CN114818139A - Aircraft structural part design method based on convolutional neural network - Google Patents

Abstract

The invention discloses an aircraft structural part design method based on a convolutional neural network, which comprises the following steps: (1) a designer collects two-dimensional design drawings of different airplane structural members, inputs characteristic points of the airplane structural members through a man-machine interaction interface, and builds a convolutional neural network model; (2) reasoning analysis is carried out through a rule reasoning machine to obtain relative spatial position information and curve information among the characteristic points in the characteristic information; (3) the CNN convolutional neural network trains the data set, detects key points and establishes a convolutional neural network training library; (4) and processing the characteristic information, automatically detecting important parts in the nodes, establishing a parameterized three-dimensional model, and storing the example in a knowledge-model library in real time. The invention fully combines the convolutional neural network with the design of the aircraft structural part, and can quickly establish the three-dimensional parameterized model of the aircraft structural part by the two-dimensional design drawing, thereby shortening the design period and the research and development period of the aircraft structural part.

Description

A Design Method of Aircraft Structural Parts Based on Convolutional Neural Network

technical field

The invention relates to the technical field of aircraft design, in particular to a design method for aircraft structural parts based on a convolutional neural network.

Background technique

With the rapid development of aircraft advanced manufacturing technology, higher requirements are put forward for the design and manufacture of aircraft structural parts. In order to improve the design efficiency of aircraft and save economic costs, the United States first adopted digital technology in the field of aircraft structural parts design, and at the same time modified and optimized the design process of traditional aircraft structural parts through digital technology. Aircraft structural parts are gradually developing towards an integrated and large-scale part structure, and the complexity is getting higher and higher. In the field of aircraft structural parts design, there are problems such as low design efficiency, cumbersome design modeling process and low knowledge utilization rate. It has become an important obstacle on the road to realize intelligent manufacturing of large-scale aircraft structural parts.

The current patent publications and literature data show that: 1) The patent (CN201911335707.7) image-based automatic construction method for the processing area of the aircraft structural parts web, the method is suitable for the construction of parts models that contain broken surfaces, broken edges, opening and closing angles, The web processing area of aircraft structural parts with various complex elements such as clamping bosses, but it has not realized various types of structures of aircraft structural parts, and cannot meet the current needs of various types and designs of aircraft structural parts. 2) The patent (CN201510577537.9) is an automatic collection system and collection method for the feature points of aircraft structural parts. The automatic collection system and collection method for the feature points of aircraft structural parts provided by the method can realize the automatic collection of feature points after the reference position is set, It solves the problem of large error in the collection of feature points by traditional workers, but this method can only be applied to the collection of feature points, the scope of application is narrow, and the knowledge utilization rate of aircraft structural parts design is low.

To sum up, although the existing research results and methods can realize the intelligent design of complex aircraft structural parts to a certain extent, there are problems of high fault tolerance and limited design methods, and there are many types of aircraft structural parts. Complex, the existing methods have low utilization of design knowledge and low design efficiency, which cannot meet the current requirements of high efficiency, high design quality and parametric design in the field of aircraft structural design in my country.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is: based on the problems of low utilization rate and low design efficiency of aircraft structural parts design knowledge based on designers or systems, the present invention proposes a convolutional neural network-based aircraft structural part design method, the method By fully combining the convolutional neural network with the design of aircraft structural parts, the three-dimensional parametric model of aircraft structural parts can be quickly established from the two-dimensional design drawings, which shortens the design cycle and development cycle of aircraft structural parts.

The present invention achieves the above objects through the following technical solutions: a convolutional neural network-based aircraft structural component design method, the technical framework of which is divided into four parts, namely a knowledge acquisition and building module, a rule-based reasoning and feature information integration module, Convolutional neural network model training module, 3D model establishment and storage system module, specifically:

(1) Knowledge acquisition and building modules

The knowledge acquisition and building module is to collect a large number of two-dimensional design drawings of different aircraft structural parts. The designer inputs the feature points of the aircraft structural parts through the human-machine interface, and describes the characteristic parameters of the aircraft structural parts to build a convolutional neural network model.

(2) Rule-based reasoning and feature information integration module

The rule-based reasoning and feature information integration module performs reasoning and analysis through the rule inference engine, and obtains the relative spatial position information and curve information between the feature points in the feature information.

(3) Convolutional Neural Network Model Training Module

The convolutional neural network model training module uses the relative spatial position information and curve information of the feature points as a data set to screen and train through the CNN convolutional neural network algorithm, detect key points, and realize the refinement and typicalization of the feature information of aircraft structural parts. , connect the multi-level feature information, so that the convolutional network has the ability to map between the input and output pairs, so as to establish the training library of the convolutional neural network model.

(4) 3D model establishment and storage system module

The 3D model establishment and storage system module processes the feature information in the training library of the convolutional neural network model, detects important parts of the nodes through OpenCV, and processes through the information modules of nodes, curves, building elements, and size elements. , establish a parametric 3D model, complete the design of aircraft structural parts, save this instance to the knowledge-model library in real time, and continuously update the knowledge-model library.

Further, the two-dimensional design drawings of typical aircraft structural parts in the knowledge acquisition and building module include a front view, a top view, an arrow view, and a partial view of the aircraft structural part.

Further, the principles for establishing the feature points in the knowledge acquisition and building modules should include: 1) the selection of each feature point can carry out parametric design; 2) the selection of the feature points can preliminarily determine the type of a certain aircraft structure. .

Further, the relative spatial position information in the rule-based reasoning and feature information integration module is the coordinate value under the spatial Cartesian coordinate system of x, y, and z.

Further, in the described rule-based reasoning and feature information integration module, the rule-based reasoning is to include the domain knowledge related to the characteristics of the aircraft structure, with an IF (condition) THEN (behavior) structure, when the condition of the rule is satisfied, trigger the rule, and then perform behavior.

Further, the feature information integration in the rule-based reasoning and feature information integration module refers to obtaining feature parameters through rule base inference based on structural component design requirements, and then analyzing the spatial position information and curve information of feature points to preliminarily determine structural components. The characteristic parameters of the structure are finally optimized and determined by manual evaluation.

Further, at least three feature points need to be reserved in the relative spatial position information and the curve information in the convolutional neural network model training module.

Further, the feature information refinement in the convolutional neural network model training module is realized by the pooling layer in the CNN convolutional network structure, that is, there is a lot of feature information, and some information is not too much for building a three-dimensional model. The purpose may be repeated. The pooling layer collects features, and then extracts the features of the image frequency domain through Fourier transform for sparse processing, thereby removing redundant information and extracting the most important features.

Further, the parameterized model in the three-dimensional model establishment and storage system module is based on CATIA CAA in the human-computer interaction system by using the spatial relative position constraints between the nodes and the feature information of the aircraft structural parts to complete the aircraft structural parts. The parameters are optimized to form a 3D model.

Compared with the prior art, the present invention has the following beneficial effects:

The method for designing aircraft structural parts based on the convolutional neural network of the present invention can quickly establish a three-dimensional parameterized model of the aircraft structural parts through the two-dimensional design drawing, thereby improving the utilization rate of the design knowledge of the aircraft structural parts, improving the design efficiency, and shortening the aircraft structure. part design cycle and development cycle.

Description of drawings

Fig. 1 is the operation flow chart of the present invention;

2 is a schematic diagram of a human-computer interaction system of the present invention;

Fig. 3 is the characteristic information integration flow chart of the present invention;

Fig. 4 is the rule base schematic diagram of rule reasoning in the present invention;

FIG. 5 is a schematic diagram of the feature information integration of the present invention.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings. The drawings described here are a part of the present application, and are used to further explain the present invention, but do not constitute a limitation of the present invention.

The present invention provides a method for designing an aircraft structural part based on a convolutional neural network, comprising the following steps:

(1) Knowledge acquisition and construction is to collect a large number of two-dimensional design drawings of different aircraft structural parts, including the main view, top view, direction view, and partial view of the aircraft structural part, and then the designer enters the aircraft structural part through the human-computer interface. Feature points, describe the characteristic parameters of aircraft structural parts to build a convolutional neural network model, and the human-computer interaction system is shown in Figure 2.

(2) The rule-based reasoning and feature information integration are performed by the rule inference engine. The rule inference engine has an IF (condition) THEN (behavior) structure. When the conditions of the rule are satisfied, the rule is triggered, and then the behavior is executed, and then the result is obtained. The relative spatial position information and curve information between the feature points in the feature information, the rule base is shown in Figure 4.

(3) The training of the convolutional neural network model is to use the relative spatial position information and curve information of the feature points as a data set to screen and train through the CNN convolutional network structure, detect key points, and realize the refinement of the feature information of aircraft structural parts and typicalization, connect the multi-level feature information, so that the convolutional network has the ability to map between the input and output pairs, so as to establish the training library of the convolutional neural network model. The feature information integration steps are shown in Figure 3.

(4) The three-dimensional model establishment and storage system is to process the feature information in the training library of the convolutional neural network model, detect the important parts of the nodes through OpenCV, and pass the information modules of nodes, curves, building elements, and size elements. Process, use the spatial relative position constraints between nodes and the feature information of aircraft structural parts, complete the parameter optimization of aircraft structural parts based on CATIA CAA in the human-computer interaction system, establish a parametric 3D model, and complete the design of aircraft structural parts, And save this instance to the knowledge-model repository in real time, and continuously update the knowledge-model repository.

To sum up, the present invention combines the convolutional neural network and the design of aircraft structural parts for product design, converts two-dimensional drawings into three-dimensional parametric modeling models, greatly improves design efficiency, and the utilization rate of design knowledge is improved. Significantly improved, can shorten the development cycle of aircraft structural parts.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still understand the specific embodiments of the present invention. Modifications or equivalent substitutions, which do not depart from the spirit and scope of the present invention, are all within the protection scope of the claims of the present invention for which the application is pending.

Claims (9)

1. A method for designing an aircraft structural part based on a convolutional neural network is characterized by comprising the following steps: the method comprises the following steps:

(1) the knowledge acquisition and construction module is used for collecting a large number of two-dimensional design drawings of different airplane structural members, and a designer inputs characteristic points of the airplane structural members through a man-machine interaction interface and describes characteristic parameters of the airplane structural members to construct a convolutional neural network model;

(2) the rule-based reasoning and characteristic information integration module carries out reasoning analysis through a rule reasoning machine to obtain relative spatial position information and curve information between characteristic points in the characteristic information;

(3) the convolutional neural network model training module is used for screening and training relative spatial position information and curve information of feature points serving as a data set through a CNN convolutional neural network algorithm, detecting key points, realizing feature information refinement and representativeness of an airplane structural part, and connecting multi-layer feature information to enable the convolutional network to have the mapping capacity between input and output pairs, so that a training library of the convolutional neural network model is established;

(4) the three-dimensional model building and storage system module processes characteristic information in a training library of a convolutional neural network model, detects important parts in nodes through OpenCV, processes information modules of the nodes, curves, construction elements and size elements, builds a parameterized three-dimensional model, completes design of an airplane structural part, stores the instance in a knowledge-model library in real time, and continuously updates the knowledge-model library.

2. The method for designing the aircraft structural part based on the convolutional neural network as claimed in claim 1, wherein: the two-dimensional design drawing of the typical aircraft structural part in the knowledge acquisition and construction module comprises a front view, a top view, a directional view and a partial view of the aircraft structural part.

3. The method for designing the aircraft structural part based on the convolutional neural network as claimed in claim 1, wherein: the principle of establishing the feature points in the knowledge acquisition and construction module includes:

1) the selection of the characteristic points can be parameterized and designed each time;

2) the selection of the characteristic points can preliminarily determine the type of a certain airplane structural part.

4. The method for designing the aircraft structural part based on the convolutional neural network as claimed in claim 1, wherein: and the relative spatial position information in the rule-based reasoning and characteristic information integration module is coordinate values under a spatial rectangular coordinate system of x, y and z.

5. The method for designing the aircraft structural part based on the convolutional neural network as claimed in claim 1, wherein: the rule reasoning in the rule-based reasoning and characteristic information integration module comprises the related domain knowledge of the structural part characteristics of the airplane, has an IF (condition) THEN (behavior) structure, and triggers the rule when the condition of the rule is met, and THEN executes the behavior.

6. The method for designing an aircraft structural member based on a convolutional neural network as claimed in claim 1, wherein: the rule-based reasoning and characteristic information integration in the characteristic information integration module refers to obtaining characteristic parameters through rule base reasoning based on structural part design requirements, then analyzing the spatial position information and curve information of characteristic points, preliminarily determining the characteristic parameters of the structural part, and finally optimizing and determining the final parameter scheme of the structural part through manual evaluation.

7. The method for designing an aircraft structural member based on a convolutional neural network as claimed in claim 1, wherein: at least three feature points need to be reserved in the relative space position information and the curve information in the convolutional neural network model training module.

8. The method for designing an aircraft structural member based on a convolutional neural network as claimed in claim 1, wherein: the characteristic information refinement in the convolutional neural network model training module is realized through a pooling layer in a CNN convolutional network structure, namely, the characteristic information is more, some information has no too much use or repetition for building a three-dimensional model, the pooling layer collects characteristics, and then the characteristics of an image frequency domain are extracted through Fourier transform for sparse processing, so that redundant information is removed, and the most important characteristics are extracted.

9. The method for designing the aircraft structural part based on the convolutional neural network as claimed in claim 1, wherein: the parameterized model in the three-dimensional model establishing and storing system module is used for completing parameter optimization of the aircraft structural part based on CATIA CAA in a man-machine interaction system by utilizing space relative position constraint between nodes and characteristic information of the aircraft structural part to form a three-dimensional model.

Images