Special Issue: Aircraft Digital Twin Technology

A deep feature fusion network based on multi-scale kernel construction for filling wing stress field data

  • Lin LIN ,
  • Shiwei SUO ,
  • Dan LIU ,
  • Yinxuan ZHANG ,
  • Lingyu YUE ,
  • Sihao ZHANG ,
  • Yikun LIU ,
  • Song FU
Expand
  • 1.School of Mechanical and Electrical Engineering,Harbin Institute of Technology,Harbin 150001,China
    2.AVIC Shenyang Aircraft Design and Research Institute,Shenyang 110035,China

Received date: 2025-05-30

  Revised date: 2025-06-24

  Accepted date: 2025-09-04

  Online published: 2025-09-10

Supported by

National Key Research and Development Project(2023YFB3308900);National Natural Science Foundation of China(52305570);Heilongjiang Province Natural Science Foundation(LH2024F014)

Abstract

Stress monitoring of aircraft wings is crucial for ensuring flight safety and maintaining the structural reliability of aircraft. In practical engineering applications, sensors must be installed at sparse and fixed locations, which limits their ability to capture stress data across the entire wing structure. As a result, comprehensive monitoring of the wing’s structural health cannot be ensured. To address this issue, this paper proposes a deep feature fusion network based on multi-scale kernel construction to reconstruct the wing stress field in cases where sensor placement is spatially sparse or incomplete. First, based on the similarity among stress fields, the data is clustered into several stress field groups, and the field closest to the center of each cluster is selected as the reference set. Then, by extracting components of varying scales from each reference stress field, multi-scale convolution kernels are constructed to perceive the missing data points from different directions, enabling the capture of multi-scale features conducive to information completion. Finally, a parallel channel attention module is employed to adaptively select salient features captured by the convolutional kernels and map them into a unified feature space for fusion, thereby generating the imputed values for the missing points. Additionally, comparative experiments were conducted on wing stress field datasets with different missing ratios. The proposed method outperforms mainstream approaches in terms of MAE, RMSE, and MAPE, achieving the best overall filling performance.

Cite this article

Lin LIN , Shiwei SUO , Dan LIU , Yinxuan ZHANG , Lingyu YUE , Sihao ZHANG , Yikun LIU , Song FU . A deep feature fusion network based on multi-scale kernel construction for filling wing stress field data[J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2025 , 46(19) : 532343 -532343 . DOI: 10.7527/S1000-6893.2025.32343

References

[1] 王彬, 郑建军, 刘玮, 等. 机翼部段静力试验优化设计方法[J]. 航空学报202344(18): 228065.
  WANG B, ZHENG J J, LIU W, et al. Optimum design method for static test of aircraft wing segment[J]. Acta Aeronautica et Astronautica Sinica202344(18): 228065 (in Chinese).
[2] 李文龙, 吴波, 谢帅. 有起落架布置的翼身整体结构机翼载荷测量技术[J]. 航空学报202445(1): 229525.
  LI W L, WU B, XIE S. Technology of wing load measurement for wing-body integrated structure with landing gear layout[J]. Acta Aeronautica et Astronautica Sinica202445(1): 229525 (in Chinese).
[3] 张晟斐, 李天梅, 胡昌华, 等. 基于深度卷积生成对抗网络的缺失数据生成方法及其在剩余寿命预测中的应用[J]. 航空学报202243(8): 225708.
  ZHANG S F, LI T M, HU C H, et al. Missing data generation method and its application in remaining useful life prediction based on deep convolutional generative adversarial network[J]. Acta Aeronautica et Astronautica Sinica202243(8): 225708 (in Chinese).
[4] CRESPO TURRADO C, SáNCHEZ LASHERAS F, CALVO-ROLLé J L, et al. A new missing data imputation algorithm applied to electrical data loggers[J]. Sensors201515(12): 31069-31082.
[5] GUO Z J, WAN Y M, YE H. A data imputation method for multivariate time series based on generative adversarial network[J]. Neurocomputing2019360: 185-197.
[6] FIERO M H, HUANG S, OREN E, et al. Statistical analysis and handling of missing data in cluster randomized trials: A systematic review[J]. Trials201617: 72.
[7] 任超, 李慧琴, 李天梅, 等. 基于退化模型动态校准的设备剩余寿命预测方法[J]. 航空学报202344(19): 228345.
  REN C, LI H Q, LI T M, et al. Equipment remaining useful life prediction method with dynamic calibration of degradation model[J]. Acta Aeronautica et Astronautica Sinica202344(19): 228345 (in Chinese).
[8] CHE Z P, PURUSHOTHAM S, CHO K, et al. Recurrent neural networks for multivariate time series with missing values[J]. Scientific Reports20188: 6085.
[9] LIU M S, CLARIDGE D E. The maximum potential energy savings from optimizing cold deck and hot deck reset schedules for dual duct VAV systems[J]. Journal of Solar Energy Engineering1999121(3): 171-175.
[10] 崔西明, 邱志鹏, 魏嘉, 等. 基于数据驱动的结构钢表面应力磁巴克豪森噪声表征方法[J]. 航空学报202344(8): 427237.
  CUI X M, QIU Z P, WEI J, et al. Data-driven method for characterization of structural steel surface stress of magnetic Barkhausen noise[J]. Acta Aeronautica et Astronautica Sinica202344(8): 427237 (in Chinese).
[11] ZHAO L, CHEN Z K, YANG Z N, et al. Local similarity imputation based on fast clustering for incomplete data in cyber-physical systems[J]. IEEE Systems Journal201812(2): 1610-1620.
[12] 孙彬, 游航, 李文博, 等. 双光载荷图像融合及其在低空遥感中的应用[J]. 航空学报202546(11): 531343.
  SUN B, YOU H, LI W B, et al. Dual-band payload image fusion and its applications in low-altitude remote sensing[J]. Acta Aeronautica et Astronautica Sinica202546(11): 531343 (in Chinese).
[13] ZHANG J H, AYTUG H. Comparison of imputation methods for discriminant analysis with strategically hidden data[J]. European Journal of Operational Research2016255(2): 522-530.
[14] 林杰, 唐志共, 钱炜祺, 等. 飞行器生成式模型气动设计研究进展与展望[J]. 航空学报202546(10): 631679.
  LIN J, TANG Z G, QIAN W Q, et al. Research progress and prospects of aircraft aerodynamic design based on generative models[J]. Acta Aeronautica et Astronautica Sinica202546(10): 631679 (in Chinese).
[15] LIU D, ZHONG S S, LIN L, et al. Feature-level SMOTE: Augmenting fault samples in learnable feature space for imbalanced fault diagnosis of gas turbines[J]. Expert Systems with Applications2024238: 122023.
[16] SALINAS D, FLUNKERT V, GASTHAUS J, et al. DeepAR: Probabilistic forecasting with autoregressive recurrent networks[J]. International Journal of Forecasting202036(3): 1181-1191.
[17] 赵燕, 宋江涛, 唐宁. 某机翼的安全预测载荷模型建立[J]. 航空学报202041(10): 223852.
  ZHAO Y, SONG J T, TANG N. Construction of safety-predicting load model on certain wing[J]. Acta Aeronautica et Astronautica Sinica202041(10): 223852 (in Chinese).
[18] 吴跃腾, 巴顿, 杜娟, 等. 基于深度注意力网络的压气机流场重构方法[J]. 航空学报202445(24): 630580.
  WU Y T, BA D, DU J, et al. Compressor flow field reconstruction method based on deep attention networks[J]. Acta Aeronautica et Astronautica Sinica202445(24): 630580 (in Chinese).
[19] 张音旋, 张起, 许镇勇, 等. 一种基于残差网络的飞机力学响应预测方法[J]. 航空学报202546(19):531295.
  ZHANG Y X, ZHANG Q, XU Z Y, et al. A method for predicting aircraft mechanical response Based on residual networks[J]. Acta Aeronautica et Astronautica Sinica202546(19):531295 (in Chinese).
[20] 王鹏飞, 曾丽芳, 邵雪明, 等. 基于预训练微调的机翼气动载荷多源数据融合建模方法研究[J]. 航空学报202546(19):532297.
  WANG P F, ZENG L F, SHAO X M, et al. Research on multi-source data fusion modeling method for aerodynamic load of aircraft wing based on pre-training and fine-tuning[J]. Acta Aeronautica et Astronautica Sinica202546(19):532297 (in Chinese).
[21] COMBERT F K, XIE S K, LAWNICZAK A T. Bi-partitioned feature-weighted K-means clustering for detecting insurance fraud claim patterns[J]. Mathematics202513(3): 434.
[22] LIN L, ZU L Z, FU S, et al. Integrating adversarial training strategies into deep autoencoders: A novel aeroengine anomaly detection framework[J]. Engineering Applications of Artificial Intelligence2024136: 108856.
[23] HAN B K, YIN P W, ZHANG Z Z, et al. Remaining useful life prediction of turbofan engines based on dual attention mechanism guided parallel CNN-LSTM[J]. Measurement Science and Technology202536(1): 016160.
[24] ROBESON S M, WILLMOTT C J. Decomposition of the mean absolute error (MAE) into systematic and unsystematic components[J]. PLoS One202318(2): e0279774.
[25] QIAN L M, CAO W R, CHEN L F. Influence of artificial intelligence on higher education reform and talent cultivation in the digital intelligence era[J]. Scientific Reports202515: 6047.
[26] DE MYTTENAERE A, GOLDEN B, LE GRAND B, et al. Mean absolute percentage error for regression models[J]. Neurocomputing2016192: 38-48.
[27] FU S, LIN L, WANG Y, et al. MCA-DTCN: A novel dual-task temporal convolutional network with multi-channel attention for first prediction time detection and remaining useful life prediction[J]. Reliability Engineering & System Safety2024241: 109696.
[28] RAZALI M N, ARBAIY N, LIN P C, et al. Optimizing multiclass classification using convolutional neural networks with class weights and early stopping for imbalanced datasets[J]. Electronics202514(4): 705.
[29] 李左飙, 温风波, 唐晓雷, 等. 基于深度学习的单排孔气膜冷却性能预测[J]. 航空学报202142(4): 524331.
  LI Z B, WEN F B, TANG X L, et al. Prediction of single-row hole film cooling performance based on deep learning[J]. Acta Aeronautica et Astronautica Sinica202142(4): 524331 (in Chinese).
[30] 崔江, 周凡, 陈永凡, 等. 一种基于GAMF-CNN的航空发电机整流器故障诊断技术[J]. 航空学报202445(24): 330398.
  CUI J, ZHOU F, CHEN Y F, et al. A technique for aerospace generator rectifier fault diagnosis based on GAMF-CNN[J]. Acta Aeronautica et Astronautica Sinica202445(24): 330398 (in Chinese).
[31] 何磊, 钱炜祺, 董康生, 等. 基于卷积神经网络的结冰翼型气动特性建模[J]. 航空学报202344(5): 126434.
  HE L, QIAN W Q, DONG K S, et al. Aerodynamic characteristics modeling of iced airfoil based on convolution neural networks[J]. Acta Aeronautica et Astronautica Sinica202344(5): 126434 (in Chinese).
[32] 王志刚, 王业光, 杨宁, 等. 基于LSTM的飞行数据挖掘模型构建方法[J]. 航空学报202142(8): 525800.
  WANG Z G, WANG Y G, YANG N, et al. Construction method of flight data mining model based on LSTM[J]. Acta Aeronautica et Astronautica Sinica202142(8): 525800 (in Chinese).
[33] MUNIR H S, REN S B, MUSTAFA M, et al. Attention based GRU-LSTM for software defect prediction[J]. PLoS One202116(3): e0247444.
[34] SUN B, HU W T, WANG H, et al. Remaining useful life prediction of rolling bearings based on CBAM-CNN-LSTM[J]. Sensors202525(2): 554.
Outlines

/