Data processing method for building structure model design

Abstract

The invention relates to the technical field of image processing, in particular to a data processing method for building structure model design. Obtaining a plurality of historical product models and a current product design sketch, and respectively constructing corresponding undirected graphs; taking each node in the undirected graph of the current product design sketch as a first initial node, and obtaining a path set of the first initial node; acquiring a second initial node corresponding to the first initial node, and acquiring a path set of the second initial node; obtaining the similarity between the first initial node and each corresponding second initial node; arranging the part types corresponding to the second initial nodes to obtain a type recommendation sequence of the part types corresponding to the first initial nodes; and obtaining a model recommendation sequence of each part type in the current product design sketch to carry out product model design. According to the method, the models of the parts in the current product design sketch are recommended through the attribute association among the parts in the historical products, so that the model establishment is more convenient.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a data processing method for building structure model design. Background Art

[0002] The building structure model is a kind of product model. In industrial production, the product model plays an important role in the manufacturing industry. It helps production personnel to test the rationality and accuracy of product design. In the design process of product models, with the continuous development of computer technology, product models have gradually changed from being manually manufactured by production designers to being simulated using three-dimensional data models. This not only saves materials for making product models, but also helps to improve the design efficiency of product models, making product models more modifiable during the design process, thereby improving the design efficiency of product models.

[0003] In the process of product design, the rationality and reliability of structural design are more important. At this time, it is necessary to design the product model more accurately and reasonably so that production personnel can estimate the quality of the produced products. In the current product model design process, generally, a product design sketch is drawn first, and then the designer further designs the three-dimensional model of the product according to the two-dimensional design sketch. Since the parts are the units that constitute the product, in order to speed up the design efficiency of the designers, the parts that constitute the three-dimensional model of the product are generally recommended. In the prior art, replaceable parts are usually recommended after the specific properties of the parts of each part of the model have been known. At the same time, when recommending parts in the prior art, the model is selected only according to the properties of the parts themselves, without considering the combination relationship between parts. When replacing, each part is often compared separately, which cannot play a good auxiliary role in the establishment of the three-dimensional model of the product, and may even reduce the efficiency of the entire three-dimensional model establishment. Summary of the invention

[0004] In order to solve the problem that in the prior art, when recommending parts, the model is selected only according to the properties of the parts themselves, the combination relationship between the parts is not considered, and the establishment of the three-dimensional model of the product cannot be well supplemented, the present invention provides a data processing method for building structure model design, including: obtaining multiple historical product models and current product design sketches and constructing corresponding undirected graphs respectively; taking each node in the undirected graph of the current product design sketch as the first initial node, obtaining the path set of the first initial node; obtaining the second initial node corresponding to the first initial node, obtaining the path set of the second initial node; obtaining the similarity between the first initial node and each corresponding second initial node; arranging the part models corresponding to each second initial node to obtain the model recommendation sequence of the part type corresponding to the first initial node; obtaining the model recommendation sequence of each part type in the current product design sketch to design the product model. The present invention recommends the model of the part in the current product design sketch through the attribute association between the parts in the historical products, so that the establishment of the model is more convenient.

[0005] The present invention adopts the following technical solution, a data processing method for building structure model design, comprising: Acquire multiple historical product models, and construct an undirected graph of each historical product model by taking the model number of each part type in each historical product model as a node; Obtain the current product design sketch, and construct an undirected graph of the current product design sketch with each part type in the current product design sketch as a node; Taking each node in the undirected graph of the current product design sketch as a first initial node, and grading the nodes in the undirected graph of the current product design sketch according to the first initial nodes; Starting from the first initial node, sequentially connect the next level nodes until there is no secondary node in the next level node, take the next level node without a secondary node as the terminal node, obtain the path connecting the first initial node to each terminal node, and obtain the path set of the first initial node; Obtain a node of the same part type as that corresponding to the first initial node in the undirected graph of each historical product model as a second initial node corresponding to the first initial node; obtain a path set of each second initial node; The similarity between the first initial node and each corresponding second initial node is obtained according to the number of nodes contained in each path and the attribute association between two nodes in each path in the path set of the first initial node and the path set corresponding to each second initial node; Arrange the part models corresponding to each second initial node from large to small according to the similarity between the first initial node and each corresponding second initial node, and obtain a model recommendation sequence of the part type corresponding to the first initial node in the current product design sketch; Product model design is performed based on the model recommendation sequence for each part type in the current product design sketch.

[0006] Furthermore, a data processing method for building structure model design, a method for grading nodes in an undirected graph of a current product design sketch according to a first initial node is: In the undirected graph of the current product design sketch, the node directly connected to the first initial node is taken as a first-level node; In the undirected graph of the current product design sketch, the ungraded nodes directly connected to each first-level node are used as second-level nodes; All nodes in the undirected graph of the current product design sketch are graded in turn until all nodes have completed the grade division.

[0007] Furthermore, a data processing method for building structure model design, a method for obtaining a path connecting a first initial node to each terminal node is: Connect each first-level node starting from the first initial node to obtain multiple first-node paths; Multiple second node paths are obtained by connecting each first-level node to the corresponding second-level node; Connect downward in sequence until the next level node is a terminal node, obtain all node paths from the first initial node to each terminal node, and obtain the path from the first initial node to each terminal node.

[0008] Furthermore, in a data processing method for designing a building structure model, the method for obtaining the path set of each second initial node is the same as the method for obtaining the path set of the first initial node.

[0009] Furthermore, in a data processing method for building structure model design, a method for obtaining the attribute correlation between two nodes connected in sequence in each path is as follows: According to the number of occurrences of each connection relationship between the part types corresponding to the two nodes connected in sequence in each path in the historical product model, the attribute association degree between the two nodes connected in sequence in each path is obtained; wherein the connection relationship includes attribute association connection and non-attribute association connection.

[0010] Furthermore, in a data processing method for building structure model design, a method for obtaining the similarity between a first initial node and each corresponding second initial node is: in, represents the similarity between the first initial node and the second initial node in the undirected graph of the p-th historical product model, represents the number of paths in the path set of the first initial node, represents the number of paths in the path set of the second initial node in the undirected graph of the p-th historical product model, Represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model.

[0011] Furthermore, in a data processing method for building structure model design, a method for obtaining the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model is as follows: in, represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model, represents the maximum level of the terminal nodes in the jth path of the second initial node in the undirected graph of the pth historical product model, represents the maximum degree of the terminal nodes in the ith path of the first initial node, represents the level of the node, The first path of the second initial node in the undirected graph of the p-th historical product model nodes, The first path of the second initial node in the undirected graph of the p-th historical product model nodes, The first path of the second initial node in the undirected graph of the p-th historical product model The node and The attribute correlation between nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path The node and The attribute correlation between nodes, The first path of the second initial node in the undirected graph of the p-th historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, The first path of the second initial node in the undirected graph of the p-th historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, is the exclusive OR operator.

[0012] The beneficial effects of the present invention are as follows: the present invention first establishes an undirected graph according to the part models in the historical product model, thereby comparing it with the undirected graph constructed by the part types in the current design sketch, which is an important basis for the combined use of parts in the present invention. The nodes in the undirected graph are graded according to the part types of unknown models in the current sketch, thereby specifically reflecting the degree of connection between each part and other parts, that is, it can reflect the part combination structure based on each part, and find similar structures in the undirected graph of the historical product model. Finally, the part models are sorted by similarity values, so that each part in the current design sketch can obtain the best part model recommendation in the historical model based on the combination information of the structure between the parts, so that the recommendation of the part model is more in line with the design logic of the model, and the efficiency of establishing the three-dimensional model is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0014] Figure 1 The present invention is a flowchart of a data processing method for building structure model design according to an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] like Figure 1 As shown, a schematic diagram of a data processing method for building structure model design according to an embodiment of the present invention is provided, comprising: 101. Obtain multiple historical product models and current product design sketches, and construct an undirected graph of each historical product model and an undirected graph of the current product design sketch respectively; The present invention first obtains multiple historical product models, and then analyzes the two-dimensional design sketch of each historical product model. Since the historical product model is a product that has been designed, the historical product model can be used to obtain information on the types of parts, assembly connection information between parts, and model information corresponding to the assembly of parts. In order to improve the efficiency of identifying information, the present invention uses a DNN network to identify the two-dimensional design sketch of the historical product model, and uses the information on the types of parts, assembly connection information between parts, and model information corresponding to the assembly of parts in the historical product model to train the DNN network, and uses the trained DNN network to identify the types of parts, assembly connection information between parts, and model information corresponding to the assembly of parts. It should be noted that the DNN network used in the present invention is an existing recognition network, that is, any DNN network used for identification and acquisition of information in the prior art can be applied to the present invention. The specific training method of the DNN network can refer to the prior art, and the present invention does not make any limitations or changes.

[0017] The specific information of the parts in the current product design sketch does not need to be obtained through the DNN network. Since the current product design sketch only contains the assembly information between the parts and the part types, the present invention needs to compare the information of the parts with the historical product model, so as to recommend the model based on the similar part structure in the historical product model.

[0018] For the historical product model Parts Type For example, the connection and assembly methods used between it and other types of parts in model design are usually divided into two types, one is non-attribute association connection, and the other is attribute association connection; non-attribute association connection means that the connection method of the two parts will not have any mutual influence of attributes, such as welding, bonding, etc. This connection method means that no matter what the specifications and attributes of the two parts themselves are, they will not affect each other. Therefore, when recommending part models, for parts without attribute association connection, the influence of their structural information on model recommendation is relatively low, and the other is attribute association connection, which means that the connection method of the two parts will have mutual influence of attributes, such as screw connection. This connection method will unify the specification attributes of the two parts. For example, the inner diameter of the nut must be unified with the outer diameter of the bolt. Therefore, when the structural information between the parts is attribute association connection, when recommending the model of the part, the structural information between the part and other parts will be combined. Therefore, by calculating the attribute association between different types of parts, the degree of mutual influence of specification attributes between different types of parts can be measured. The specific calculation process is: in, Indicates the category number of the part, i.e. The corresponding parts type, Indicates the type of parts , In the historical product model, the connection between the two is the number of attribute association connections and non-attribute association connections. Indicates the type of parts , The attribute correlation between the two parts is normalized data. The closer it is to 1, the more attribute correlation there is, and the closer it is to 0, the more non-attribute correlation there is. It can be seen that the more times the non-attribute correlation between the two parts is connected, the more it indicates that there is no necessary connection between the model matching of the two parts, that is, there is no fixed model matching. Therefore, the attribute correlation between the two parts will be lower, and the influencing factor in the subsequent part model recommendation will be lower. Conversely, the more times the attribute correlation between the two parts is connected, the more it indicates that there may be a fixed model matching between the two parts, that is, when one part model is selected, the model of the other part will most likely be determined accordingly. Therefore, the influencing factor in the subsequent part model recommendation will be higher.

[0019] For the historical product model, its part type is the attribute of the part itself. Since the present invention needs to recommend part models through the historical product model, the model of its part type is used as the undirected graph node, and the part type nodes with connection relationships are connected. The attribute correlation between the part types corresponding to the two nodes is used as the edge weight between the nodes to construct an undirected graph for each historical product model.

[0020] For the current design sketch, since the specific model of each part cannot be obtained directly, for example, the part type is a nut, but there is no specific model information, the part's own attributes, that is, the type of each part, are used as nodes of the undirected graph. Similarly, the part type nodes with connection relationships are connected, and the attribute correlation between the part types corresponding to the two nodes is used as the edge weight between the nodes to construct the undirected graph corresponding to the current design sketch.

[0021] 102. Taking each node in the undirected graph of the current product design sketch as the first initial node, grading the nodes in the undirected graph of the current product design sketch; Since the design purpose of the present invention is to find an undirected graph structure similar to the undirected graph corresponding to the current design sketch from the undirected graph of the historical product model, so as to make specific part model recommendations, the undirected graph corresponding to the design sketch For example, the designer needs to select a part type node as the design node (i.e., the first starting node or the second starting node in the present invention), and then select an undirected graph structure with nodes of the same part type as the design node from the historical database as comparison data, and use all nodes of the same part type as the design node on the undirected graph in the comparison data (i.e., the undirected graph of the historical product model) as reference nodes. At this time, the part type structure undirected graph corresponding to the design sketch The part types of the design nodes on are the same as the part types of the reference nodes of each undirected graph in the comparison data.

[0022] For the undirected graph of the pth historical product model in the comparison data ( , For the total number of data undirected graphs) and the undirected graph G of the current design sketch, the design node is taken as the starting node, and the first-level node is calibrated first. The first-level node is the node directly connected to the design node. After the first-level node is calibrated, the second-level node is calibrated. The second-level node is the uncalibrated node directly connected to the first-level node. After the second-level node is calibrated, the third-level node is calibrated. The third-level node is the uncalibrated node directly connected to the second-level node. The calibration continues until and All nodes above are calibrated, thus completing the level division of nodes in the undirected graph.

[0023] The method for grading the nodes in the undirected graph of the current product design sketch according to the first initial node is: In the undirected graph of the current product design sketch, the node directly connected to the first initial node is taken as a first-level node; In the undirected graph of the current product design sketch, the ungraded nodes directly connected to each first-level node are used as second-level nodes; All nodes in the undirected graph of the current product design sketch are graded in turn until all nodes have completed the grade division.

[0024] 103. Obtain a path set of the first initial node and a path set of each second initial node; After calibration and After finding all the nodes, start from the design node to find the secondary node, that is, start from the first starting node or the second starting node, find the first-level node, the second-level node, and so on to form a node path until there are no connected secondary nodes around the node. At this time, stop searching and use the current node as the path end node. At this time, a path containing the level order will be formed from the starting node to the path end node. This path represents the connection characteristics of the first starting node or the second starting stage in the undirected graph. By searching and Design all paths of nodes on the and The path set corresponding to the above designed node.

[0025] The method for obtaining the path connecting the first initial node to each terminal node is: Connect each first-level node starting from the first initial node to obtain multiple first-node paths; Multiple second node paths are obtained by connecting each first-level node to the corresponding second-level node; Connect downward in sequence until the next level node is a terminal node, obtain all node paths from the first initial node to each terminal node, and obtain the path from the first initial node to each terminal node.

[0026] The method for obtaining the path set of each second initial node is the same as the method for obtaining the path set of the first initial node.

[0027] 104. Obtain the similarity between the first initial node and each corresponding second initial node; The method for obtaining the attribute association between two nodes connected in sequence in each path is as follows: According to the number of occurrences of each connection relationship between the part types corresponding to the two nodes connected in sequence in each path in the historical product model, the attribute association degree between the two nodes connected in sequence in each path is obtained; wherein the connection relationship includes attribute association connection and non-attribute association connection.

[0028] The method for obtaining the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model is: in, represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model, represents the maximum level of the terminal nodes in the jth path of the second initial node in the undirected graph of the pth historical product model, represents the maximum degree of the terminal nodes in the ith path of the first initial node, represents the level of the node, The first path of the second initial node in the undirected graph of the p-th historical product model nodes, The first path of the second initial node in the undirected graph of the p-th historical product model nodes, The first path of the second initial node in the undirected graph of the p-th historical product model The node and The attribute correlation between nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path The node and The attribute correlation between nodes, The first path of the second initial node in the undirected graph of the p-th historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, The first path of the second initial node in the undirected graph of the p-th historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, is the exclusive OR operator.

[0029] Indicates the path With path The product of the attribute association between nodes of the same level above, which means that if two nodes are more similar, that is, The more the node path on The node path reference on the node path, and the condition for being a reference is that the attribute correlation between the corresponding part types of the two nodes should be close to 1, that is, if the attribute correlation between the two path segments is small, it means that the corresponding model of the part type on the starting node is not greatly affected by the structure, that is, the reference degree is not large, so if The closer it is to 1, the more it indicates the path With path At the node level arrive The greater the degree of mutual reference on this path, and Indicates that this node level is arrive The node parts of this path should be of the same type. If they are different, there is no reference. If they are the same, it will be explained. Makes sense.

[0030] The method for obtaining the similarity between the first initial node and each corresponding second initial node is: in, represents the similarity between the first initial node and the second initial node in the undirected graph of the p-th historical product model, represents the number of paths in the path set of the first initial node, represents the number of paths in the path set of the second initial node in the undirected graph of the p-th historical product model, represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model, is a normalized value. The closer it is to 1, the and The more similar, that is, by comprehensively considering the similarity of all paths of the same design node in the undirected graph of the current design sketch and the undirected graph of the historical product model, as the similarity between the first initial node in the undirected graph of the current design sketch and the second initial node in the undirected graph of the historical product model, to determine whether the part model corresponding to the second initial node in the undirected graph of the historical product meets the modeling needs of the current design sketch.

[0031] It is worth noting that if there are multiple nodes in a comparison data undirected graph that meet the conditions for being a reference node, the similarity between each qualified node and the reference node is calculated separately, and the reference node setting method with the greatest similarity is selected as the final result.

[0032] 105. Obtain the recommended model sequence for each part type in the current product design sketch to design the product model.

[0033] By calculating the undirected graph of the current design sketch The similarity between each design node in the undirected graph and the corresponding design node in all the comparison data (i.e., historical product models) is calculated, and then arranged in descending order according to the similarity, thus completing the recommended sorting of the corresponding models for each part type in the historical product model. When the designer selects the model of the part in the current design sketch, after selecting a part model according to the ranking in the recommendation list, the next part model can be obtained in the model recommendation sequence with the same ranking. At the same time, certain part model adjustments can be made according to the specific situation, thereby realizing the design of the entire product model.

[0034] The present invention first establishes an undirected graph according to the part models in the historical product model, thereby comparing it with the undirected graph constructed by the part types in the current design sketch, which is an important basis for the combined use of parts in the present invention. The nodes in the undirected graph are graded according to the part types of unknown models in the current sketch, thereby specifically reflecting the degree of connection between each part and other parts, that is, it can reflect the part combination structure based on each part, and find similar structures in the undirected graph of the historical product model. Finally, the part models are sorted by similarity values, so that each part in the current design sketch can obtain the best part model recommendation in the historical model based on the combination information of the structure between the parts, so that the recommendation of the part model is more in line with the design logic of the model, and the efficiency of establishing the three-dimensional model is improved.

[0035] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A data processing method for building structure model design, characterized in that: include: Acquire multiple historical product models, and construct an undirected graph of each historical product model by taking the model number of each part type in each historical product model as a node; Obtain the current product design sketch, and construct an undirected graph of the current product design sketch with each part type in the current product design sketch as a node; Taking each node in the undirected graph of the current product design sketch as a first initial node, and grading the nodes in the undirected graph of the current product design sketch according to the first initial nodes; Starting from the first initial node, sequentially connect the next level nodes until there is no secondary node in the next level node, take the next level node without a secondary node as the terminal node, obtain the path connecting the first initial node to each terminal node, and obtain the path set of the first initial node; Obtain a node in the undirected graph of each historical product model that has the same part type as that corresponding to the first initial node, as the second initial node corresponding to the first initial node; Get the path set of each second initial node; The similarity between the first initial node and each corresponding second initial node is obtained according to the number of nodes contained in each path and the attribute association between two nodes in each path in the path set of the first initial node and the path set corresponding to each second initial node; Arrange the part models corresponding to each second initial node from large to small according to the similarity between the first initial node and each corresponding second initial node, and obtain a model recommendation sequence of the part type corresponding to the first initial node in the current product design sketch; Product model design is performed based on the model recommendation sequence for each part type in the current product design sketch.

2. A data processing method for building structure model design according to claim 1, characterized in that: The method for grading the nodes in the undirected graph of the current product design sketch according to the first initial node is: In the undirected graph of the current product design sketch, the node directly connected to the first initial node is taken as a first-level node; In the undirected graph of the current product design sketch, the ungraded nodes directly connected to each first-level node are used as second-level nodes; All nodes in the undirected graph of the current product design sketch are graded in turn until all nodes have completed the grade division.

3. A data processing method for building structure model design according to claim 2, characterized in that: The method for obtaining the path connecting the first initial node to each terminal node is: Connect each first-level node starting from the first initial node to obtain multiple first-node paths; Multiple second node paths are obtained by connecting each first-level node to the corresponding second-level node; Connect downward in sequence until the next level node is a terminal node, obtain all node paths from the first initial node to each terminal node, and obtain the path from the first initial node to each terminal node.

4. A data processing method for building structure model design according to claim 1, characterized in that: The method for obtaining the path set of each second initial node is the same as the method for obtaining the path set of the first initial node.

5. The data processing method for building structure model design according to claim 1, characterized in that: The method for obtaining the attribute association between two nodes connected in sequence in each path is as follows: According to the number of occurrences of each connection relationship between the part types corresponding to the two nodes connected in sequence in each path in the historical product model, the attribute association degree between the two nodes connected in sequence in each path is obtained; wherein the connection relationship includes attribute association connection and non-attribute association connection.

6. A data processing method for building structure model design according to claim 1, characterized in that: The method for obtaining the similarity between the first initial node and each corresponding second initial node is: in, represents the similarity between the first initial node and the second initial node in the undirected graph of the p-th historical product model, represents the number of paths in the path set of the first initial node, represents the number of paths in the path set of the second initial node in the undirected graph of the p-th historical product model, Represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model.

7. A data processing method for building structure model design according to claim 6, characterized in that: The method for obtaining the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model is: in, represents the similarity between the i-th path of the first initial node and the j-th path of the second initial node in the undirected graph of the p-th historical product model, represents the maximum number of terminal nodes in the jth path from the second initial node in the undirected graph of the pth historical product model, represents the maximum degree of the terminal nodes in the ith path of the first initial node, represents the level of the node, The first path of the second initial node in the undirected graph of the pth historical product model nodes, The first path of the second initial node in the undirected graph of the pth historical product model nodes, The first path of the second initial node in the undirected graph of the pth historical product model The node and The attribute correlation between nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path nodes, Indicates the first initial node in the i-th path The node and The attribute correlation between nodes, The first path of the second initial node in the undirected graph of the pth historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, The first path of the second initial node in the undirected graph of the pth historical product model The part type corresponding to each node, Indicates the first initial node in the i-th path The part type corresponding to each node, is the exclusive OR operator.

Images