Abstract:It is very important for the stable operation of spacecraft to realize the fault diagnosis of aluminum alloy stiffened plate, but it may be difficult to accurately identify the damage with some traditional methods due to its special rib structure. This paper presents a fault diagnosis method based on convolution neural network (CNN) to deal with impact damage recognition of aluminum alloy stiffened plates, which can accurately identify the impact location and damage degree (including no damage, pothole and perforation). For this purpose, the damage identification model was constructed by using CNN with double output, and the impact signal containing the impact position and damage degree information was obtained by combining the shock stress wave detection system and finite element simulation. Furthermore, the frequency data set is generated through the fast Fourier transform of the impact signal, and it is input into CNN to further extract the features. Finally, the position and damage degree are identified by two output layers respectively. To validate the effectiveness of the proposed model, the models of neural network and support vector machine are used for comparison. The results show that the method is well performed in accuracy and rapidity. The accuracies of this method reach 99.85% and 99.61% in impact location and damage degree.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve The paper aims to address the issue of fault diagnosis in aluminum alloy stiffened panels in spacecraft. Due to the special rib structure of aluminum alloy stiffened panels, traditional fault diagnosis methods face difficulties in identifying the location and extent of damage. Therefore, the paper proposes a method based on Convolutional Neural Networks (CNN) to identify impact damage in aluminum alloy stiffened panels, including the impact location and damage extent (no damage, dent, and perforation). ### Summary of Main Content 1. **Research Background**: - Spacecraft are subjected to various physical impacts (such as space debris, hail, etc.) during long-term operations, leading to damage or destruction of aluminum alloy stiffened panels, thus posing safety risks to space missions. - Common non-destructive testing methods (such as ultrasound, X-rays, computed tomography, etc.) are time-consuming and difficult to monitor in real-time. 2. **Problems with Existing Methods**: - Model-driven methods require precise mathematical models, but it is challenging to obtain accurate speed information in practical operations. - Data-driven methods (such as neural networks, support vector machines, etc.) can handle large amounts of data but face challenges in runtime and memory usage. 3. **Proposed Solution**: - Utilize an impact stress wave detection system combined with numerical simulation to obtain signals containing information on impact location and damage extent. - Convert time-domain signals to frequency-domain signals using Fast Fourier Transform (FFT) and use them as input to the CNN to extract feature information. - Design a dual-output CNN model to simultaneously identify impact location and damage extent. 4. **Experimental Validation**: - Generate time-domain signals of different damage extents through experiments and simulations, and convert them to frequency-domain signals as input to the CNN. - Construct the CNN model and optimize model parameters through training and validation sets. - Experimental results show that the method performs excellently in terms of accuracy and speed, with recognition accuracies of 99.85% and 99.61% for impact location and damage extent, respectively.

Impact-Induced Damage Recognition of Aluminium Alloy Stiffened Plate Structure Based on Convolutional Neural Network

Damage identification based on the inner product matrix and parallel convolution neural network for frame structure

Investigations on Plate Crack Damage Detection Using Convolutional Neural Networks

Structural Damage Identification Based on Convolutional Neural Networks and Improved Hunter–Prey Optimization Algorithm

Structural damage identification using strain mode differences by the iFEM based on the convolutional neural network (CNN)

Deep Learning-Based Subsurface Damage Localization Using Full-Field Surface Strains

Damage Quantification Method for Aircraft Structures Based on Multi-Task CNN-LSTM and Transfer Learning

Structural Damage Identification Based on Convolutional Neural Network Group Considering the Sensor Fault

Damage Quantification Method for Aircraft Structures Based on Multitask CNN-LSTM and Transfer Learning

A Convolutional Neural Network for Impact Detection and Characterization of Complex Composite Structures

Structural Damage Identification Based on Convolution Neural Network

A Two-Stage Structural Damage Detection Method Based on 1D-CNN and SVM

Improved Damage Localization and Quantification of CFRP Using Lamb Waves and Convolution Neural Network

Structural damage identification under nonstationary excitations through recurrence plot and multi-label convolutional neural network

Civil Engineering Structural Damage Identification by Integrating Benchmark Numerical Model and PCNN Network

A novel damage identification algorithm by combing the boundary element method and a series connection neural network

Structural damage identification based on variational mode decomposition–Hilbert transform and CNN

Impact Damage Identification of Aluminum Alloy Reinforced Plate Based on GWO-ELM Algorithm

Parallel Convolutional Neural Network Toward High Efficiency and Robust Structural Damage Identification

Structural damage identification and classification from time-frequency impact echo data using deep learning

One-dimensional Convolutional Neural Network for Damage Detection of Jacket-Type Offshore Platforms