the noise points are removed again.

[0023] In one embodiment, the S4 includes:

[0024] The X target blade profiles sorted according to the average height are grouped according to each cross section to obtain Y groups of target blade profiles;

[0025] in, Each group of target blade profiles corresponds one-to-one to a theoretical blade profile cross section.

[0026] In one embodiment, after S2 and before S3, the method further includes:

[0027] The initial blade profile is sorted from small to large according to the average height of each measuring point, and the sorted initial blade profile satisfies H(C j ) is the average height of the initial blade profile with serial number j.

[0028] In one embodiment, before S4, the method further includes:

[0029] The target blade profile is sorted from small to large according to the average height of each measuring point, and the sorted target blade profile satisfies H'(C j ) is the average height of the target blade profile with serial number j.

[0030] According to another aspect of the present invention, there is provided a profile segmentation device for an aviation multi-blade, comprising:

[0031] A construction module, used to construct a KD-tree corresponding to the measuring points on the aviation multi-joint blade, and traverse the KD-tree to calculate the distance density of each measuring point;

[0032] A clustering module, used to divide all the measuring points on the aviation multi-joint blade into a plurality of clusters according to the distance density and connectivity, and determine an initial blade profile based on the chord length of the measuring points of each cluster;

[0033] A deletion module, used for deleting noise points and overflow points on each of the initial blade profiles to obtain a target blade profile;

[0034] The grouping module is used to group the target blade profiles sorted according to the height average values ​​into groups per section, so that each group of the target blade profiles obtained corresponds one-to-one to a theoretical blade profile section.

[0035] According to another aspect of the present invention, a profile segmentation system for aviation multi-blades is provided, comprising: a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the profile segmentation method.

[0036] In general, the above technical solution conceived by the present invention has the following beneficial effects compared with the prior art:

[0037] (1) The present invention provides a method for segmenting the profile of an aviation multi-joint blade. First, the corresponding KD-tree of the aviation multi-joint blade is traversed to calculate the distance density of each measuring point; then, all measuring points on the aviation multi-joint blade are divided into multiple clusters according to the distance density and connectivity, and the initial blade profile is determined based on the chord length of the measuring points of each cluster; secondly, the noise points and overflow points on each initial blade profile are deleted to obtain the target blade profile; finally, the target blade profile sorted according to the height average value is grouped per section to obtain the target blade profile. The above method obtains a one-to-one correspondence between the multi-joint blade profile and the theoretical blade profile section in a manner of first stratification and then grouping. It can solve the problem of the mismatch between the blade profile measuring points and the theoretical blade profile section during the segmentation of the multi-joint blade section, thereby improving the segmentation accuracy and preparing for subsequent matching evaluation work.

[0038] (2) The present invention adopts the DBSCAN clustering method and uses the measurement point density to distinguish different cross-sections, especially for the blade profile measurement points in the case of flow channel interference. It can avoid the interference problem of flow channel measurement points as spatial free curves on various planes, and delete redundant flow channel measurement points after segmentation to achieve a more accurate segmentation effect.

[0039] (3) The present invention completes the denoising of internal and external noise points in the measuring points during the segmentation process, especially for external noise points such as overflow points caused by the measurement trajectory of multi-joint blades, the denoising is achieved by adopting a segmented single connection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flow chart of a method for segmenting a profile of a multi-jointed aviation blade in one embodiment of the present invention;

[0041] Figure 2 It is a flow chart of a method for segmenting a profile of a multi-jointed aviation blade in another embodiment of the present invention;

[0042] Figure 3 This is a diagram showing the effect of using DBSCAN clustering and segmentation for an aviation multi-joint blade with flow channel data in one embodiment of the present invention;

[0043] Figure 4 This is a diagram showing the effect of multi-link segmentation in one embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0045] like Figure 1 As shown, the present invention provides a method for segmenting a profile of a multi-jointed aviation blade, comprising:

[0046] S1: Construct the KD-tree corresponding to the measuring points on the aviation multi-joint blade, and traverse the KD-tree to calculate the distance density of each measuring point;

[0047] S2: Divide all the measuring points on the aviation multi-joint blade into multiple clusters according to the distance density and connectivity, and determine the initial blade profile based on the chord length of the measuring points in each cluster;

[0048] S3: Delete the noise points and overflow points on each initial blade profile to obtain the target blade profile;

[0049] S4: grouping the target blade profiles sorted according to the average height values ​​into groups per section, so that each group of target blade profiles obtained corresponds one-to-one to the theoretical blade profile section.

[0050] Specifically, first, in the order of stratification and then grouping, a KD-tree is constructed for the aviation multi-blade profile measurement points based on the point-to-point distance, and the distance density of each point is calculated based on the KD-tree. Then, the DBSCAN method is used to divide the aviation multi-blade measurement points into multiple clusters according to the density and connectivity principles, and the blade profile is judged according to the chord length of the cluster measurement points. Secondly, the RANSAC method is used to judge and remove noise points for the measurement points of each profile, and the overflow points of each profile are identified and deleted based on the continuity of the two points before and after the measurement trajectory. Finally, all profiles are sorted according to the average height of the profile, and all profiles are grouped according to each section. Each group of blade profiles corresponds one-to-one to the theoretical blade profile section.

[0051] In one example, the measured points of the aviation blade profile are collected by, but not limited to, a three-dimensional coordinate measuring machine, a point laser displacement sensor, and an area array scanner, and the blade profile measurement points are given by coordinate information or control parameters of points, line segments, arcs, and spline curves.

[0052] Furthermore, if Figure 2As shown in the figure, the specific order of stratification and then grouping is as follows: first, all the sections included in the measuring point data are separated according to the measuring point density and connectivity, and then the noise points and overflow points are deleted according to the RANSAC method and the continuity of the front and back points of the measuring trajectory. Finally, the grouping operation is performed on all the stratified result sections, and they are matched one by one with the theoretical blade profiles.

[0053] The method of the invention is suitable for the profile segmentation of various forms of aviation multi-blade blades.

[0054] In one embodiment, S1 includes:

[0055] S1: Select a split node from the measurement point set based on the split dimension, divide the measurement point set into two sub-measurement point sets according to the split node, and then recursively calculate the two sub-measurement point sets to obtain a KD-tree; the measurement point set is a collection of all measurement points on the aviation multi-joint blade;

[0056] S2: Based on the KD-tree, find the N closest points corresponding to each measuring point and obtain the corresponding N distance values; determine the number of distance values ​​less than the distance threshold from the N distance values ​​to obtain the distance density of each measuring point.

[0057] Specifically, the specific method of constructing KD-tree is: construct the tree node of KD-tree, select the split dimension according to the current measurement point set, and then select the split node from the measurement point set according to the split dimension, divide the measurement point set into two sub-measurement point sets according to the split node, and use the above method to perform recursive calculation on the two sub-measurement point sets respectively, and finally obtain the KD-tree structure of the entire measurement point set.

[0058] The specific implementation method of calculating the distance density of each point is as follows: traverse each point of the blade surface measurement point, find the corresponding N points that meet the closest distance according to the constructed KD-tree, and obtain N distance values ​​arranged from small to large. Compare these N distance values ​​with the given distance threshold, calculate the number of distances that meet the condition of being less than the distance threshold, and obtain the distance density of each point.

[0059] In one embodiment, S2 includes:

[0060] For any measuring point P, multiple measuring points {P0, P1, P2, ..., P m} and marked as P density reachable, the measuring points that are P density reachable are taken as a cluster, thereby dividing all measuring points into multiple clusters; the clusters whose number of measuring points exceeds the threshold M are regarded as the undetermined profile clusters, the chord length of the measuring points of each undetermined profile cluster is calculated, and the undetermined profile clusters whose measuring point chord length is greater than the chord length threshold are regarded as the initial blade profile.

[0061] The DBSCAN clustering method is: starting from a measurement point P, marked as P density reachable, searching for measurement points P0, P1, P2, etc. that meet the distance condition less than the distance threshold according to KD-tree, all marked as P density reachable. Recursively traverse the measurement points retrieved above, and take the points marked as P density reachable in the entire measurement point set as a cluster. Then use the above method for other unmarked measurement point sets until all measurement points are divided into different cluster sets.

[0062] Furthermore, the threshold method for determining blade profile is as follows: for all cluster sets, clusters that satisfy the number of cluster measurement points greater than or equal to the cluster threshold number of points M are regarded as pending measurement point profile clusters. The chord lengths of all pending measurement point profile clusters are calculated, and pending measurement point profiles that satisfy the chord length threshold are regarded as measurement point profile clusters.

[0063] In one embodiment, S3 includes:

[0064] S31: using the RANSAC method to identify and delete noise points on each initial blade profile;

[0065] S32: For each initial blade profile, starting from a measuring point Q, when the distance between the next node Q1 and Q is greater than the distance threshold, the passed measuring point is regarded as a measurement trajectory; the initial blade profile is traversed to obtain several measurement trajectories, and the intersection part is clipped based on the first and last coordinates of each measurement trajectory to obtain the target blade profile; wherein, there are overflow points in the intersection part.

[0066] In one embodiment, S31 includes:

[0067] The initial blade profiles that contain a number of measurement points less than the threshold M are regarded as external noise points and deleted;

[0068] The envelope curve of the initial blade profile is solved by the rolling ball method, and the measurement points not on the envelope curve are regarded as internal noise points and deleted;

[0069] The RANSAC method is used to randomly fit the initial blade profile to the theoretical blade profile, and the noise points are removed again.

[0070] Specifically, the methods for determining noise points include:

[0071] (1) Delete the noise points outside the surface measurement points by using the cluster threshold number M. The clusters that satisfy the number of cluster measurement points less than the cluster threshold number M are regarded as noise points outside the surface measurement points. It should be noted that when the number of measurement points is less than the threshold M, the cluster is regarded as an external noise point and deleted; when the number of measurement points is greater than the threshold M, the cluster is regarded as a pending surface cluster.

[0072] (2) Delete the noise points inside the surface measurement points by using the concave hull. The envelope curve of the surface measurement point contour is solved by the rolling ball method, and the points not on the envelope curve are regarded as noise points inside the surface measurement points.

[0073] (3) For the surface that completes (1) and (2) above, the RANSAC method is used to randomly fit it to the theoretical blade surface to remove noise points. Figure 3 This is a DBSCAN clustering segmentation effect diagram of an aviation multi-joint blade with flow channel data in one embodiment of the present invention, where the lines in the diagram are the contours of the initial blade profiles obtained; Figure 4 This is a diagram showing the effect of multi-link segmentation in one embodiment of the present invention.

[0074] Furthermore, the method of deleting the overflow points of the profile includes: dividing the profile measurement points into several segments, starting from a measurement point P, when the distance between the next node P1 and P is greater than a threshold, the above measurement points are regarded as one segment. The entire profile measurement points are traversed to obtain several measurement tracks, the beginning and end of each measurement track are determined, and the intersection part with the overflow points is cut out.

[0075] In one embodiment, S4 includes: grouping the X target blade profiles sorted according to the average height according to each cross section to obtain Y groups of target blade profiles;

[0076] in, Each set of target blade profiles corresponds one-to-one to the theoretical blade profile section.

[0077] Specifically, the corresponding relationship between the multi-blade profile and the theoretical blade profile section is obtained by:

[0078] The surface measurement points that have been arranged from large to small according to the average height are divided into Y groups, where Each group of profile measurement points corresponds one-to-one to the theoretical blade profile section.

[0079] In one embodiment, after S2 and before S3, the method further includes:

[0080] The initial blade profile is sorted from small to large according to the average height of each measuring point. The sorted initial blade profile meets H(C j ) is the average height of the initial blade profile with serial number j.

[0081] In one embodiment, before S4, the method further includes: sorting the target blade profiles from small to large according to the average heights of the included measuring points, and the sorted target blade profiles satisfy H'(C j ) is the average height of the target blade profile with serial number j.

[0082] According to another aspect of the present invention, there is provided a profile segmentation device for an aviation multi-blade, comprising:

[0083] A construction module is used to construct the KD-tree corresponding to the measuring points on the aviation multi-joint blade, and traverse the KD-tree to calculate the distance density of each measuring point;

[0084] A clustering module is used to divide all the measurement points on the aviation multi-joint blade into multiple clusters according to the distance density and connectivity, and determine the initial blade profile based on the chord length of the measurement points of each cluster;

[0085] A deletion module is used to delete noise points and overflow points on each initial blade profile to obtain a target blade profile;

[0086] The grouping module is used to group the target blade profiles sorted according to the average height into groups per section, so that each group of target blade profiles obtained corresponds one to one with the theoretical blade profile section.

[0087] According to another aspect of the present invention, a profile segmentation system for aviation multi-blades is provided, comprising: a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the profile segmentation method.

[0088] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for segmenting the profile of a multi-jointed aviation blade, characterized in that: include: S1: constructing a KD-tree corresponding to the measuring points on the aviation multi-joint blade, traversing the KD-tree to calculate the distance density of each measuring point; S2: dividing all the measuring points on the aviation multi-joint blade into a plurality of clusters according to the distance density and connectivity, and determining an initial blade profile based on the chord length of the measuring points of each cluster; S3: deleting noise points and overflow points on each of the initial blade profiles to obtain a target blade profile; S4: grouping the target blade profiles sorted according to the height average values ​​into groups per section, so that each group of the target blade profiles obtained corresponds one-to-one to a theoretical blade profile section; The S1 includes: S11: selecting a splitting node from a measuring point set based on a splitting dimension, dividing the measuring point set into two sub-measuring point sets according to the splitting node, and then recursively calculating the two sub-measuring point sets to obtain the KD-tree; the measuring point set is a set of all measuring points on the aviation multi-joint blade; S12: searching for the N closest points corresponding to each measuring point based on the KD-tree and obtaining the corresponding N distance values; determining the number of distance values ​​less than the distance threshold from the N distance values ​​to obtain the distance density of each measuring point; The step S2 includes: for any measuring point P, retrieving multiple measuring points whose distance is less than a distance threshold from the KD-tree. and marked as P density reachable, taking the P density reachable measuring points as a cluster, thereby dividing all the measuring points into multiple clusters; taking the clusters including the measuring points exceeding the threshold M as the undetermined profile clusters, calculating the measuring point chord length of each of the undetermined profile clusters, and taking the undetermined profile clusters with the measuring point chord length greater than the chord length threshold as the initial blade profile; The S3 includes: S31: using the RANSAC method to identify and delete noise points for each of the initial blade profiles; S32: starting from a measuring point Q for each of the initial blade profiles, when the distance between the next node Q1 and Q is greater than a distance threshold, the passed measuring point is regarded as a measurement trajectory; traversing the initial blade profile to obtain a number of measurement trajectories, and cutting the intersection part based on the first and last coordinates of each measurement trajectory to obtain the target blade profile; wherein the intersection part has the overflow point; The step S4 includes: grouping the X target blade profiles sorted by the average height according to each cross section to obtain Y groups of target blade profiles; wherein: , each group of target blade profiles corresponds one-to-one to the theoretical blade profile cross-section.

2. The method for dividing the profile of a multi-jointed aviation blade according to claim 1, characterized in that: The S31 includes: The initial blade profile including the number of measurement points less than the threshold M is regarded as an external noise point and deleted; The envelope curve of the initial blade profile is solved by using a rolling ball method, and the measurement points not on the envelope curve are regarded as internal noise points and deleted; The RANSAC method is used to randomly fit the initial blade profile to the blade theoretical profile, and the noise points are removed again.

3. The method for dividing the profile of a multi-jointed aviation blade according to claim 1 or 2, characterized in that: After S2 and before S3, the method further includes: The initial blade profile is sorted from small to large according to the average height of each measuring point, and the sorted initial blade profile satisfies , is the average height of the initial blade profile with serial number j.

4. The method for dividing the profile of a multi-jointed aviation blade according to claim 1 or 2, characterized in that: Before S4, the method further includes: The target blade profile is sorted from small to large according to the average height of each measuring point, and the sorted target blade profile satisfies , is the average height of the target blade profile with serial number j.

5. A profile segmentation device for aviation multi-blades, characterized in that: The method for performing the profile segmentation of the aviation multi-blade according to claim 1 comprises: A construction module, used to construct a KD-tree corresponding to the measuring points on the aviation multi-joint blade, and traverse the KD-tree to calculate the distance density of each measuring point; A clustering module, used to divide all the measuring points on the aviation multi-joint blade into a plurality of clusters according to the distance density and connectivity, and determine an initial blade profile based on the chord length of the measuring points of each cluster; A deletion module, used for deleting noise points and overflow points on each of the initial blade profiles to obtain a target blade profile; The grouping module is used to group the target blade profiles sorted according to the height average values ​​into groups per section, so that each group of the target blade profiles obtained corresponds one-to-one to a theoretical blade profile section.

6. A profile segmentation system for aviation multi-blades, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the profile segmentation method according to any one of claims 1 to 4.

Images