Automatic dimensioning method for three-dimensional parts

Abstract

The application discloses a method for automatically marking the size of a three-dimensional part, and relates to the technical field of computers, and comprises the following steps: S1, constructing a hierarchical expression model, a feature type library and a feature marking rule library; S2, obtaining a three-dimensional part to be marked, and extracting geometric elements of the three-dimensional part to be marked and structural relationships between the elements as a geometric element set; S3, identifying features of the three-dimensional part to be marked by using the hierarchical expression model, and establishing a feature structure tree of the three-dimensional part to be marked; and S4, marking the size of the three-dimensional part to be marked by using the feature marking rule library and the feature structure tree. The method establishes a feature type library and a marking rule library of the three-dimensional part, takes modeling features as basic marking units, converts the size marking problem of the three-dimensional part into the problems of fixed-size marking and positioning-size marking of the modeling features, improves the accuracy of part feature recognition, guarantees the correctness and integrity of the marking result, and can effectively meet the use of actual engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer technology, and in particular to a method for automatically marking the size of a three-dimensional part. BACKGROUND

[0002] Three-dimensional marking is to transform two-dimensional engineering drawings, integrate all process sizes, technical requirements and tolerance information to three-dimensional models, and replace the role of two-dimensional drawings. Traditional manual marking work is repetitive and time-consuming, sometimes even more than modeling workload, and the marking efficiency is low, and it is easy to miss and make mistakes. Although three-dimensional marking technology has matured, three-dimensional automatic marking technology is still in the development stage. Through three-dimensional automatic marking, the need for two-dimensional drawings can be weakened, and the work efficiency and accuracy of the marking results can be greatly improved.

[0003] The research on three-dimensional model size automatic marking at home and abroad is mainly divided into two parts, which are from the geometric information (surface, edge and vertex) of the model and the modeling features of the model, and the theoretical method of size automatic marking is given. For the definition of modeling features, most researchers define it as a region with certain processing significance in a three-dimensional model, which is composed of a series of geometric information.

[0004] The size automatic marking based on model geometric information takes the geometric elements such as points, lines and surfaces of three-dimensional models as the research object, and the size marking is mainly to shape and position the geometric elements. Since it needs to extract all geometric elements of the model, it can completely describe the shape of the model from the pure geometric point of view, and the completeness of the marking result is high. However, this method extracts scattered points, lines and surfaces in the part, which cannot be directly reflected in the part process size and other engineering information because these information are not structured combined. Based on this, the marking elements obtained lack the corresponding engineering meaning. And in the subsequent calculation, the search of the marking elements is from the global range of the whole part, which makes the search space too wide, and thus the extraction efficiency of the marking elements is very low. In addition, since the size marking directly aims at the positioning and shaping of geometric elements such as points, lines and surfaces, the marking result is easy to appear size redundancy and "hedgehog" phenomenon, and the size marking is difficult to meet the actual engineering standard, and the engineering practicability is low.

[0005] The size automatic labeling based on the feature model takes part features as the research object, takes the features as the basic labeling unit, and performs shape and position labeling on the features. The labeling result is more accurate for the organization and expression of part engineering information, is easier to realize the labeling mode conforming to the national standard, and is more suitable for the whole life cycle of a product. However, the existing research has a technical difficulty in the acquisition and identification of the model features, and the accuracy of the labeling result depends on the accuracy of the part feature identification to a great extent. Since the existing feature identification technology cannot realize completely accurate feature identification, the correctness and integrity of the labeling result cannot be guaranteed. SUMMARY

[0006] The purpose of the present application is to solve the above problems by designing a three-dimensional part size automatic labeling method.

[0007] The present application realizes the above-mentioned purpose by the following technical solutions:

[0008] The three-dimensional part size automatic labeling method comprises:

[0009] S1, a hierarchical expression model, a feature type library and a feature labeling rule library are constructed, the hierarchical expression model is used for classifying geometric elements, the classification result of the hierarchical expression model comprises a part layer, a feature layer and a geometry layer, the geometry layer is used for describing basic geometric elements of a part and topological relations between the geometric elements, the feature layer is used for describing each sub-feature and mutual relations between the features; the feature type library comprises basic features, additional features and editing features; the feature labeling rule library comprises a labeling rule corresponding to each feature type;

[0010] S2, a three-dimensional part to be labeled is acquired, and geometric elements of the three-dimensional part to be labeled and structural relations between the elements are extracted as a geometric element set;

[0011] S3, features of the three-dimensional part to be labeled are identified by using the hierarchical expression model, and a feature structure tree of the three-dimensional part to be labeled is established;

[0012] S4, the three-dimensional part to be labeled is subjected to shape size labeling and position size labeling by using the feature labeling rule library and the feature structure tree.

[0013] The present application has the beneficial effects that the feature type library and the labeling rule library of the three-dimensional part are established, the modeling features are taken as the basic labeling unit, the size labeling problem of the three-dimensional part is converted into the shape size labeling and position size labeling problem of the modeling features, the accuracy of the part feature identification is improved, the correctness and integrity of the labeling result are guaranteed, and the actual engineering use can be effectively met. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1is a flow chart of the automatic dimensioning method of the three-dimensional part of the present application;

[0015] Figure 2 is a flow chart of the geometric element extraction of the three-dimensional part of the present application;

[0016] Figure 3 is a classification table of the feature type library of the present application;

[0017] Figure 4 is a classification type chart of the hierarchical expression model of the present application;

[0018] Figure 5 is the composition form of the part modeling feature;

[0019] Figure 6 is a flow chart of the condensed hierarchical clustering algorithm of the present application;

[0020] Figure 7 is a flow chart of the establishment of the feature structure tree of the present application;

[0021] Figure 8 is a flow chart of the feature recognition of the present application;

[0022] Figure 9 is a barrel tooling part drawing of the embodiment of the present application;

[0023] Figure 10 is a barrel tooling part establishment history record drawing of the embodiment of the present application;

[0024] Figure 11 is a feature recognition drawing of the barrel tooling part of the present application;

[0025] Figure 12 is the labeling result of the barrel tooling part of the present application. DETAILED DESCRIPTION

[0026] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0027] Therefore, the detailed description of the embodiments of the present application provided in the drawings below is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative labor are within the scope of protection of the present application.

[0028] It should be noted that similar reference numerals and letters refer to like items in the accompanying drawings, and once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0029] In the description of the present application, it needs to be understood that the terms "upper", "lower", "inner", "outer", "left", "right" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship commonly placed when the product of the application is used, or the orientation or positional relationship commonly understood by those skilled in the art, and are only for the convenience of describing the application and simplifying the description, and are not intended to indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application.

[0030] In addition, the terms "first", "second" and the like are only used for differentiation in description and cannot be understood as indicating or implying relative importance.

[0031] In the description of the present application, it also needs to be explained that, unless otherwise explicitly specified and limited, the terms "arranged", "connected" and the like should be understood broadly, for example, "connected" can be fixedly connected, or detachably connected, or integrally connected; can be mechanically connected, or electrically connected; can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0032] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0033] The size automatic labeling method of the three-dimensional part comprises:

[0034] S1, constructing a hierarchical expression model, a feature type library and a feature labeling rule library, the hierarchical expression model is used for classifying geometric elements, the classification results of the hierarchical expression model include a part layer, a feature layer and a geometric layer, the geometric layer is used for describing basic geometric elements of the part and the topological relationship between the geometric elements, the feature layer is used for describing each sub-feature and the mutual relationship between the features; the feature type library includes basic features, additional features and editing features; the feature labeling rule library includes the labeling rule corresponding to each feature type;

[0035] S2, acquiring a three-dimensional part to be labeled, and extracting geometric elements of the three-dimensional part to be labeled and the structural relationship between the elements as a geometric element set; specifically comprising:

[0036] S21, reading part information of the three-dimensional part to be labeled, the part information being three-dimensional part parameter information;

[0037] S22, calling a three-dimensional modeling software function to obtain geometric elements of a three-dimensional part to be labeled and structural relationships between the elements;

[0038] S23, structurally storing the geometric elements and the structural relationships to form a geometric element structured data set.

[0039] S3, identifying features of the three-dimensional part to be labeled using a hierarchical expression model, and establishing a feature structure tree of the three-dimensional part to be labeled; specifically comprising:

[0040] S31, identifying features of the geometric element set using a deep convolutional neural network;

[0041] S32, using a convolutional neural network to individually judge the feature identification results according to the similarity relationship between the features; the geometric elements contained in the same modeling feature are classified into a class, so that the geometric element set is classified into n classes;

[0042] S33, inputting the geometric elements contained in any two classes into the convolutional neural network for identification, calculating the similarity with each modeling feature in the feature library, and judging whether to merge the two input classes to form a new class according to the similarity;

[0043] S34, repeatedly executing S33 until the number of classes is 1, forming a feature structure tree.

[0044] S4, using a feature labeling rule library and a feature structure tree to perform form dimension labeling and positioning dimension labeling on the three-dimensional part to be labeled; specifically, traversing each feature of the feature structure tree, performing form dimension labeling on each feature according to the feature labeling rule library, and when performing positioning dimension labeling between features, performing priority sorting on the positioning dimension labeling between features according to the feature structure tree, and sequentially performing positioning dimension labeling on the features.

[0045] The working principle of the three-dimensional part dimension automatic labeling method of the application is as follows:

[0046] Automatic extraction of three-dimensional part geometric elements

[0047] The design of mechanical parts is mostly based on feature-based modeling of parts in a three-dimensional software environment, and Boolean operation is performed between features to form the final modeling of the part. No matter how complex a part is, through the parameterized design and feature coding functions provided by the system software during the modeling process, the geometric elements, constraint relationships between features, and shape parameters inside the features during the part modeling process are recorded in the parameterized design history of the part. By reading the three-dimensional part parameter information through the system software, the three-dimensional part geometric elements and their structural relationships can be obtained. The three-dimensional dimension automatic extraction algorithm flow is as shown in Figure 2 ​

[0048] Existing three-dimensional modeling software provides an open secondary development interface, which can achieve the acquisition of three-dimensional model internal geometric elements. After obtaining the geometric elements and their structural relationships of three-dimensional parts through the open interface (API) provided by the three-dimensional design software, the geometric elements are first stored in a structured manner to form a structured dataset of geometric elements. The structured dataset of part geometric elements is the basis for subsequent hierarchical clustering and feature recognition.

[0049] Feature modeling of three-dimensional parts

[0050] Features are composed of specific positional relationships of geometric elements and a certain number of faces and edges. The geometric modeling feature specification part of GB / T24734-2009 standard gives the standardization requirements for terms and definitions, classification, etc. of geometric modeling features in three-dimensional CAD applications. The standard divides geometric modeling features into three categories: basic modeling features, additional modeling features, and editing operation features. Referring to the classification principles of this standard, the basic feature type library of three-dimensional parts is established based on these features as the basic unit of automatic labeling of three-dimensional parts; the basic feature types of three-dimensional parts are shown in Table 1. Figure 3

[0051] Referring to the GB / T24734-2009 standard, geometric modeling features are divided into three categories: basic modeling features, additional modeling features, and editing modeling features. Feature labeling elements and labeling rule libraries are established for different types of features.

[0052] Most features (hole features, fillet features, chamfer features, and array features, etc.) have mature engineering labeling standards. For features with mature labeling standards, the feature is taken as a whole, and the feature labeling rule library is established in a fixed mode according to the engineering labeling standards. For features without mature labeling standards, the feature surface is taken as the basic unit of dimension labeling, the labeling elements are extracted for each feature surface, grouped, and the dimension labeling set is divided according to the grouping results, and a fixed mode feature labeling rule library is established.

[0053] Basic modeling features, also known as main modeling features, are used to construct the main shape or basic elements of a part. Basic modeling features are generally obtained from sketch features through stretching, rotating, scanning, and lofting, etc.

[0054] a) Sketch Feature

[0055] A sketch is a parametric feature that is used to express design intent after adding constraints after drawing a curve contour using a sketch tool. As shown in Table 1, the labeling elements of a sketch feature include: sketch drawing surface, sketch geometry, and sketch dimensions.

[0056] Table 1 Geometric elements of sketch feature ​

[0057]

[0058] b) Extrude Feature

[0059] A geometry generated by linearly translating a sketch on a sketch plane along the normal direction of the sketch plane is called an extrude feature. As shown in Table 2, the common parameters of an extrude feature include (but are not limited to): sketch feature, extrude start face, extrude end position (or extrude distance), extrude direction, extrude manner.

[0060] Table 2. Extrude Feature Geometry Elements

[0061]

[0062] c) Revolve Feature

[0063] A geometry generated by rotating a sketch profile on one side of a straight axis about the axis by a certain angle is called a revolve feature. As shown in Table 3, the common parameters of a revolve feature include (but are not limited to): sketch feature, revolve axis, revolve start face (or start angle), revolve end position (or end angle), revolve direction, revolve manner (single direction or double direction).

[0064] Table 3. Revolve Feature Geometry Elements

[0065]

[0066] d) Sweep Feature

[0067] A geometry generated by moving a sketch on a plane perpendicular to a certain trajectory line while keeping the position and direction of the intersection point of the sketch plane and the trajectory line unchanged is called a sweep feature. As shown in Table 4, the common parameters of a sweep feature include (but are not limited to): sweep trajectory line, sketch feature and specified direction, sweep start direction, sweep end point.

[0068] Table 4. Sweep Feature Geometry Elements

[0069]

[0070]

[0071] e) Loft Feature

[0072] A continuous geometric body or patch formed by connecting the outlines of two or more planar sketches according to certain rules is called a lofted feature, which should satisfy the defined sketch outline shape and size on the specified section. As shown in Table 5 below, the commonly used parameters of a lofted feature include (but are not limited to): the number of sections n, each section sketch, the control position of each lofted section, and the lofting type. Basic voxels are used to obtain the lofted feature.

[0073] Table 5 Geometric elements of a lofted feature

[0074]

[0075] Additional modeling features, also known as auxiliary modeling features, are usually not the first feature to appear. Additional modeling features are modifications or refinements of basic features or other additional modeling features, such as chamfers, fillets, ribs, etc.

[0076] a) Hole Feature

[0077] A hole feature refers to a geometric hole generated on a specified geometric body by a Boolean difference operation according to given parameters (such as diameter, depth, etc.). As shown in Table 6 below, different types of hole features have different parameter definitions. For example, a simple hole, its parameters commonly include (but are not limited to) hole diameter, hole depth, and end angle.

[0078] Table 6 Geometric elements of a hole feature

[0079]

[0080] b) Rib Feature

[0081] A feature of a rib-shaped protrusion generated on a geometric body. As shown in Table 7 below, the parameters of a rib feature commonly include (but are not limited to) rib sketch, thickening direction, thickening type, and rib thickness.

[0082] Table 7 Geometric elements of a rib feature

[0083]

[0084] c) Thread Feature

[0085] A feature that represents a thread feature on a cylindrical or conical geometric surface. As shown in Table 8 below, the commonly used parameters of a thread feature include (but are not limited to) major diameter, minor diameter, pitch, number of threads, lead, thread type, thread engagement length, and thread rotation direction, etc.

[0086] Table 8 Geometric elements of a thread feature

[0087]

[0088] d) Round Feature

[0089] A feature with round feature is created at the junction of different surfaces on a geometric body. The parameters of round feature include (but not limited to) round edge, round radius.

[0090] e) Chamfer Feature

[0091] A feature with chamfer feature is created at the junction of different surfaces on a geometric body. For chamfer feature, the parameters of different chamfer types are different, as shown in Table 9 below. For example, the parameters of equal chamfer (DxD type) include (but not limited to) chamfer edge, chamfer distance.

[0092] Table 9 Chamfer Feature Geometric Element

[0093] Serial number Label parameter Description Restriction condition 1 Chamfered edge Edge with chamfered transition -- 2 Chamfer distance D Chamfer distance D>0

[0094] f) Shell Feature

[0095] A feature with shell feature is created by hollowing a geometric body into a shell with a certain thickness and direction. As shown in Table 10 below, the parameters of shell feature include shell face, shell thickness, etc.

[0096] Table 10 Shell Feature Geometric Element

[0097] Serial number Label parameter Description Restriction condition 1 Shell thickness S Shell thickness S > 0 or S < 0 2 Shell surface SF Shell surface --

[0098] g) Draft Feature

[0099] A feature with draft feature is created on a face or a series of faces on a geometric body with given parameters. As shown in Table 11 below, the parameters of draft feature often include neutral face (edge), draft degree, draft face, draft direction.

[0100] Table 11 Draft Feature Geometric Element

[0101] Serial number Parameter Description Restriction condition 1 Neutral surface (edge) Neutral surface (edge) -- 2 Draft degree Draft angle 30°>draft degree>0 3 Draft surface Finish surface One or more 4 Draft direction Draft direction --

[0102] Establishing a Hierarchical Model of Part Geometric Information

[0103] During the design in a three-dimensional design software, the feature modeling process not only records the feature information of a part, but also records the point, line, face and their topological information of the part. Therefore, the three-dimensional part feature library and feature labeling rule library formed by the feature modeling process provide a technical basis for automatic dimension labeling. Through the hierarchical model of "part-feature-geometry", the feature information and geometric information of a part can be obtained and organized by layers, as shown in Table 12 below. Figure 4

[0104] ​The geometric layer mainly describes the basic geometric elements such as points, lines and surfaces of the part and the topological relationship between the geometric elements, and is a component element of the part feature model. The feature layer mainly describes the sub-features and the mutual relationship between the features, and is the core of the part feature modeling and the basic unit of the automatic dimension marking.

[0105] Part geometric element hierarchical clustering method based on modeling features

[0106] The classification of part geometric elements is an unsupervised learning problem. The purpose of geometric element classification is to classify the points, lines, surfaces and their mutual topological information of the part by using a set of rules and an unsupervised learning algorithm: to identify geometric elements with similar distances in the feature space and assign them to clusters or groups with similar elements, and the clusters or groups of similar geometric elements constitute the modeling features of the part, which are the basic units of automatic three-dimensional marking, as shown in Figure 5 .

[0107] The agglomerative hierarchical clustering (AGNES) algorithm works in a bottom-up manner, first letting all geometric elements become a separate cluster, and then combining them through "similarity" until there is only one cluster left, as shown in Figure 6 . Using the agglomerative hierarchical clustering algorithm, it is assumed that there is a hierarchical structure between the classes of geometric elements, and the samples are clustered into hierarchical classes. A feature recognition algorithm is used to identify the subset of geometric elements that can constitute the modeling features, and a merging method is used to cluster the samples from each sample to a cluster to all samples in a cluster. The agglomerative hierarchical clustering method is established to classify the geometric elements of the three-dimensional part in the geometric space, and the result of the classification forms a tree-shaped cluster of geometric elements, and finally the hierarchical modeling features of the part are realized.

[0108] In the geometric space, the geometric elements of the part are clustered by the agglomerative hierarchical clustering method, and the feature structure tree of the three-dimensional model is finally established, as shown in Figure 7 .

[0109] The hierarchical geometric element cluster tree of the three-dimensional model is established by the agglomerative hierarchical clustering algorithm. The classification rule of each layer uses a feature recognition algorithm, and the main steps of the algorithm are:

[0110] a) According to the similarity relationship between the features, the geometric elements are discriminated one by one using a convolutional neural network; the geometric elements contained in the same modeling feature are classified into a class, so that the geometric element set is classified into n classes;

[0111] b) The geometric elements contained in any two classes in the previous step are jointly input into a convolutional neural network for recognition, and the similarity to each modeling feature in the feature library is calculated. The highest similarity means the shortest distance between the two classes, thereby merging the two classes with the shortest distance to form a new class.

[0112] c) Repeat the previous step to calculate the distance between the new class and each class in the layer. If the number of classes is 1, terminate the calculation, otherwise repeat the previous step.

[0113] The AgglomerativeClustering class in the open-source machine learning library sklearn has implemented the AGNES algorithm framework.

[0114] The inter-class distance calculation method of the AGNES algorithm is changed to the feature similarity index described in the present application, thereby completing the feature-based hierarchical clustering of the geometric elements of the part.

[0115] Feature recognition of three-dimensional parts using convolutional neural network (CNN)

[0116] Features are composed of specific positional relationships of geometric elements and a certain number of faces and edges. Based on the geometric information and topological information of the features generated during modeling, a convolutional neural network is built based on the open-source neural network architecture TensorFlow and Keras, as shown in Figure 8 The feature library of the three-dimensional model formed by feature modeling is used for model training to complete the recognition of the modeling features of the three-dimensional model.

[0117] Vectorization of the geometric element set of the three-dimensional model and feature label encoding

[0118] Current deep learning networks all use tensors as the basic data structure. Therefore, the geometric element set of the three-dimensional model needs to be represented as vector data and encoded first. Since the feature recognition of the three-dimensional model is a multi-classification problem, the feature label list of the model is one-hot encoded,

[0119] Feature recognition of three-dimensional parts is a single-label multi-classification problem, and each sample has only one class. The last layer of the neural network uses a fully connected layer (Dense layer) with a number of units equal to the number of classes of the model features.

[0120] Compile the model. Since the feature categories of the three-dimensional parts are encoded using one-hot during the calculation of the loss function, the sparse_categorical_crossentropy is used as the loss function to train the convolutional neural network.

[0121] Optimizer, an improved variant of stochastic gradient descent (SGD) optimization method, an optimization method suitable for convolutional neural networks; using a categorical cross-entropy loss function, suitable for multi-classification problems.

[0122] Training a convolutional neural network

[0123] Input the training sample set into the convolutional neural network built in the previous step for training, obtain a trained model, and save the model. During the training process, overfitting may occur after a certain generation of training. A callback function is used, and this framework automatically saves the model with the optimal number of iterations during the training process.

[0124] Feature recognition of three-dimensional parts using a convolutional neural network

[0125] Read the geometry element set of the three-dimensional model to be recognized, use the trained convolutional neural network model for feature recognition, and establish a feature structure tree according to the steps of agglomerative hierarchical clustering.

[0126] Automatic labeling of three-dimensional parts

[0127] For automatic labeling of the model, first, all features of the part are obtained, and a feature structure tree of the three-dimensional model is established. Then, each feature is traversed in turn, and according to the established feature labeling rule library, each feature is labeled respectively, and the positioning between features is labeled, thereby completing the dimension labeling of the entire model part.

[0128] For the labeling of the dimension of the shape, the feature labeling rule established by the present application is used for labeling.

[0129] For the positioning dimension labeling between features, the features need to be prioritized. After sorting the features according to the hierarchical three-dimensional feature tree structure obtained by hierarchical clustering, the features are labeled in turn. The priority of features of the same type is the same, and the priority of features of different types is different. The value determines the order of feature dimension labeling. Features of the same type are added to the corresponding array. When sorting, the priority is from large to small, and when the priority is the same, the axial coordinate value is from small to large. Finally, the positioning dimension labeling between features is completed.

[0130] Automatic labeling of three-dimensional models using three-dimensional design software open interface

[0131] On the basis of the above research work, by building a three-dimensional design software secondary development environment, using the rules and methods described in the present application, automatic dimension labeling of three-dimensional parts can be realized.

[0132] Embodiment

[0133] As Figure 9A barrel-shaped tooling kit is shown, with an inner diameter of 500 mm, an outer diameter of 600 mm, and a height of 470 mm. The barrel has an array of holes with a diameter of 40 mm in the radial direction and an array of holes with a diameter of 20 mm in the axial direction. In the three-dimensional design software, it is built through 11 feature modeling processes such as sketch features, stretch features, array features, feature merging, and mirror features, as shown in Figure 10 .

[0134] From Figure 10 , it can be seen that for a specific three-dimensional part, the user can clearly understand the specific types of features 1 to 11, but for the automatic labeling algorithm, based on the existing secondary development functions of the three-dimensional design software, only the number of features of the barrel-shaped tooling part is 11, and the specific feature types cannot be obtained.

[0135] Extraction of three-dimensional part modeling features and geometric elements

[0136] The current three-dimensional design software provides secondary development interfaces, which facilitates users to customize their needs based on the business needs. Build a secondary development environment for three-dimensional design software, obtain the feature set (feature name unknown) and geometric element set of the three-dimensional part, and store them in a structured manner.

[0137] Figure 9 The feature set, geometric element set and their mutual relationship of the barrel-shaped tooling part shown in Table 12 are shown in Table 12.

[0138] Table 12 Feature set and geometric element set of barrel-shaped tooling part

[0139]

[0140]

[0141] It is worth noting that only the number and number of features can be obtained, and the specific feature type cannot be obtained. The obtained feature set and geometric element set are stored in a linked list structure.

[0142] Three-dimensional part feature modeling and establishment of feature labeling rule library

[0143] The feature modeling refers to the establishment of a feature type library for three-dimensional modeling according to GB / T 24734-2009 standard. The feature labeling rule library refers to the establishment of labeling rules for each type of feature based on the feature type library, thereby forming a feature labeling rule library.

[0144] The feature labeling rule library of the three-dimensional part is established based on the features of the three-dimensional part as the basic labeling unit and the geometric element set contained in the features as the labeling object.

[0145] Establish a hierarchical expression model of three-dimensional part features and geometric elements

[0146] Based on the geometric feature set and the geometric element set in table 12, a layered expression model composed of 'part layer-feature layer-geometric element layer' is established, as shown in Figure 11 .

[0147] Three-dimensional part feature recognition method based on feature modeling and feature labeling rule library

[0148] Based on the formed feature type library, a deep learning algorithm is used to recognize the feature types of feature 1 to feature 11 in Figure 11 , and the feature types contained in the barrel-shaped tooling part are recognized.

[0149] Automatic dimension labeling of three-dimensional parts

[0150] After completing feature recognition, based on the formed feature labeling rule library, by building a secondary development environment of three-dimensional design software, using secondary development functions, automatic three-dimensional dimension labeling of the barrel-shaped tooling part is completed.

[0151] Finally, the automatic dimension labeling of the three-dimensional part is completed, as shown in Figure 12 .

[0152] The technical solutions of the present application are not limited to the limitations of the above specific embodiments, and any technical variations made according to the technical solutions of the present application fall within the protection scope of the present application.

Claims

1. A method for automatically marking the dimensions of three-dimensional parts, characterized in that: include: S1. Construct a hierarchical expression model, a feature type library, and a feature annotation rule library. The hierarchical expression model is used to classify geometric elements. The classification results of the hierarchical expression model include part layer, feature layer, and geometry layer. The geometry layer is used to describe the basic geometric elements of the part and the topological relationships between geometric elements. The feature layer is used to describe each sub-feature and the relationship between features. The feature type library includes basic features, additional features and editing features; The feature annotation rule library includes annotation rules corresponding to each feature type; S2. Obtain a 3D part to be annotated, and extract geometric elements of the 3D part to be annotated and structural relationships between the elements as a geometric element set; S3. Using a hierarchical expression model to identify features of the 3D part to be annotated, and establishing a feature structure tree of the 3D part to be annotated; specifically, including: S31. Feature recognition of geometric element sets using deep convolutional neural networks; S32, using a convolutional neural network to discriminate the results of feature recognition one by one according to the similarity relationship between the features; classifying the geometric elements contained in the same modeling feature into one category, thereby classifying the geometric element set into n categories; S33, inputting the geometric elements contained in any two classes into the convolutional neural network for recognition, calculating the similarity with each modeling feature in the feature library, and judging whether to merge the two input classes into a new class based on the similarity; S34, repeatedly executing S33 until the number of classes is 1, forming a feature structure tree; S4. Use the feature annotation rule library and feature structure tree to perform shape dimensioning and positioning dimensioning on the three-dimensional parts to be annotated; specifically: traverse each feature of the feature structure tree, perform shape dimensioning on each feature according to the feature annotation rule library, and when performing positioning dimensioning between features, prioritize the positioning dimensioning between features according to the feature structure tree, and perform positioning dimensioning on the features in turn.

2. The automatic dimensioning method for three-dimensional parts according to claim 1, characterized in that: Extracting the geometric elements of the 3D parts to be annotated and the structural relationships between the elements includes: S21, reading part information of the 3D part to be annotated, where the part information is 3D part parameter information; S22, calling a 3D modeling software function to obtain geometric elements of the 3D part to be annotated and the structural relationships between the elements; S23. Perform structured storage on the geometric elements and structural relationships to form a geometric element structured data set.

Images