CN-121980913-A - Aircraft structure assembly clearance prediction method and device based on generation countermeasure network

CN121980913ACN 121980913 ACN121980913 ACN 121980913ACN-121980913-A

Abstract

The invention discloses an aircraft assembly clearance prediction method based on a generated countermeasure network, which comprises the steps of acquiring three-dimensional point cloud data and tooling movement data of a part to be assembled; inputting data into a multisource information fusion network, extracting and fusing geometric shape characteristics and component deformation characteristics, outputting condition characteristics, inputting the condition characteristics into a condition guide assembly clearance generating network, outputting an assembly clearance image, forming an initial model by the multisource information fusion network and the condition guide assembly clearance generating network, and training the initial model by utilizing pre-constructed data to obtain a prediction model for predicting the assembly clearance image. The invention also provides an aircraft assembly clearance prediction device. The method provided by the invention has the advantages of considering prediction precision and efficiency, ensuring visual and comprehensive prediction results, and providing powerful technical support for on-line guidance and quality control of aircraft assembly.

Inventors

WANG QING
SUN HAORUI
LUO QUN
ZHANG YIFAN
KE YINGLIN

Assignees

浙江大学

Dates

Publication Date: 20260505
Application Date: 20251218

Claims (9)

1. An aircraft structure assembly clearance prediction method based on generation of an countermeasure network, comprising the steps of: acquiring three-dimensional point cloud data of a plurality of parts to be assembled in an airplane structure and tool movement data for adjusting the pose of the parts to be assembled; Inputting three-dimensional point cloud data and tooling movement data into a pre-trained multi-source information fusion network to extract geometric shape features in the three-dimensional point cloud data and deformation features of parts to be assembled of the tooling movement data, and fusing the geometric shape features and the deformation features of the parts to be assembled to generate condition features; inputting the condition characteristics into a condition guiding assembly clearance generating network to output an assembly clearance image of the distribution of the clearance of the fit surface among all the parts to be assembled; And forming an initial model by the multisource information fusion network and the conditional guidance assembly clearance generation network, and training the initial model by utilizing pre-constructed data to obtain a prediction model for predicting an assembly clearance image.
2. The aircraft structure assembly gap prediction method based on generation of an countermeasure network of claim 1, wherein the multi-source information fusion network comprises: the point cloud encoder is used for extracting geometric shape features in the input three-dimensional point cloud data; the tool movement quantity encoder is used for extracting deformation characteristics of the assembly parts in the tool movement data; And the feature fusion module is used for fusing the geometric feature and the deformation feature of the assembly part to generate a corresponding condition feature.
3. The aircraft structure assembly gap prediction method based on generation of countermeasure network of claim 2, wherein the point cloud encoder is constructed based on PointNet ++ architecture.
4. An aircraft structure assembly gap prediction method based on generating an countermeasure network according to claim 1, wherein the conditional guided assembly gap generating network comprises: a generator to output a predicted fit-up gap image based on the inputted condition features; and the discriminator is used for evaluating the authenticity of the output result of the generator and guiding the iterative optimization of the generator in a training stage.
5. The aircraft structure assembly gap prediction method based on a generation countermeasure network of claim 4, wherein the conditional guided assembly gap generation network is constructed using a generation countermeasure network architecture.
6. The method of generating an countermeasure network based aircraft structure assembly gap prediction of claim 4, wherein the generator employs a U-Net architecture with enhanced attention mechanisms.
7. The method for predicting the assembly clearance of an aircraft structure based on generating an countermeasure network according to claim 1 or 4, wherein the assembly clearance image is used for representing the gap distribution of a plurality of parts to be assembled in the current assembly state, wherein the intensity values of the pixels in the image correspond to the gap sizes of sampling points on the gap distribution of the matching surface.
8. The aircraft structure assembly gap prediction method based on generation of an countermeasure network according to claim 1, wherein the assembly gap data set is constructed as follows: modeling the geometric deviation of the matching surface by adopting a skin model shape method to obtain a corresponding structure model; establishing a kinematic relation between the pose of the component and tool movement data; Based on the kinematic relationship and the structural model, performing assembly simulation through finite element analysis to obtain corresponding gap distribution; and forming a data set by the pose of the part, the tool movement data and the gap distribution.
9. An aircraft structure assembly gap predicting device, characterized by the steps for performing the aircraft structure assembly gap predicting method based on generating an countermeasure network as claimed in any one of claims 1 to 8.

Description

Aircraft structure assembly clearance prediction method and device based on generation countermeasure network Technical Field The invention belongs to the field of aircraft automatic assembly, and particularly relates to an aircraft structure assembly clearance prediction method and device based on a generated countermeasure network. Background Aircraft manufacturing is a typical representative of high-end equipment manufacturing, and the level of accuracy control of the assembly links is one of the core indicators measuring the manufacturing capacity of national high-end equipment. In modern aircraft manufacturing, precise assembly is a key element for ensuring the performance and structural safety of the whole aircraft. When the aircraft structural members (such as wing panels and wing spars, a fuselage skin and frame ribs, etc.) are assembled in a butt joint mode, uneven gaps are inevitably generated between the matching surfaces due to the comprehensive influences of various factors such as manufacturing errors, gravity deformation, clamping force, etc. These assembly clearances are a major factor affecting assembly accuracy, and if improperly handled, can result in detrimental residual stresses within the components, which in severe cases can affect the life and flight safety of the aircraft. This is directly related to the reliability and core competitiveness of high-end aircraft equipment. Thus, to control the assembly gap, the prior art generally employs the following several approaches: And the physical measurement and manual adjustment are that in the assembly site, a craftsman uses tools such as a feeler gauge to measure key parts, and then the gap is eliminated by adjusting a tool or adding a gasket. The method is severely dependent on manual experience, has low efficiency, and cannot realize global prediction and control of the gap distribution of the whole fit surface. Virtual assembly based on three-dimensional scanning, namely acquiring an accurate three-dimensional point cloud model of a part to be assembled by using a laser scanner and other equipment before assembly, and then performing virtual assembly in a computer to predict gaps. This approach only considers the initial geometric deviation of the part, but ignores the structural deformation of the part during assembly due to tooling movement. Based on the simulation of finite element analysis, the method can accurately simulate various physical effects in the assembly process, including geometric errors, gravity deformation, contact force, clamping force and the like, so that a high-precision gap prediction result is obtained. However, for large complex aircraft components, a complete simulation analysis often takes hours or even more. The calculation efficiency makes it difficult to meet the real-time and quick decision-making requirement of an assembly site, and cannot be used for guiding assembly adjustment on line. The patent document CN119803381A discloses a method for measuring the assembly clearance of a complex-profile aircraft structure, which comprises the steps of cleaning an installation surface when a connecting piece and a connected piece are assembled, uniformly coating a layer of lubricating grease on the installation surfaces of the connecting piece and the connected piece after the connecting piece and the connected piece are cleaned, uniformly coating filler on the installation surface of the connecting piece and the connected piece, installing the connecting piece and the connected piece, after the filler is solidified, decomposing and disassembling the connecting piece and the connected piece, taking out a filler molding piece solidified between the connecting piece and the connected piece, and measuring the wall thickness of the filler molding piece to obtain the installation clearance. Patent document CN119477794a discloses a digital measurement method for assembly gaps of components with large aspect ratio, which comprises the following steps of scanning to obtain point cloud data P of a cold plate and a component in an assembly region of the TR component, preliminarily splitting the point cloud data into point cloud Pz of the TR component and point cloud Pl of the cold plate through clustering, respectively extracting mounting hole characteristics and positioning plane characteristics from the point cloud Pz and the point cloud Pl, and corresponding theoretical model pairs Ji Zhidian cloud data of the TR component and the cold plate according to the extracted hole characteristics and positioning plane characteristics, and calculating gap distribution between a receiving and transmitting component and the cold plate according to relative pose of the theoretical model. Disclosure of Invention The invention aims to provide an aircraft structure assembly clearance prediction method and device based on a generated countermeasure network, which can comprehensively consider the initial geometric deviation of componen