CN115238591B - Dynamic parameter checking and driving CAD automatic modeling engine system - Google Patents
Dynamic parameter checking and driving CAD automatic modeling engine system Download PDFInfo
- Publication number
- CN115238591B CN115238591B CN202210966599.9A CN202210966599A CN115238591B CN 115238591 B CN115238591 B CN 115238591B CN 202210966599 A CN202210966599 A CN 202210966599A CN 115238591 B CN115238591 B CN 115238591B
- Authority
- CN
- China
- Prior art keywords
- product
- training
- requirement
- unit
- understanding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to the technical field of software modeling, and particularly discloses a dynamic parameter checking and driving CAD automatic modeling engine system which can be used for driving a CAD system to automatically model based on intelligent analysis and checking of user requirements, so that an accurate and reasonable drawing is obtained, and the efficiency of workers is improved.
Description
Technical Field
The present application relates to the field of software modeling technologies, and more particularly, to a dynamic parameter verification and driving CAD automatic modeling engine system.
Background
The advent of computer aided design software (e.g., CAD software) has brought many conveniences to designers. However, the existing computer aided design software is not intelligent. Specifically, because the manual mechanical design is easy to make mistakes, the parameter relationships are complicated, the efficiency of designing the CAD drawing is extremely low, the accuracy cannot be guaranteed, the complete cycle of an accurate and proper drawing is long, and a great deal of effort and manpower are needed to complete the drawing.
Therefore, a more intelligent CAD automatic modeling engine scheme is desired, which can drive the CAD system to perform the automatic modeling engine to obtain an accurate and reasonable drawing, so as to improve the efficiency of the worker.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a dynamic parameter verification and driving CAD automatic modeling engine system, which uses a trained context encoder model containing an embedded layer and a trained convolutional neural network model with a multi-scale convolutional structure to perform context encoding and multi-scale feature extraction on a product requirement input by a user, then uses the trained context encoder containing the embedded layer to perform context encoding on a retrieval result to be matched, and finally fuses a multi-scale product requirement association understanding feature matrix and a retrieval result understanding feature vector, and obtains a classification result used for representing whether the retrieval result to be matched is matched with the product requirement input by the user through a classifier, so that a CAD block is more accurately matched based on the user requirement.
According to one aspect of the present application, there is provided a dynamic parameter verification and driving CAD automatic modeling engine system, which includes:
the system comprises a user demand acquisition unit, a product customization parameter template form acquisition unit and a product customization parameter acquisition unit, wherein the user demand acquisition unit is used for acquiring product demands input by a user, and the product demands comprise product models and demand parameters input by the user in the product customization parameter template form;
a user requirement semantic coding unit, configured to pass the product model and the requirement parameters in the product requirement through a trained context encoder including an embedded layer to obtain a plurality of product requirement understanding feature vectors;
the user demand representation unit is used for arranging the product demand understanding feature vectors into a two-dimensional feature matrix and then obtaining a multi-scale product demand association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolutional structure;
the device comprises a to-be-matched retrieval result acquisition unit, a matching unit and a matching unit, wherein the to-be-matched retrieval result acquisition unit is used for acquiring a to-be-matched retrieval result, and the to-be-matched retrieval result comprises text description, block parameters and block coordinates of a CAD block;
a to-be-matched retrieval result encoding unit, configured to pass the text description of the CAD block, the block parameter, and the block coordinate in the to-be-matched retrieval result through the trained context encoder including the embedded layer to obtain a plurality of retrieval result label understanding feature vectors, and cascade the plurality of retrieval result label understanding feature vectors to obtain a retrieval result understanding feature vector;
the measurement fusion unit is used for multiplying the multi-scale product demand association understanding feature matrix and the retrieval result understanding feature vector to obtain a classification feature vector; and
and the modeling result generating unit is used for enabling the classification characteristic vectors to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the retrieval result to be matched is matched with the product requirement input by the user or not.
In the above CAD automatic modeling engine system for dynamic parameter verification and driving, the user requirement semantic coding module includes:
a product requirement embedding vector generating unit, configured to convert the product model and the requirement parameters in the product requirement into a sequence of product requirement embedding vectors using the trained embedding layer of the context encoder, respectively;
a product requirement understanding feature vector generating unit, configured to perform global context-based semantic coding on the sequence of product requirement embedding vectors using the trained Bert model of the context encoder to obtain the plurality of product requirement understanding feature vectors.
In the above CAD-automatic dynamic parameter verification and driving system, the user requirement representing module is further configured to perform, in forward pass of layers, respectively on the input data using each layer of the convolutional neural network model with the multi-scale convolutional structure:
performing convolution processing on the input data based on a first convolution core to obtain a first convolution feature map;
performing convolution processing on the input data based on a second convolution kernel to obtain a second convolution characteristic diagram;
performing convolution processing on the input data based on a third convolution kernel to obtain a third convolution characteristic diagram;
performing convolution processing on the input data based on a fourth convolution kernel to obtain a fourth convolution feature map, wherein the first convolution kernel, the second convolution kernel, the third convolution kernel and the fourth convolution kernel have different sizes;
cascading the first convolution feature map, the second convolution feature map, the third convolution feature map and the fourth convolution feature map to obtain a multi-scale convolution feature map;
performing mean pooling along channel dimensions on the multi-scale convolution feature map to obtain a pooled feature map; and
carrying out nonlinear activation processing on the pooled feature map to obtain an activated feature map;
wherein the output of the last layer of the convolutional neural network model with the multi-scale convolutional structure is the multi-scale product demand correlation understanding feature matrix.
In the above CAD automatic modeling engine system for dynamic parameter verification and driving, the module for encoding the search result to be matched includes:
an embedding vectorization unit, configured to convert, using the trained embedding layer of the context encoder, the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched into a search result embedding vector to obtain a sequence of search result embedding vectors;
a context coding unit, configured to perform global context-based semantic coding on the sequence of the search result embedded vectors using a trained Bert model of the context coder to obtain a plurality of search result label understanding feature vectors; and
and the cascading unit is used for cascading the plurality of feature vectors to obtain the retrieval result understanding feature vector.
In the above CAD engine system for dynamic parameter verification and driving, the modeling result generating module is further configured to: processing the classification feature vector by using the classifier according to the following formula to obtain the classification result;
wherein the formula is:softmax{(W n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, beta 1 To B n Is the bias vector and X is the classification feature vector.
In the above dynamic parameter verification and driving CAD automatic modeling engine system, further comprising a training module, configured to train the convolutional neural network model with the multi-scale convolutional structure and the context encoder including the embedded layer;
wherein the training module comprises:
the system comprises a training data acquisition unit, a searching unit and a matching unit, wherein the training data acquisition unit is used for acquiring product requirements input by a user and search results to be matched, the product requirements comprise product models and required parameters input by the user in a product self-defined parameter template form, and the search results to be matched comprise text description, block parameters and block coordinates of CAD blocks;
a training user requirement semantic coding unit, configured to pass the product model and the requirement parameters in the product requirement through an untrained context encoder including an embedded layer to obtain a plurality of training product requirement understanding feature vectors;
the training user requirement representing unit is used for arranging the training product requirement understanding feature vectors into a two-dimensional feature matrix and then obtaining a training multi-scale product requirement association understanding feature matrix through the untrained convolutional neural network model with the multi-scale convolutional structure;
the training search result coding unit is used for enabling the text description, the block parameters and the block coordinates of the CAD block in the search result to be matched to pass through the untrained context coder comprising the embedded layer to obtain a plurality of training search result label understanding feature vectors, and cascading the plurality of training search result label understanding feature vectors to obtain the training search result understanding feature vectors;
the training metric fusion unit is used for multiplying the training multi-scale product demand association understanding feature matrix and the training retrieval result understanding feature vector to obtain a training classification feature vector;
the classification loss unit is used for enabling the training classification feature vectors to pass through a classifier to obtain a classification loss function value; and
the frugal decomposition incentive loss function value calculation unit is used for calculating a frugal decomposition incentive loss function value of the training multi-scale product requirement association understanding feature matrix, wherein the frugal decomposition incentive loss function value is related to the weighted sum of natural index function values taking the negative value of the feature value of each position in each column vector of the training multi-scale product requirement association understanding feature matrix as power.
In the above CAD engine system, the classification loss unit includes:
a training classification result obtaining unit, configured to input the training classification feature vector into a Softmax classification function of the classifier to obtain a training classification result; and
and the classification loss function value calculating unit is used for calculating a cross entropy value between the training classification result and the real value as the classification loss function value.
In the above dynamic parameter verification and driving CAD automatic modeling engine system, the frugal decomposition incentive loss function value calculating unit is further configured to: calculating a frugal decomposition incentive loss function value of the training multi-scale product demand correlation understanding feature matrix according to the following formula;
wherein the formula is:
wherein m is i,j And associating and understanding the characteristic value of the characteristic matrix for the multi-scale product demand, wherein tau is a penalty factor.
Compared with the prior art, the dynamic parameter verification and driving CAD automatic modeling engine system uses a trained context encoder model containing an embedded layer and a trained convolutional neural network model with a multi-scale convolutional structure to perform context encoding and multi-scale feature extraction on a product requirement input by a user, then uses the trained context encoder containing the embedded layer to perform context encoding on a retrieval result to be matched, and finally fuses a multi-scale product requirement association understanding feature matrix and a retrieval result understanding feature vector, and obtains a classification result used for representing whether the retrieval result to be matched and the product requirement input by the user are matched based on the user requirement through a classifier.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
FIG. 1 illustrates a block diagram schematic of a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application.
FIG. 2 illustrates a block diagram of a user requirement semantic code module in a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application.
Fig. 3 is a block diagram illustrating a to-be-matched search result encoding module in a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application.
Fig. 4 illustrates a block diagram view of a training module in a dynamic parameter verification and driving CAD automated modeling engine system according to an embodiment of the application.
Fig. 5 illustrates a block diagram view of a classification loss element in a dynamic parameter verification and driving CAD automated modeling engine system according to an embodiment of the application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of scenes
Correspondingly, in the technical scheme of the application, the research of the application finds that: an accurate and reasonable drawing consists of a plurality of CAD blocks (basic design elements), and the CAD blocks are distributed at the correct positions (the position relation between the design elements) of the CAD drawing, that is, when a user designs the CAD drawing, the user can select the proper CAD blocks based on the design requirements and arrange the CAD blocks in the CAD drawing in a preset pattern. Therefore, the construction of the CAD automatic modeling engine solution is essentially a CAD block result matching problem based on the user's needs, i.e., the user's product needs are analyzed to obtain the correct CAD block combination strategy result.
Correspondingly, in the technical scheme of the application, firstly, product requirements input by a user are obtained, wherein the product requirements comprise product models and requirement parameters input by the user in a product self-defined parameter template form. Then, the product model number and the requirement parameters in the product requirement pass through a trained context encoder comprising an embedding layer to obtain a plurality of product requirement understanding feature vectors. Namely, semantic understanding is carried out on each description in the product requirement by a semantic encoder to obtain a plurality of product requirement understanding feature vectors. And then, arranging the product demand understanding feature vectors into a two-dimensional feature matrix, and then obtaining a multi-scale product demand association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolution structure. That is, a convolutional neural network model (text convolutional neural network) is used as a feature extractor to extract high-dimensional semantic association information of each item description in the product requirement.
In particular, in the technical solution of the present application, because different semantic associations exist among text descriptions of different spans in the product requirement, for example, there may exist a strong semantic association between a first item parameter and a fifth item parameter, and a weak semantic association between the first item parameter and a second item, in the requirement parameters input by the user in the product customized parameter template form. Therefore, in order to capture semantic association information of different text spans, in the technical solution of the present application, a text convolution neural network with a multi-scale convolution structure is used as a feature extractor to extract high-dimensional semantic association information of each description in a product requirement.
Compared with a conventional statistical rule model, the semantic encoder and the text convolution neural network model can be used for extracting semantic features more implicit in the description of the product requirement so as to improve the accuracy of understanding the product requirement.
And then, performing semantic understanding on the search result to be matched by using a semantic encoder to obtain a search result understanding feature vector, wherein the search result comprises the text description, the block parameters and the block coordinates of the CAD block. Compared with matching in a source domain space, in the technical scheme of the application, the retrieval result to be matched is mapped to the high-dimensional semantic feature space, so that the retrieval result to be matched can be expressed more abundantly, and the adaptation accuracy between the product requirement and the retrieval result to be matched is improved.
Specifically, in a specific example of the present application, the text description of the CAD block, the block parameters, and the block coordinates in the search result to be matched are passed through the trained context encoder including the embedded layer to obtain a plurality of search result label understanding feature vectors, and the plurality of search result label understanding feature vectors are concatenated to obtain the search result understanding feature vector.
Then, matching the multi-scale product demand correlation understanding feature matrix used for representing the high-dimensional feature representation of the product demand and the retrieval result understanding feature vector used for representing the high-dimensional feature representation of the retrieval result to be matched in a high-dimensional feature space. Specifically, the multi-scale product demand association understanding feature matrix is multiplied by the retrieval result understanding feature vector to obtain a classification feature vector, and then the classification feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the retrieval result to be matched is matched with the product demand input by the user or not.
Particularly, in the technical solution of the present application, for a multi-scale product demand association understanding feature matrix obtained by a convolutional neural network model with a multi-scale convolutional layer structure, assuming that a product demand association understanding feature vector is a row vector, the multi-scale product demand association understanding feature matrix represents associated features of parameter semantics between different scales in a column direction, that is, when the multi-scale product demand association understanding feature matrix is optimized, it is desirable to improve a multi-scale-based classification expression capability in the column direction.
Therefore, a parsimonious decomposition incentive loss function is introduced for the multi-scale product demand correlation understanding feature matrix, and is expressed as:
m i,j and associating and understanding the characteristic value of the characteristic matrix for the multi-scale product demand, wherein tau is a penalty factor.
The parsimony decomposition encourages the loss function to be used for grouping the features in the column direction, and imposes punishment on overlapping of elements in a row vector, so as to promote parsimony decomposition of a high-dimensional manifold in the column direction by calculating distance type combination of the symbolization function, and the high-dimensional manifold can be simply understood as being decomposed into a set of convex polyhedrons (covex polytope) in the column direction, so that the dimensionality monotonicity of the multi-scale product demand correlation understanding feature matrix in the column direction is improved, and the classification expression capability of the multi-scale product demand correlation understanding feature matrix is improved. Therefore, the adaptation accuracy between the user requirement and the retrieval result to be matched is improved, and the reasonability and intelligence of automatic drawing construction of the CAD automatic modeling engine system are improved.
Based on this, the present application provides a dynamic parameter verification and driving CAD automatic modeling engine system, which includes: the system comprises a user demand acquisition unit, a product customization parameter template form acquisition unit and a product customization parameter acquisition unit, wherein the user demand acquisition unit is used for acquiring product demands input by a user, and the product demands comprise product models and demand parameters input by the user in the product customization parameter template form; the user requirement semantic coding unit is used for enabling the product model and the requirement parameters in the product requirement to pass through a trained context coder containing an embedded layer so as to obtain a plurality of product requirement understanding feature vectors; the user demand representation unit is used for arranging the product demand understanding feature vectors into a two-dimensional feature matrix and then obtaining a multi-scale product demand association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolutional structure; the device comprises a to-be-matched retrieval result acquisition unit, a matching unit and a matching unit, wherein the to-be-matched retrieval result acquisition unit is used for acquiring a to-be-matched retrieval result, and the to-be-matched retrieval result comprises text description, block parameters and block coordinates of a CAD block; a to-be-matched retrieval result encoding unit, configured to pass the text description of the CAD block, the block parameter, and the block coordinate in the to-be-matched retrieval result through the trained context encoder including the embedded layer to obtain a plurality of retrieval result label understanding feature vectors, and cascade the plurality of retrieval result label understanding feature vectors to obtain a retrieval result understanding feature vector; the measurement fusion unit is used for multiplying the multi-scale product demand association understanding feature matrix and the retrieval result understanding feature vector to obtain a classification feature vector; and the modeling result generating unit is used for enabling the classification characteristic vectors to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the retrieval result to be matched is matched with the product requirement input by the user or not.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary System
FIG. 1 illustrates a block diagram schematic of a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application. As shown in fig. 1, the dynamic parameter verification and driving CAD automatic modeling engine system 100 according to the embodiment of the present application includes: the system comprises a user demand acquisition unit 110, a product demand acquisition unit and a product processing unit, wherein the user demand acquisition unit is used for acquiring product demands input by a user, and the product demands comprise product models and demand parameters input by the user in a product self-defined parameter template form; a user requirement semantic coding unit 120, configured to pass the product model and the requirement parameters in the product requirement through a trained context encoder including an embedded layer to obtain a plurality of product requirement understanding feature vectors; the user demand representation unit 130 is configured to arrange the product demand understanding feature vectors into a two-dimensional feature matrix, and then obtain a multi-scale product demand association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolutional structure; the to-be-matched retrieval result acquisition unit 140 is configured to acquire a to-be-matched retrieval result, where the to-be-matched retrieval result includes a text description, a block parameter, and a block coordinate of a CAD block; a to-be-matched search result encoding unit 150, configured to pass the text description, the block parameters, and the block coordinates of the CAD block in the to-be-matched search result through the trained context encoder including the embedded layer to obtain a plurality of search result label understanding feature vectors, and cascade the plurality of search result label understanding feature vectors to obtain a search result understanding feature vector; the measurement fusion unit 160 is configured to multiply the multi-scale product demand association understanding feature matrix with the retrieval result understanding feature vector to obtain a classification feature vector; and a modeling result generating unit 170, configured to pass the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether the search result to be matched is adapted to the product requirement input by the user.
In this embodiment, the user requirement acquiring unit 110 is configured to acquire a product requirement input by a user, where the product requirement includes a product model and a requirement parameter input by the user in a product-defined parameter template form. As described above, in the technical solution of the present application, it is found through the research of the present application that: an accurate and reasonable drawing is composed of a plurality of CAD blocks (basic design elements), and the CAD blocks are distributed at correct positions (position relations among the design elements) of the CAD drawing, that is, when a user designs the CAD drawing, the user can select proper CAD blocks based on the design requirements and arrange the CAD blocks in the CAD drawing in a preset pattern. Therefore, the construction of the CAD automatic modeling engine solution is essentially a CAD block result matching problem based on the user's needs, i.e., the user's product needs are analyzed to obtain the correct CAD block combination strategy result.
In this embodiment of the present application, the user requirement semantic encoding unit 120 is configured to pass the product model number and the requirement parameters in the product requirement through a trained context encoder including an embedded layer to obtain a plurality of product requirement understanding feature vectors. It should be appreciated that in view of the relationship between product model and demand parameters and between different demand parameters, the product model and the demand parameters are context coded using a converter-based encoder model to mine global association information of the product model and the demand parameters to obtain a plurality of product demand understanding feature vectors. That is, all descriptions in the product requirement are regarded as a text sequence, and a semantic encoder performs semantic understanding on the descriptions in the product requirement to obtain a plurality of product requirement understanding feature vectors.
More specifically, the trained context encoder with the embedded layer is used for carrying out context semantic encoding on the product model and the requirement parameters in the product requirement, and it can be understood that the trained context encoder can extract features which are beneficial to matching judgment between the retrieval result to be matched and the product requirement input by the user.
FIG. 2 illustrates a block diagram of a user requirement semantic code module in a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application. As shown in fig. 2, in a specific embodiment of the present application, the user requirement semantic code module 120 includes: a product requirement embedded vector generating unit 121, configured to use the trained embedded layers of the context encoder to respectively convert the product model and the requirement parameters in the product requirement into a product requirement embedded vector to obtain a sequence of product requirement embedded vectors; a product requirement understanding feature vector generating unit 122, configured to perform global context-based semantic encoding on the sequence of product requirement embedding vectors using the trained Bert model of the context encoder to obtain the plurality of product requirement understanding feature vectors.
In this embodiment of the present application, the context encoder is a Bert model based on a converter, where the Bert model is capable of performing context semantic coding based on the global input sequence on each input quantity in the input sequence based on an intrinsic mask structure of the converter. That is, the converter-based Bert model is able to extract a globally based feature representation of each input quantity in the input sequence. In this embodiment, the Bert model based on the converter can perform global context-based semantic coding on descriptions in the product demand to obtain a plurality of product demand understanding feature vectors, where one feature vector in the plurality of feature vectors corresponds to one description. It should be understood that each product demand understanding feature vector in the plurality of product demand understanding feature vectors is used for representing global context deep implicit features of each item description based on the whole sequence of the product demands.
In this embodiment of the application, the user requirement representing unit 130 is configured to arrange the product requirement understanding feature vectors into a two-dimensional feature matrix, and then obtain a multi-scale product requirement association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolutional structure. It should be appreciated that although the context semantic encoder can extract high-dimensional semantic association information of each description relative to other descriptions in the product requirement, it is granular with a single index, which does not perform well in association information extraction among multiple local indexes. Therefore, in the technical solution of the present application, a convolutional neural network model (text convolutional neural network) is used as a feature extractor to extract high-dimensional semantic associated information of each description in a product requirement.
In particular, in the technical solution of the present application, because different semantic associations exist among text descriptions of different spans in the product requirement, for example, there may exist a strong semantic association between a first item parameter and a fifth item parameter, and a weak semantic association between the first item parameter and a second item, in the requirement parameters input by the user in the product customized parameter template form. Therefore, in order to capture semantic association information of different text spans, convolution kernels of different sizes are required to be used for extracting features from a feature matrix, and in the technical scheme of the application, a trained convolutional neural network model with a multi-scale convolution structure is used as a feature extractor to extract high-dimensional semantic association information of each item description in a product requirement.
In a specific embodiment of the present application, in order to accurately extract semantic association features between adjacent text descriptions, a smaller convolution kernel of 1 × 1 size is added to solve this problem.
The multi-scale convolution kernel effectively solves the problem that the single-size convolution kernel cannot extract features of different scales in the feature matrix, but the convolution kernel of each size still extracts features once in the convolution layer. In order to achieve the purpose of extracting features for multiple times, the probability of packet convolution is provided: and extracting features from the feature maps by utilizing multiple groups of convolutions, dividing a convolution kernel into multiple groups according to channels by grouping convolution, and performing convolution operation on the feature maps respectively.
Compared with a conventional statistical rule model, the semantic encoder and the text convolution neural network model can be used for extracting more implicit semantic features in the description of the product requirement so as to improve the accuracy of understanding the product requirement.
In a specific embodiment of the present application, the user requirement representing module 130 is further configured to perform, in a layer forward pass, the following operations on the input data respectively using the layers of the convolutional neural network model with the multi-scale convolutional structure: performing convolution processing on the input data based on a first convolution core to obtain a first convolution feature map; performing convolution processing on the input data based on a second convolution kernel to obtain a second convolution characteristic diagram; performing convolution processing on the input data based on a third convolution kernel to obtain a third convolution characteristic diagram; performing convolution processing on the input data based on a fourth convolution kernel to obtain a fourth convolution characteristic map, wherein the first convolution kernel, the second convolution kernel, the third convolution kernel and the fourth convolution kernel have different sizes; cascading the first convolution feature map, the second convolution feature map, the third convolution feature map and the fourth convolution feature map to obtain a multi-scale convolution feature map; performing mean pooling along channel dimensions on the multi-scale convolution characteristic map to obtain a pooled characteristic map; carrying out nonlinear activation processing on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network model with the multi-scale convolutional structure is the multi-scale product demand correlation understanding feature matrix.
In this embodiment, the to-be-matched search result acquisition unit 140 is configured to acquire a to-be-matched search result, where the to-be-matched search result includes a text description, a block parameter, and a block coordinate of a CAD block. As described above, the present application is essentially a solution to the problem of matching CAD block results based on user requirements, and therefore, in addition to analyzing the product requirements of the user, the search results to be matched also need to be analyzed and then compared with the product requirements of the user.
In this embodiment of the present application, the to-be-matched search result encoding unit 150 is configured to pass the text description of the CAD block, the block parameter, and the block coordinate in the to-be-matched search result through the trained context encoder including the embedded layer to obtain a plurality of search result label understanding feature vectors, and concatenate the plurality of search result label understanding feature vectors to obtain the search result understanding feature vector. It should be understood that, considering that there is semantic association between the text description of the CAD block, the block parameter and the block coordinate in the search result to be matched, the text description of the CAD block, the block parameter and the block coordinate in the search result to be matched are also regarded as a text sequence, and the trained context encoder including the embedding layer performs context semantic encoding on each description in the search result to be matched to obtain a search result understanding feature vector.
Fig. 3 is a block diagram illustrating a to-be-matched search result encoding module in a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application. As shown in fig. 3, in a specific embodiment of the present application, the to-be-matched search result encoding module 150 includes: an embedded vectorization unit 151, configured to convert, using the trained embedding layer of the context encoder, the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched into a search result embedded vector to obtain a sequence of search result embedded vectors; a context encoding unit 152, configured to perform global context-based semantic encoding on the sequence of search result embedded vectors using the trained transformer-based Bert model of the context encoder to obtain a plurality of search result label understanding feature vectors; and a concatenation unit 153, configured to concatenate the plurality of feature vectors to obtain the retrieval result understanding feature vector.
In this embodiment of the present application, the metric fusion unit 160 is configured to multiply the multi-scale product requirement association understanding feature matrix and the retrieval result understanding feature vector to obtain a classification feature vector. It should be understood that, in view of the fact that the essence of the technical solution of the present application is to match the product requirement of the user with the search result to be matched, the multi-scale product requirement association understanding feature matrix for representing the high-dimensional feature representation of the product requirement and the search result understanding feature vector for representing the high-dimensional feature representation of the search result to be matched are matched in the high-dimensional feature space. Compared with matching in a source domain space, in the technical scheme of the application, the search results to be matched are mapped to the high-dimensional semantic feature space, so that the search results to be matched can be expressed more abundantly, and the matching accuracy between product requirements and the search results to be matched is improved.
In the technical scheme, the multi-scale product demand association understanding feature matrix and the retrieval result understanding feature vector are subjected to matrix multiplication to obtain a classification feature vector containing high-dimensional feature information of the product demand and high-dimensional feature information of the retrieval result to be matched.
In a specific embodiment of the present application, the metric fusion unit 160 calculates a product between the multi-scale product requirement association understanding feature matrix and the retrieval result understanding feature vector according to the following formula to obtain a classification feature vector;
wherein the formula is:
wherein, V 2 Is the classification feature vector, M is the multi-scale product demand association understanding feature matrix, V 1 Is the search result understanding feature vector.
In this embodiment of the application, the modeling result generating unit 170 is configured to pass the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether a search result to be matched is adapted to a product requirement input by a user.
In a specific embodiment of the present application, the modeling result generating module 170 is further configured to: processing the classification feature vector by using the classifier according to the following formula to obtain a classification result; wherein the formula is: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is the bias vector and X is the classification feature vector.
Fig. 4 illustrates a block diagram view of a training module in a dynamic parameter verification and driving CAD automated modeling engine system according to an embodiment of the application. As shown in fig. 4, in the embodiment of the present application, the dynamic parameter verification and driving CAD automatic modeling engine system 100 further includes a training module 180 for training the convolutional neural network model with the multi-scale convolutional structure and the context encoder with embedded layer; wherein the training module 180 includes: the training data acquisition unit 181 is configured to acquire a product requirement input by a user and a search result to be matched, where the product requirement includes a product model and a requirement parameter input by the user in a product-defined parameter template form, and the search result to be matched includes a text description, a block parameter, and a block coordinate of a CAD block; a training user requirement semantic encoding unit 182, configured to pass the product model and the requirement parameters in the product requirement through an untrained context encoder including an embedded layer to obtain a plurality of training product requirement understanding feature vectors; the training user requirement representing unit 183 is used for arranging the training product requirement understanding feature vectors into a two-dimensional feature matrix and then obtaining a training multi-scale product requirement association understanding feature matrix through the untrained convolutional neural network model with the multi-scale convolutional structure; a training search result encoding unit 184, configured to pass text descriptions, block parameters, and block coordinates of the CAD blocks in the search results to be matched through the untrained context encoder including the embedding layer to obtain a plurality of training search result label understanding feature vectors, and cascade the plurality of training search result label understanding feature vectors to obtain training search result understanding feature vectors; a training metric fusion unit 185, configured to multiply the training multi-scale product requirement association understanding feature matrix with the training search result understanding feature vector to obtain a training classification feature vector; a classification loss unit 186, configured to pass the training classification feature vector through a classifier to obtain a classification loss function value; and a parsimonious decomposition incentive loss function value calculation unit 187 for calculating a parsimonious decomposition incentive loss function value of the training multi-scale product demand correlation understanding feature matrix, wherein the parsimonious decomposition incentive loss function value is related to a weighted sum of natural index function values raised by powers of negative values of feature values of respective positions in respective column vectors of the training multi-scale product demand correlation understanding feature matrix.
More specifically, in an example of the present application, the training data acquisition unit 181 is configured to acquire a product requirement input by a user and a search result to be matched, where the product requirement includes a product model and a requirement parameter input by the user in a product customized parameter template form, and the search result to be matched includes a text description, a block parameter, and a block coordinate of a CAD block. That is, it is considered that the present application is a technical solution for solving the problem of matching CAD block results based on user requirements, and therefore, product requirements input by a user and search results to be matched need to be obtained.
More specifically, in an example of the present application, the training user requirement semantic encoding unit 182 is configured to pass the product model number and the requirement parameter in the product requirement through an untrained context encoder including an embedded layer to obtain a plurality of training product requirement understanding feature vectors. That is, considering the relationship between the product model and the requirement parameters and between different requirement parameters, all descriptions in the product requirement are regarded as a text sequence, and the untrained context encoder containing the embedded layer is used for performing semantic understanding on the descriptions in the product requirement to obtain a plurality of training product requirement understanding feature vectors.
More specifically, in an example of the present application, the training user requirement representing unit 183 is configured to arrange the training product requirement understanding feature vectors into a two-dimensional feature matrix, and then obtain a training multi-scale product requirement association understanding feature matrix through the untrained convolutional neural network model with a multi-scale convolution structure. That is, the same consideration is that although the use of the context semantic encoder can extract high-dimensional semantic relation information of each description relative to other descriptions in the product requirement, it is granular with a single index, which does not perform well in relation information extraction among a plurality of local indexes. Therefore, in the technical solution of the present application, a convolutional neural network model (text convolutional neural network) is used as a feature extractor to extract high-dimensional semantic associated information of each description in a product requirement.
In particular, in the technical solution of the present application, because there are different semantic associations between text descriptions of different spans in the product requirement, for example, there may be strong semantic association between a first item parameter and a fifth item parameter, and weak semantic association between the first item parameter and a second item, in the requirement parameters input by the user in the product customized parameter template form. Therefore, in order to capture semantic association information of different text spans, in the technical solution of the present application, the untrained convolutional neural network model with a multi-scale convolution structure is used as a feature extractor to extract high-dimensional semantic association information of each description in a product requirement.
More specifically, in an example of the present application, the training to-be-matched search result encoding unit 184 is configured to pass the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched through the untrained context encoder including the embedding layer to obtain a plurality of training search result label understanding feature vectors, and concatenate the plurality of training search result label understanding feature vectors to obtain a training search result understanding feature vector. That is, similarly, in consideration of semantic association among the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched, regarding the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched as a text sequence, and performing context semantic encoding on each description in the search result to be matched with the untrained context encoder including the embedding layer to obtain a trained search result understanding feature vector.
More specifically, in an example of the present application, the training metric fusion unit 185 is configured to multiply the training multi-scale product requirement association understanding feature matrix and the training search result understanding feature vector to obtain a training classification feature vector. Similarly, considering that the essence of the technical solution of the present application is to match the product requirement of the user with the search result to be matched, the training multi-scale product requirement association understanding feature matrix for representing the high-dimensional feature representation of the product requirement and the training search result understanding feature vector for representing the high-dimensional feature representation of the search result to be matched are matched in the high-dimensional feature space to obtain the training classification feature vector.
More specifically, in an example of the present application, the classification loss unit 186 is configured to pass the training classification feature vector through a classifier to obtain a classification loss function value.
Fig. 5 illustrates a block diagram view of a classification loss unit in a dynamic parameter verification and driving CAD automatic modeling engine system according to an embodiment of the present application. As shown in fig. 5, in a specific embodiment of the present application, the classification loss unit 186 includes: a training classification result obtaining unit 1861, configured to input the training classification feature vector into a Softmax classification function of the classifier to obtain a training classification result; and a classification loss function value calculation unit 1862, configured to calculate a cross entropy value between the training classification result and the real value as the classification loss function value.
In this embodiment, the parsimonial incentive loss function value calculating unit 187 is configured to calculate a parsimonial incentive loss function value of the training multi-scale product demand relevance understanding feature matrix, where the parsimonial incentive loss function value is related to a weighted sum of natural index function values raised by powers of negative values of feature values at respective positions in respective column vectors of the training multi-scale product demand relevance understanding feature matrix. Particularly, in the technical solution of the present application, for a multi-scale product demand association understanding feature matrix obtained by a convolutional neural network model with a multi-scale convolutional layer structure, assuming that a product demand association understanding feature vector is a row vector, the multi-scale product demand association understanding feature matrix represents association features of parameter semantics between different scales in a column direction, that is, when the multi-scale product demand association understanding feature matrix is optimized, it is desirable to improve multi-scale-based classification expression capability in the column direction. Therefore, an incentive loss function is introduced for the multiple-scale product demand correlation understanding feature matrix according to frugal decomposition.
In a specific embodiment of the present application, the frugal decomposition incentive loss function value calculating unit 187 is further configured to: calculating a frugal decomposition incentive loss function value of the training multi-scale product demand correlation understanding feature matrix according to the following formula;
wherein the formula is:
wherein m is i,j And correlating and understanding the characteristic value of the characteristic matrix for the multi-scale product demand, wherein tau is a penalty factor.
The parsimony decomposition encourages the loss function to be used for grouping the features in the column direction, and applying punishment to overlapping of elements in a row vector, so as to promote parsimony decomposition of a high-dimensional manifold in the column direction by calculating distance type union of the symbolization function, and the parsness decomposition can be simply understood as a set of a convex polyhedron (convex polytope) in the column direction, thus, the dimensionality monotonicity of the multi-scale product requirement correlation understanding feature matrix in the column direction is improved, and the classification expression capability of the multi-scale product requirement correlation understanding feature matrix is improved. Therefore, the adaptation accuracy between the user requirement and the retrieval result to be matched is improved, and the reasonability and the intelligence of automatic drawing construction of the CAD automatic modeling engine system are improved.
In an embodiment of the present application, the training module 180 is configured to train the convolutional neural network model with the multi-scale convolutional structure and the context encoder including the embedded layer.
In one embodiment of the present application, a weighted sum between the parsimonious decomposition incentive loss function values and the classification loss function values is calculated as a loss function value to train the convolutional neural network model with a multi-scale convolutional structure and the context encoder containing an embedded layer to accurately determine whether a match between the search result to be matched and the product demand input by the user is appropriate.
In summary, based on the dynamic parameter verification and driving CAD automatic modeling engine system according to the embodiment of the present application, a trained context encoder model including an embedded layer and a trained convolutional neural network model having a multi-scale convolutional structure are used to perform context encoding and multi-scale feature extraction on a product requirement input by a user, then a trained context encoder including an embedded layer is used to perform context encoding on a search result to be matched, and finally, a multi-scale product requirement association understanding feature matrix and a search result understanding feature vector are fused, and a classifier is used to obtain a classification result indicating whether the search result to be matched and the product requirement input by the user are matched based on a user requirement.
As described above, the dynamic parameter verification and driving CAD automatic modeling engine system 100 according to the embodiment of the present application may be implemented in various terminal devices, such as a server deployed with a dynamic parameter verification and driving CAD automatic modeling engine algorithm. In one example, the CAD-based automated modeling engine system 100 can be integrated into a terminal device as a software module and/or a hardware module based on dynamic parameter verification and driving. For example, the dynamic parameter verification and driver CAD automated modeling engine system 100 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the dynamic parameter verification and driving CAD automated modeling engine system 100 can also be one of many hardware modules of the terminal device.
Alternatively, in another example, the dynamic parameter verification and driving CAD automatic modeling engine system 100 and the terminal device may be separate devices, and the dynamic parameter verification and driving CAD automatic modeling engine system 100 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to the agreed data format.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is provided for purposes of illustration and understanding only, and is not intended to limit the application to the details which are set forth in order to provide a thorough understanding of the present application.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. As used herein, the words "or" and "refer to, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (8)
1. A dynamic parameter verification and driving CAD automatic modeling engine system is characterized by comprising:
the system comprises a user demand acquisition unit, a product customization parameter template form acquisition unit and a product customization parameter acquisition unit, wherein the user demand acquisition unit is used for acquiring product demands input by a user, and the product demands comprise product models and demand parameters input by the user in the product customization parameter template form;
a user requirement semantic coding unit, configured to pass the product model and the requirement parameters in the product requirement through a trained context encoder including an embedded layer to obtain a plurality of product requirement understanding feature vectors;
the user demand representation unit is used for arranging the product demand understanding feature vectors into a two-dimensional feature matrix and then obtaining a multi-scale product demand association understanding feature matrix through a trained convolutional neural network model with a multi-scale convolutional structure;
the device comprises a to-be-matched retrieval result acquisition unit, a matching unit and a matching unit, wherein the to-be-matched retrieval result acquisition unit is used for acquiring a to-be-matched retrieval result, and the to-be-matched retrieval result comprises text description, block parameters and block coordinates of a CAD block;
a to-be-matched retrieval result encoding unit, configured to pass the text description, the block parameters, and the block coordinates of the CAD block in the to-be-matched retrieval result through the trained context encoder including the embedded layer to obtain a plurality of retrieval result label understanding feature vectors, and cascade the plurality of retrieval result label understanding feature vectors to obtain a retrieval result understanding feature vector;
the measurement fusion unit is used for multiplying the multi-scale product demand association understanding feature matrix and the retrieval result understanding feature vector to obtain a classification feature vector; and
and the modeling result generating unit is used for enabling the classification characteristic vectors to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the retrieval result to be matched is matched with the product requirement input by the user or not.
2. The dynamic parameter verification and driven CAD automatic modeling engine system according to claim 1, wherein the user requirement semantic code module comprises:
a product requirement embedding vector generating unit, configured to convert the product model and the requirement parameter in the product requirement into a product requirement embedding vector using the trained embedding layer of the context encoder to obtain a sequence of product requirement embedding vectors;
a product requirement understanding feature vector generating unit, configured to perform global context semantic coding on the sequence of product requirement embedding vectors using the trained Bert model of the context encoder to obtain the plurality of product requirement understanding feature vectors.
3. The system of claim 2, wherein the user requirement representation module is further configured to perform the following operations on the input data in a layer forward pass using the layers of the convolutional neural network model with the multi-scale convolutional structure:
performing convolution processing on the input data based on a first convolution core to obtain a first convolution feature map;
performing convolution processing on the input data based on a second convolution kernel to obtain a second convolution characteristic diagram;
performing convolution processing on the input data based on a third convolution kernel to obtain a third convolution characteristic diagram;
performing convolution processing on the input data based on a fourth convolution kernel to obtain a fourth convolution characteristic map, wherein the first convolution kernel, the second convolution kernel, the third convolution kernel and the fourth convolution kernel have different sizes;
cascading the first convolution feature map, the second convolution feature map, the third convolution feature map and the fourth convolution feature map to obtain a multi-scale convolution feature map;
performing mean pooling along channel dimensions on the multi-scale convolution feature map to obtain a pooled feature map; and
carrying out nonlinear activation processing on the pooled feature map to obtain an activated feature map;
wherein the output of the last layer of the convolutional neural network model with the multi-scale convolutional structure is the multi-scale product demand correlation understanding feature matrix.
4. The dynamic parameter verification and driving CAD automatic modeling engine system according to claim 3, wherein said to-be-matched search result coding module comprises:
an embedded vectorization unit, configured to convert, using the trained embedded layer of the context encoder, the text description of the CAD block, the block parameter, and the block coordinate in the search result to be matched into a search result embedded vector to obtain a sequence of search result embedded vectors;
a context coding unit, configured to perform global context-based semantic coding on the sequence of the search result embedded vectors using a trained converter-based Bert model of the context coder to obtain a plurality of search result label understanding feature vectors; and
and the cascading unit is used for cascading the plurality of feature vectors to obtain the retrieval result understanding feature vector.
5. The dynamic parameter verification and driving CAD automatic modeling engine system according to claim 4, wherein the modeling result generation module is further configured to: processing the classification feature vector by using the classifier according to the following formula to obtain the classification result;
wherein the formula is: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) | X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is the bias vector and X is the classification feature vector.
6. The dynamic parameter verification and driving CAD automatic modeling engine system according to claim 1, further comprising a training module for training the convolutional neural network model with multi-scale convolutional structure and the context encoder containing embedded layer;
wherein the training module comprises:
the system comprises a training data acquisition unit, a searching data acquisition unit and a matching processing unit, wherein the training data acquisition unit is used for acquiring product requirements input by a user and a searching result to be matched, the product requirements comprise product models and required parameters input by the user in a product self-defined parameter template form, and the searching result to be matched comprises text description, block parameters and block coordinates of CAD blocks;
a training user requirement semantic coding unit, configured to pass the product model and the requirement parameters in the product requirement through an untrained context encoder including an embedded layer to obtain a plurality of training product requirement understanding feature vectors;
the training user requirement representing unit is used for arranging the training product requirement understanding feature vectors into a two-dimensional feature matrix and then obtaining a training multi-scale product requirement association understanding feature matrix through the untrained convolutional neural network model with the multi-scale convolutional structure;
the training search result coding unit is used for enabling the text description, the block parameters and the block coordinates of the CAD block in the search result to be matched to pass through the untrained context coder comprising the embedded layer to obtain a plurality of training search result label understanding feature vectors, and cascading the plurality of training search result label understanding feature vectors to obtain the training search result understanding feature vectors;
the training metric fusion unit is used for multiplying the training multi-scale product demand association understanding feature matrix and the training retrieval result understanding feature vector to obtain a training classification feature vector;
the classification loss unit is used for enabling the training classification feature vectors to pass through a classifier to obtain a classification loss function value; and
the frugal decomposition incentive loss function value calculation unit is used for calculating a frugal decomposition incentive loss function value of the training multi-scale product requirement association understanding feature matrix, wherein the frugal decomposition incentive loss function value is related to the weighted sum of natural index function values taking the negative value of the feature value of each position in each column vector of the training multi-scale product requirement association understanding feature matrix as power.
7. The dynamic parameter verification and driving CAD automatic modeling engine system according to claim 6, wherein the classification loss unit comprises:
a training classification result obtaining unit, configured to input the training classification feature vector into a Softmax classification function of the classifier to obtain a training classification result; and
and the classification loss function value calculating unit is used for calculating a cross entropy value between the training classification result and the real value as the classification loss function value.
8. The dynamic parameter verification and driving CAD automatic modeling engine system of claim 7, wherein the parsimonious incentive loss function value calculation unit is further configured to: calculating a frugal decomposition incentive loss function value of the training multi-scale product demand correlation understanding feature matrix according to the following formula;
wherein the formula is:
wherein m is i,j And associating and understanding the characteristic value of the characteristic matrix for the multi-scale product demand, wherein tau is a penalty factor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210966599.9A CN115238591B (en) | 2022-08-12 | 2022-08-12 | Dynamic parameter checking and driving CAD automatic modeling engine system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210966599.9A CN115238591B (en) | 2022-08-12 | 2022-08-12 | Dynamic parameter checking and driving CAD automatic modeling engine system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115238591A CN115238591A (en) | 2022-10-25 |
| CN115238591B true CN115238591B (en) | 2022-12-27 |
Family
ID=83679367
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210966599.9A Active CN115238591B (en) | 2022-08-12 | 2022-08-12 | Dynamic parameter checking and driving CAD automatic modeling engine system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115238591B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116127019B (en) * | 2023-03-07 | 2024-06-11 | 杭州国辰智企科技有限公司 | Dynamic parameter and visual model generation WEB 2D automatic modeling engine system |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693334A (en) * | 2012-03-01 | 2012-09-26 | 成都鹏业软件股份有限公司 | Dynamic construction member identifying method based on CAD (computer-aided design) electronic drawing |
| GB201612320D0 (en) * | 2015-07-17 | 2016-08-31 | Wal Mart Stores Inc | Shopping facility assistance systems, devices and methods to drive movable item containers |
| CN105930562A (en) * | 2016-04-13 | 2016-09-07 | 浙江大学 | Structural performance optimum design method under non-probability conditions |
| GB201613851D0 (en) * | 2015-08-14 | 2016-09-28 | Wal Mart Stores Inc | Shopping facility track system and method of routing motorized transport units |
| CN106096127A (en) * | 2016-06-07 | 2016-11-09 | 浙江大学 | Robust error estimator method containing interval parameter uncertainty structure |
| GB201703444D0 (en) * | 2016-03-03 | 2017-04-19 | Wal Mart Stores Inc | Assignment of a motorized personal assistance apparatus |
| CN107357989A (en) * | 2017-07-07 | 2017-11-17 | 电子科技大学 | A kind of digital implementation of CAE to the CAD complete structures of microwave tube |
| CN109783914A (en) * | 2018-12-29 | 2019-05-21 | 河北德冠隆电子科技有限公司 | A kind of pretreatment dynamic modelling method and device based on virtual reality emulation |
| CN110245249A (en) * | 2019-05-15 | 2019-09-17 | 西安交通大学 | A kind of three-dimensional CAD model intelligent search method based on the double-deck depth residual error network |
| CN110400370A (en) * | 2019-07-17 | 2019-11-01 | 北京航空航天大学 | A method of the semantic class component model of building three-dimensional CAD model |
| CN110941732A (en) * | 2019-11-28 | 2020-03-31 | 上海交通大学 | Adaptive three-dimensional CAD model retrieval system for standard parts |
| CN112115252A (en) * | 2020-08-26 | 2020-12-22 | 罗彤 | Intelligent auxiliary writing processing method and device, electronic equipment and storage medium |
| CN112597662A (en) * | 2020-12-30 | 2021-04-02 | 博锐尚格科技股份有限公司 | Method and system for checking correctness and mistakes of building model |
| CN113434918A (en) * | 2021-06-28 | 2021-09-24 | 北京理工大学 | Text-based three-dimensional voxel model generation method |
| CN114726870A (en) * | 2022-04-14 | 2022-07-08 | 福建福清核电有限公司 | Hybrid cloud resource arrangement method and system based on visual dragging and electronic equipment |
| CN114768279A (en) * | 2022-04-29 | 2022-07-22 | 福建德尔科技股份有限公司 | Rectification control system for preparing electronic-grade difluoromethane and control method thereof |
-
2022
- 2022-08-12 CN CN202210966599.9A patent/CN115238591B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693334A (en) * | 2012-03-01 | 2012-09-26 | 成都鹏业软件股份有限公司 | Dynamic construction member identifying method based on CAD (computer-aided design) electronic drawing |
| GB201612320D0 (en) * | 2015-07-17 | 2016-08-31 | Wal Mart Stores Inc | Shopping facility assistance systems, devices and methods to drive movable item containers |
| GB201613851D0 (en) * | 2015-08-14 | 2016-09-28 | Wal Mart Stores Inc | Shopping facility track system and method of routing motorized transport units |
| GB201703444D0 (en) * | 2016-03-03 | 2017-04-19 | Wal Mart Stores Inc | Assignment of a motorized personal assistance apparatus |
| CN105930562A (en) * | 2016-04-13 | 2016-09-07 | 浙江大学 | Structural performance optimum design method under non-probability conditions |
| CN106096127A (en) * | 2016-06-07 | 2016-11-09 | 浙江大学 | Robust error estimator method containing interval parameter uncertainty structure |
| CN107357989A (en) * | 2017-07-07 | 2017-11-17 | 电子科技大学 | A kind of digital implementation of CAE to the CAD complete structures of microwave tube |
| CN109783914A (en) * | 2018-12-29 | 2019-05-21 | 河北德冠隆电子科技有限公司 | A kind of pretreatment dynamic modelling method and device based on virtual reality emulation |
| CN110245249A (en) * | 2019-05-15 | 2019-09-17 | 西安交通大学 | A kind of three-dimensional CAD model intelligent search method based on the double-deck depth residual error network |
| CN110400370A (en) * | 2019-07-17 | 2019-11-01 | 北京航空航天大学 | A method of the semantic class component model of building three-dimensional CAD model |
| CN110941732A (en) * | 2019-11-28 | 2020-03-31 | 上海交通大学 | Adaptive three-dimensional CAD model retrieval system for standard parts |
| CN112115252A (en) * | 2020-08-26 | 2020-12-22 | 罗彤 | Intelligent auxiliary writing processing method and device, electronic equipment and storage medium |
| CN112597662A (en) * | 2020-12-30 | 2021-04-02 | 博锐尚格科技股份有限公司 | Method and system for checking correctness and mistakes of building model |
| CN113434918A (en) * | 2021-06-28 | 2021-09-24 | 北京理工大学 | Text-based three-dimensional voxel model generation method |
| CN114726870A (en) * | 2022-04-14 | 2022-07-08 | 福建福清核电有限公司 | Hybrid cloud resource arrangement method and system based on visual dragging and electronic equipment |
| CN114768279A (en) * | 2022-04-29 | 2022-07-22 | 福建德尔科技股份有限公司 | Rectification control system for preparing electronic-grade difluoromethane and control method thereof |
Non-Patent Citations (2)
| Title |
|---|
| Automatic recognition of features from freeform surface CAD models;V.B.Sunil等;《Computer-Aided Design》;20080430;第502-517页 * |
| 含未建模动态的非线性参数系统的鲁棒自适应控制;刘玉生等;《四川大学学报(工程科学版)》;20050930;第148-153页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115238591A (en) | 2022-10-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101430695B (en) | System and method for computing difference affinities of word | |
| CN115146488B (en) | Variable business process intelligent modeling system and method based on big data | |
| CN109657226B (en) | Multi-linkage attention reading understanding model, system and method | |
| CN104657350A (en) | Hash learning method for short text integrated with implicit semantic features | |
| CN113407660B (en) | Unstructured text event extraction method | |
| CN110851584B (en) | Legal provision accurate recommendation system and method | |
| CN104484380A (en) | Personalized search method and personalized search device | |
| CN115238591B (en) | Dynamic parameter checking and driving CAD automatic modeling engine system | |
| CN116720497B (en) | Semantic analysis-based power grid document relevance hierarchical analysis method and system | |
| CN103473307A (en) | Cross-media sparse Hash indexing method | |
| CN116523583A (en) | Electronic commerce data analysis system and method thereof | |
| CN115269882A (en) | Intellectual property retrieval system and method based on semantic understanding | |
| CN110866042A (en) | Intelligent table query method and device and computer readable storage medium | |
| CN112100486A (en) | Deep learning recommendation system and method based on graph model | |
| CN116821373A (en) | Map-based prompt recommendation method, device, equipment and medium | |
| CN113360654B (en) | Text classification method, apparatus, electronic device and readable storage medium | |
| CN117077586B (en) | Register transmission level resource prediction method, device and equipment for circuit design | |
| CN112905793A (en) | Case recommendation method and system based on Bilstm + Attention text classification | |
| CN116720525A (en) | Disease auxiliary analysis method, device, equipment and medium based on inquiry data | |
| CN117009516A (en) | Converter station fault strategy model training method, pushing method and device | |
| CN116124448A (en) | Fault diagnosis system and method for wind power gear box | |
| CN114240560A (en) | Product ranking method, device, equipment and storage medium based on multidimensional analysis | |
| CN110555209A (en) | Method and device for training word vector model | |
| CN108563777A (en) | A kind of method and apparatus obtaining graphical representation | |
| CN114398980A (en) | Cross-modal Hash model training method, encoding method, device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: Dynamic Parameter Verification and Driving CAD Automatic Modeling Engine System Effective date of registration: 20230720 Granted publication date: 20221227 Pledgee: Xiaoshan sub branch of Hangzhou United Rural Commercial Bank Co.,Ltd. Pledgor: Hangzhou Guochen Zhiqi Technology Co.,Ltd. Registration number: Y2023980049065 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |