Fig.1

Deviation field of thin-walled structures

Fig.2

Tolerance constraint of thin-walled curved panel

Fig.3

Random fabrication bias generation method for skin structure

Fig.4

Deviation of fixture positioning

Fig.5

Example of fixture contact mode

Fig.6

Matrix of skin product shape

Fig.7

Matrix of skin products manufacturing deviation

Fig.8

Matrix of assembly process and deviation information

Fig.9

Image fusion method of assembly process and deviation information

Fig.10

Architecture of cGAN

Fig.11

Framework of assembly deviation prediction network

Fig.12

Framework of UNet

Fig.13

Framework of PatchGAN

Tab.1

Model geometric and material parameters

Fig.14

Layout of fixture positioning

Fig.15

Visualization of assembly process and deviation information

Fig.16

Schematic of four stages in thin-walled structure assembly

Fig.17

Results of finite element simulation

Fig.18

Dataset of flexible assembly deviation

Fig.19

Samples of network training results

Fig.20

Prediction performance of flexible deviation networks

Tab.2

Computational efficiency of the prediction method

Fig.21

Plot of generator loss

Fig.22

Plot of discriminator loss

Fig.23

Ablation for positioning deviation

Fig.24

Samples of positioning deviation ablation

Fig.25

Ablation for manufacturing deviation

Fig.26

Schematic diagram of simulated sheet metal assembly experimental platform

Fig.27

Point cloud scanning system

Fig.28

Initial deviation information collection of curved plate

Fig.29

Schematic of experiments and post-processing

Fig.30

Comparison plot of bias prediction and experiment