A method and device for automatic feature recognition of molds with a large number of bosses and grooves

Abstract

The present invention provides an automatic feature recognition method and device for a mold with a large number of bosses and grooves, wherein the method comprises: S1 importing a STEP file model of the mold; S2 constructing an attribute adjacency graph according to the model; S3 decomposing the model into multiple meta-features and constructing a meta-feature feature matrix; S4 matching the meta-feature feature matrix with a feature library feature matrix; if the match is successful, feature recognition is completed, and feature information corresponding to the matched feature matrix is ​​obtained; and if the feature matrix cannot be matched, the unrecognized features are added to the feature library according to case analysis, the expansion of the feature library is completed, and then the matching and recognition of the features are performed. The present invention helps to improve the accuracy and efficiency of feature recognition.

Description

Technical Field

[0001] The invention relates to the technical field of mold feature recognition, in particular to an automatic feature recognition method and device for a mold with a large number of bosses and grooves. Background Art

[0002] With the rapid development of modern industry and innovative product design, mold shapes are complex and changeable, which requires enterprises to respond quickly to market demand and actively shorten product development and production cycles. Manufacturing feature recognition is one of the important technologies studied in CAD / CAPP / CAM integrated technology. Accurately extracting the required manufacturing features from the .STEP file model is the basis of the CAPP system. The large number of features, feature geometric complexity, and matching accuracy are currently hot topics for domestic and foreign scholars. Tang Zhihong et al. established a feature recognition algorithm based on the same side, proposed methods such as application attribute adjacency graph construction, effective adjacent face identification, and complete identification of associated faces, and gradually identified the associated faces at all levels to construct the same side on both sides. The basic feature construction and feature adjacency graph construction were realized through the matching of the same side unit. Kuang Su et al. proposed to extract the processing object information from the STEP file, reorganize the underlying basic geometric feature data framework, and proposed an improved feature recognition method based on the associated area. Zhang Yu et al. proposed a STEP-NC manufacturing feature recognition method based on STEP and improved neural network. After extracting geometric topological information from the neutral file of STEP AP203, the minimum subgraph of the part was constructed based on the convexity and concavity of the edge. By inputting the obtained minimum subgraph information data of the part model into the improved BP neural network, the STEP-NC manufacturing feature recognition was realized. At present, great progress has been made in the problem of part feature recognition, but for complex mold models with a large number of complex features such as bosses and grooves, the recognition method still has the problems of low accuracy and large amount of calculation. Summary of the invention

[0003] In response to the complex construction model proposed above, the present invention aims to provide a method and device for automatic feature recognition of molds with a large number of bosses and grooves, construct an attribute adjacency graph, decompose the complex workpiece model into multiple meta-features and calculate the feature matrix, and match the feature matrix with the feature library to accurately identify the model features.

[0004] The purpose of the present invention is achieved by the following technical solutions:

[0005] In a first aspect, the present invention provides an automatic feature recognition method for a mold with a large number of bosses and grooves, comprising:

[0006] S1 imports the STEP file model of the mold;

[0007] S2 builds an attribute adjacency graph based on the model;

[0008] S3 decomposes the model into multiple meta-features and constructs a meta-feature feature matrix;

[0009] S4 matches the meta-feature feature matrix with the feature library feature matrix; if the match is successful, feature recognition is completed and feature information corresponding to the matched feature matrix is ​​obtained; if the feature matrix cannot be matched, the unidentified features are added to the feature library based on case analysis to complete the expansion of the feature library, and then feature matching and recognition are performed.

[0010] In one implementation, in step S2, constructing an attribute adjacency graph specifically includes:

[0011] The mold structure is characterized by numbering the characteristic faces of the mold, and the concavity and convexity between adjacent faces are calculated. According to the relationship between the characteristic faces and the concavity and convexity characteristics between adjacent faces, an attribute adjacency graph is constructed.

[0012] In one embodiment, the nodes in the attribute adjacency graph represent the characteristic surfaces of the part model, the arc connections between the nodes represent the relationship between the surfaces, and the numbers on the arcs represent the concavity and convexity of adjacent surfaces, where 0 represents the concave attribute of the adjacent surfaces and 1 represents the convex attribute of the adjacent surfaces.

[0013] In one implementation, in step S3, the model is decomposed into multiple meta-features, specifically including:

[0014] The part model is decomposed into composite features and meta features, and the composite features are further decomposed based on boundary relationships and concave-convex relationships to obtain corresponding meta features; the meta features are basic geometric shapes that constitute the model, and the basic geometric shapes include blind holes, bosses, keyways, through holes, square slots, elliptical square slots, etc.; the meta features contain feature faces that constitute the corresponding geometric shapes.

[0015] In one implementation, in step S3, constructing a meta-feature matrix includes:

[0016] According to the concavity and convexity between adjacent faces of the meta-feature in the part model and the relationship between the common edge attributes, a feature matrix is ​​constructed using different matrix element attribute values; the values ​​of the main diagonal elements in the feature matrix are used to describe the attributes of the meta-feature surface, and the surface attributes include the surface attribute values ​​corresponding to different surface types; the values ​​of other elements in the matrix are used to describe the surface attributes, the concavity and convexity between adjacent faces and the relationship between the common edge attributes.

[0017] In one implementation, the values ​​of other elements in the meta-feature matrix are four-digit eigenvalues, where the ones digit of the element value is the type of common edge, the tens digit is the concavity of adjacent faces, the hundreds digit and the thousands digit are the types of faces, and the value of elements that are not adjacent to each other is 0.

[0018] In one implementation, in step S4, matching the meta-feature matrix with the feature library matrix specifically includes:

[0019] S41: Calculate the meta-feature matrix A, obtain the number of main diagonal elements m of the meta-feature matrix, and compare it with the number of main diagonal elements n of the feature matrix B in the feature matrix library. When m=n, retain the feature matrix B that meets the conditions, and the feature matrix library completes the preliminary screening. When the number of retained feature matrices B is greater than or equal to 1, go to S42; otherwise, go to S46;

[0020] S42: Calculate the trace tr(A) of the main diagonal of the element feature matrix A, and compare it with the trace tr(B) of the main diagonal of the filtered feature matrix B. When tr(A)=tr(B), retain the feature matrix B that meets the conditions, and the feature matrix library completes the secondary screening; when the number of feature matrices B retained after the secondary screening is greater than or equal to 1, go to S43; otherwise, go to S46;

[0021] S43: taking the elements of each row of the meta-feature matrix A as the basis, matching the feature matrix B after secondary screening row by row, and then going to S44 after completion;

[0022] S44: if the feature matrix A matches successfully, feature recognition is completed, and the process goes to S45; if the match fails, the process goes to S46;

[0023] S45: Match the next meta-feature with the feature matrix library and go to S41; if all features have been identified, go to S47;

[0024] S46: Prompt manual feature recognition, and add unrecognized feature matrices to the feature matrix library to complete expansion, and then go to S41; if all features have been recognized, go to S47;

[0025] S47: Feature recognition is completed and the corresponding feature recognition result is output.

[0026] In a second aspect, the present invention proposes an automatic feature recognition device for a mold with a large number of bosses and grooves, wherein the device is used to implement the automatic feature recognition method for a mold with a large number of bosses and grooves as described in any one of the embodiments of the first aspect above.

[0027] The beneficial effects of the present invention are as follows: (1) In order to solve the problems of large number of mold features, geometric complexity of features, and feature recognition and matching accuracy, a method based on attribute adjacency graph is proposed, which uses the matching recognition of meta-feature feature matrix and feature library feature matrix to accurately realize the recognition of model features.

[0028] (2) The mold model is decomposed into multiple types of meta-features by using the boundary relationship and convex-concave relationship between feature surfaces. The convex-concave properties of surfaces and adjacent surfaces and the attribute types of edges are defined. The feature matrix of each meta-feature is calculated and matched with the feature matrix of the feature library, which helps to improve the accuracy and efficiency of feature recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0030] Figure 1 A schematic diagram of an automatic feature recognition method for a mold with a large number of bosses and grooves according to an exemplary embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a feature recognition method flow chart shown in an exemplary embodiment of the present invention;

[0032] Figure 3 A schematic diagram of a mold model shown in an exemplary embodiment of the present invention;

[0033] Figure 4 for Figure 3 Property adjacency graph of the die model shown;

[0034] Figure 5 A schematic diagram of a model composition shown in an exemplary embodiment of the present invention;

[0035] Figure 6 A schematic diagram of boundary relations shown in an exemplary embodiment of the present invention;

[0036] Figure 7 A schematic diagram of determining a concave-convex relationship according to an exemplary embodiment of the present invention;

[0037] Figure 8 for Figure 3 Schematic diagram of the composite feature decomposition results of the mold model shown. DETAILED DESCRIPTION

[0038] The present invention is further described in conjunction with the following application scenarios.

[0039] See also Figure 1 , which shows an automatic feature recognition method for a mold with a large number of bosses and grooves, including:

[0040] S1 imports the STEP file part (mold) model of the mold;

[0041] S2 builds an attribute adjacency graph based on the model;

[0042] S3 decomposes the model into multiple meta-features and constructs a meta-feature feature matrix;

[0043] S4 matches the meta-feature feature matrix with the feature library feature matrix; if the match is successful, feature recognition is completed and feature information corresponding to the matched feature matrix is ​​obtained; if the feature matrix cannot be matched, the unidentified features are added to the feature library based on case analysis to complete the expansion of the feature library, and then feature matching and recognition are performed.

[0044] Among them, the feature recognition flow chart is as follows Figure 2 As shown in the figure, firstly, the .STEP file model is imported to calculate the concavity and convexity of adjacent faces of the model; secondly, the model attribute adjacency graph is constructed; then, the model is decomposed into multiple meta-features, the attribute types of faces and edges are defined, and the meta-feature feature matrix is ​​constructed; finally, the meta-feature feature matrix is ​​matched with the feature library feature matrix. If the match is successful, the feature recognition is completed; if the feature matrix cannot be matched, the unidentified features are added to the feature library according to the case analysis, the feature library is expanded, and then the feature matching and recognition are performed.

[0045] In one implementation, in step S2, constructing an attribute adjacency graph specifically includes:

[0046] The mold structure is characterized by numbering the characteristic faces of the mold, and the concavity and convexity between adjacent faces are calculated. According to the relationship between the characteristic faces and the concavity and convexity characteristics between adjacent faces, an attribute adjacency graph is constructed.

[0047] In one embodiment, the nodes in the attribute adjacency graph represent the characteristic surfaces of the part model, the arc connections between the nodes represent the relationship between the surfaces, and the numbers on the arcs represent the concavity and convexity of adjacent surfaces, where 0 represents the concave attribute of the adjacent surfaces and 1 represents the convex attribute of the adjacent surfaces.

[0048] In one implementation, taking a specific mold structure as an example, by performing feature analysis on a mold structure, such as Figure 3 As shown in the figure, it can be seen that it mainly includes blind holes, keyways, bosses, etc. Each feature surface is numbered and marked with f. n Indicates, where n = 1 to 29, representing the surface unit of the part model. 7 is the cylinder, f 8 It is a plane, and the two together constitute a cylindrical boss.

[0049] The concavity and convexity between adjacent faces is an important process of feature recognition. Since the faces in the model may be divided into many types, no matter what type of faces on the model are, the spatial relationship between adjacent faces is determined by the angle θ between the two faces. The calculation of concavity and convexity is mainly divided into four categories for discussion: intersection of planes, intersection of planes and arc edges, intersection of planes and straight edges of arcs, and intersection of arcs. The calculation methods of concavity and convexity are introduced in detail in the existing technical literature, and this application will not repeat them here.

[0050] The attribute adjacency graph is a graph-based feature recognition method. In the attribute adjacency graph, nodes represent the feature faces of the part model, faces are connected by arcs, and the numbers on the arcs represent the concavity and convexity of adjacent faces. It is defined that 0 represents the concave attribute of adjacent faces, and 1 represents the convex attribute of adjacent faces. Figure 4 for Figure 3 In the attribute adjacency graph, connected face numbers represent two adjacent faces.

[0051] In one implementation, in step S3, the model is decomposed into multiple meta-features, specifically including:

[0052] The part model is decomposed into composite features and meta features, and the composite features are further decomposed based on boundary relationships and concave-convex relationships to obtain corresponding meta features; wherein the meta features are basic geometric shapes constituting the model, wherein the basic geometric shapes include blind holes, bosses, keyways, through holes, square slots, and elliptical square slots; wherein the meta features include feature faces constituting the corresponding geometric shapes;

[0053] Any complex model can be considered as composed of multiple features of different shapes. Feature recognition starts from the solid model of the part. The model can be divided into composite features and meta-features. Meta-features are the most basic geometric shapes that constitute the model. The feature faces of various meta-features can intersect and share a feature face, but meta-features cannot be further refined and decomposed. Figure 3 Medium 26 ~f 29 Constitute the composite feature, f 26 is a cylindrical surface, f 27 is a plane, f 26 and f 27 Composition element feature blind hole; f 28 is a cylindrical surface, f 29 is a plane, f 28 and f 29 Any other complex model can be represented by these meta-features, such as Figure 5 shown.

[0054] In order to achieve high accuracy in feature recognition, the model needs to be decomposed into multiple meta-features. The boundary relationship and concave-convex relationship between feature faces are used as the judgment criteria for model decomposition.

[0055] (1) Boundary relations

[0056] Each feature surface in the model has an outer boundary and one or more inner boundaries. Composite features can be decomposed into multiple meta-features through boundary relationships. Figure 6 As shown in the figure, the outer boundary of the blind hole coincides with the inner boundary of the plane, and the composite feature can be decomposed into two parts: the plane and the blind hole.

[0057] (2) Concave-convex relationship

[0058] There is a concave-convex relationship between the feature faces of each meta-feature, and the overall model and composite features are decomposed through the concave-convex relationship between adjacent faces. Figure 7 As shown in the figure, if the angle between surface A and surface B of the part's outer contour is less than 180°, the two surfaces are in a concave relationship. If the angle between surface A and surface B is greater than 180°, the two surfaces are in a convex relationship. The adjacent feature faces within each meta-feature only contain concave or convex relationships.

[0059] Define f 1 The bottom surface is used as the reference surface, f 1 Not used as meta-feature feature surface. According to the above composite feature decomposition method, Figure 3 The overall decomposition of the model can obtain 8 meta-features, such as Figure 8 shown.

[0060] In one implementation, in step S3, constructing a meta-feature matrix includes:

[0061] According to the concavity and convexity between adjacent faces of the meta-feature in the part model and the relationship between the common edge attributes, a feature matrix is ​​constructed using different matrix element attribute values; the values ​​of the main diagonal elements in the feature matrix are used to describe the attributes of the meta-feature surface, and the surface attributes include the surface attribute values ​​corresponding to different surface types; the values ​​of other elements in the matrix are used to describe the surface attributes, the concavity and convexity between adjacent faces and the relationship between the common edge attributes.

[0062] In one implementation, the values ​​of other elements in the meta-feature matrix are four-digit eigenvalues, where the ones digit of the element value is the type of common edge, the tens digit is the concavity of adjacent faces, the hundreds digit and the thousands digit are the types of faces, and the value of elements that are not adjacent to each other is 0.

[0063] The present invention proposes to construct a feature matrix based on the attributes of the surface, the concavity and convexity between adjacent surfaces and the relationship between the common edge attributes. The feature matrix of different element features has unique certainty, which makes the accurate matching of features in the instance model fully feasible. The types of surfaces include planes, cylinders, semi-cylinders, elliptical cylinders, cones, and spheres. The type attributes of the surfaces are defined as shown in Table 1.

[0064] Table 1 Type attribute definition

[0065]

[0066] Specifically, 1 to 6 represent plane, cylinder, semi-cylinder, ellipse, cone, and sphere, respectively. i With f j When the faces are adjacent, get the face attribute value corresponding to the face type, such as 11 for two planes connected, 12 for a plane and a cylinder connected; if f i With f j When the faces are not adjacent, the corresponding matrix element is 0

[0067] The attribute types of edges include straight lines, circular arcs, semicircular arcs, elliptical arcs, and B-spline curves. The type attributes of edges are defined as shown in Table 2.

[0068] Table 2 Edge type attribute definition

[0069]

[0070] Specifically, 1 to 5 represent straight lines, circular arcs, semicircular arcs, elliptical arcs, and B-spline curves, respectively. i With f j When the faces are adjacent, get the edge attribute value. For example, if the common edge between two adjacent planes is a straight line, the attribute value is 1. When the plane and the cylindrical surface are adjacent, the common edge is an arc, and the attribute value is 2. i With f j When the faces are not adjacent, the corresponding matrix element value is 0.

[0071] The method is applied to create a feature matrix. The values ​​of the main diagonal elements of the matrix are used to describe the properties of the surface, ensuring that the elements adjacent to the main diagonal are adjacent feature surfaces, followed by curved surfaces. The values ​​of other elements of the matrix are used to describe the properties of the surface, the concavity and convexity between adjacent surfaces, and the relationship between the properties of the common edges. The ones digit of the element value is the type of the common edge, the tens digit is the concavity and convexity of the adjacent surfaces, and the hundreds and thousands digits are the types of the surfaces. The values ​​of elements that are not adjacent to each other are defined as 0.

[0072] The construction of feature library is open and expandable. It is usually based on the existing feature library and continuously expands the feature library in combination with specific case analysis. Through the analysis of a large number of mold examples, a feature library for mold parts is constructed. Table 3 shows some typical meta-features in the feature library and their feature matrix.

[0073] Table 3 Feature prototypes in the feature library

[0074]

[0075]

[0076] In one implementation, in step S4, matching the meta-feature matrix with the feature library matrix specifically includes:

[0077] S41: Calculate the meta-feature matrix A, obtain the number of main diagonal elements m of the meta-feature matrix, and compare it with the number of main diagonal elements n of the feature matrix B in the feature matrix library. When m=n, retain the feature matrix B that meets the conditions, and the feature matrix library completes the preliminary screening. When the number of retained feature matrices B is greater than or equal to 1, go to S42; otherwise, go to S46;

[0078] S42: Calculate the trace tr(A) of the main diagonal of the element feature matrix A, and compare it with the trace tr(B) of the main diagonal of the filtered feature matrix B. When tr(A)=tr(B), retain the feature matrix B that meets the conditions, and the feature matrix library completes the secondary screening; when the number of feature matrices B retained after the secondary screening is greater than or equal to 1, go to S43; otherwise, go to S46;

[0079] S43: taking the elements of each row of the meta-feature matrix A as the basis, matching the feature matrix B after secondary screening row by row, and then going to S44 after completion;

[0080] S44: if the feature matrix A matches successfully, feature recognition is completed, and the process goes to S45; if the match fails, the process goes to S46;

[0081] S45: Match the next meta-feature with the feature matrix library and go to S41; if all features have been identified, go to S47;

[0082] S46: Prompt manual feature recognition, and add unrecognized feature matrices to the feature matrix library to complete expansion, and then go to S41; if all features have been recognized, go to S47;

[0083] S47: Feature recognition is completed, the result is output, and the process ends.

[0084] In a second aspect, the present invention proposes an automatic feature recognition device for a mold with a large number of bosses and grooves, wherein the device is used to implement the automatic feature recognition method for a mold with a large number of bosses and grooves as described in any one of the embodiments of the first aspect above.

[0085] Computational Construction Figure 7 The meta-feature matrix in is shown in Table 4.

[0086] Table 4 Calculation of meta-feature matrix

[0087]

[0088] According to the above process, the meta-feature feature matrix is ​​matched with the feature matrix library. (a) The feature matrix is ​​a square boss feature, (b), (c), and (h) The feature matrix is ​​a cylindrical boss feature, (d) The feature matrix is ​​a keyway feature, (e) The feature matrix is ​​a square slot feature, (f) The feature matrix is ​​an elliptical square slot feature, and (g) The feature matrix is ​​a blind hole feature. All meta-features are matched, and the result accuracy is 100%.

[0089] Aiming at the problems of large number of mold features, geometric complexity of features and accuracy of feature recognition and matching, the present invention proposes a method based on attribute adjacency graph, which utilizes the matching recognition of meta-feature feature matrix and feature library feature matrix, and accurately realizes the recognition of model features.

[0090] The mold model is decomposed into multiple types of meta-features by using the boundary relationship and convex-concave relationship between feature surfaces. The concave-convexity of surfaces and adjacent surfaces and the attribute type of edges are defined. The feature matrix of each meta-feature is calculated and matched with the feature matrix of the feature library. It is proved that this type of feature recognition method can be accurately and efficiently recognized.

[0091] At the same time, based on the mold automatic feature recognition method shown in the above embodiment, the present invention also proposes an automatic feature recognition device for a mold with a large number of bosses and grooves, wherein the device is used to realize the above Figure 1 The automatic feature recognition method for a mold with a large number of bosses and grooves described in any of the embodiments is not repeated in this application.

[0092] It should be noted that each functional unit / module in each embodiment of the present invention may be integrated into one processing unit / module, or each unit / module may exist physically separately, or two or more units / modules may be integrated into one unit / module. The above-mentioned integrated unit / module may be implemented in the form of hardware or in the form of software functional unit / module.

[0093] Through the description of the above implementation mode, it can be clearly understood by those skilled in the art that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described herein or their combination. For software implementation, part or all of the process of the embodiment can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein the communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The storage medium can be any available medium that can be accessed by a computer. Computer-readable media can include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should analyze that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. An automatic feature recognition method for a mold with a large number of bosses and grooves, characterized in that: include: S1 imports the STEP file model of the mold; S2 builds an attribute adjacency graph based on the model; S3 decomposes the model into multiple meta-features and constructs a meta-feature feature matrix; Specifically include: The part model is decomposed into composite features and meta features, and the composite features are further decomposed based on boundary relationships and concave-convex relationships to obtain corresponding meta features; wherein the meta features are basic geometric shapes constituting the model, wherein the basic geometric shapes include blind holes, bosses, keyways, through holes, square slots, and elliptical square slots; wherein the meta features include feature faces constituting the corresponding geometric shapes; According to the relationship between the concavity and convexity of adjacent faces of the meta-feature in the part model and the common edge attributes, the feature matrix is ​​constructed by using different matrix element attribute values; the values ​​of the main diagonal elements in the feature matrix are used to describe the attributes of the meta-feature faces, and the attributes of the faces include the face attribute values ​​corresponding to different face types; the values ​​of other elements in the matrix are used to describe the attributes of the faces, the concavity and convexity between adjacent faces, and the relationship between the common edge attributes; S4 matches the meta-feature feature matrix with the feature library feature matrix; if the match is successful, feature recognition is completed to obtain feature information corresponding to the matched feature matrix; if the feature matrix cannot be matched, the unrecognized features are added to the feature library according to case analysis to complete the expansion of the feature library, and then feature matching and recognition are performed; Among them, matching the meta-feature feature matrix with the feature library feature matrix specifically includes: S41: Calculate the meta-feature feature matrix A, obtain the number m of the main diagonal elements of the meta-feature feature matrix, and compare it with the number n of the main diagonal elements of the feature matrix B in the feature matrix library. When m=n, retain the feature matrix B that meets the conditions, and the feature matrix library completes the preliminary screening. When the number of retained feature matrices B is greater than or equal to 1, go to S42; otherwise, go to S46; S42: Calculate the trace tr(A) of the main diagonal of the element feature matrix A, and compare it with the trace tr(B) of the main diagonal of the filtered feature matrix B. When tr(A)=tr(B), retain the feature matrix B that meets the conditions, and the feature matrix library completes the secondary screening; when the number of feature matrices B retained after the secondary screening is greater than or equal to 1, go to S43; otherwise, go to S46; S43: taking the elements of each row of the meta-feature matrix A as the basis, matching the feature matrix B after secondary screening row by row, and then going to S44 after completion; S44: if the feature matrix A matches successfully, feature recognition is completed, and the process goes to S45; if the match fails, the process goes to S46; S45: Match the next meta-feature with the feature matrix library and go to S41; if all features have been identified, go to S47; S46: Prompt manual feature recognition, and add unrecognized feature matrices to the feature matrix library to complete expansion, and then go to S41; if all features have been recognized, go to S47; S47: Feature recognition is completed and the corresponding feature recognition result is output.

2. The automatic feature recognition method for a mold with a large number of bosses and grooves according to claim 1 is characterized in that: In step S2, an attribute adjacency graph is constructed, which specifically includes: The mold structure is characterized by numbering the characteristic faces of the mold, and the concavity and convexity between adjacent faces are calculated. According to the relationship between the characteristic faces and the concavity and convexity characteristics between adjacent faces, an attribute adjacency graph is constructed.

3. The automatic feature recognition method for a mold with a large number of bosses and grooves according to claim 2 is characterized in that: in, The nodes in the attribute adjacency graph represent the characteristic surfaces of the part model, the arc connections between the nodes represent the relationship between the surfaces, and the numbers on the arcs represent the concavity and convexity of the adjacent surfaces, where 0 represents the concave attribute of the adjacent surface and 1 represents the convex attribute of the adjacent surface.

4. The automatic feature recognition method for a mold with a large number of bosses and grooves according to claim 1 is characterized in that: The values ​​of other elements in the meta-feature matrix are four-digit eigenvalues, where the ones digit of the element value is the type of common edge, the tens digit is the concavity of adjacent faces, the hundreds digit and the thousands digit are the types of faces, and the value of elements that are not adjacent to each other is 0.

5. An automatic feature recognition device for a mold with a large number of bosses and grooves, characterized in that: The device is used to implement the automatic feature recognition method for a mold with a large number of bosses and grooves as described in any one of claims 1 to 4.

Images