Restricted accessResearch articleFirst published online 2025
Research on acoustic emission monitoring of mechanical behavior of differentiated aging composites based on improved Laplacian score feature selection algorithm
As a fundamental material in the aerospace sector, the long-term service performance of composites is inseparable from their internal damage evolution. Therefore, studying the relationship between damage evolution and the mechanical behavior of composites in different aging states is of great significance. Acoustic emission (AE) has emerged as an effective method for investigating the damage evolution of composites due to its high sensitivity to material microscopic damage and the superiority of real-time monitoring, and the key to studying damage evolution lies in the accurate selection of AE features. The Laplacian score (LS) is a traditional algorithm for feature selection in AE, but its limitations severely hinder the application of the algorithm. In this study, an improved LS feature selection algorithm (Imp-LS) is proposed, which enhances the ability to identify feature correlations and improves the robustness of feature selection. The robustness of the Imp-LS algorithm and the effectiveness of AE for whole-life health monitoring of composites are validated through monitoring the mechanical behavior of double cantilever beam testing on various aging composites.
ZhangJLinGVaidyaU, et al. Past, present and future prospective of global carbon fibre composite developments and applications. Compos Part B Eng2023; 250: 110463.
2.
RavichandranBBalasubramanianM. Joining methods for fiber reinforced polymer (FRP) composites − a critical review. Compos Part A Appl Sci Manuf2024; 186: 108394.
3.
DeBBeraMBhattacharjeeD, et al. A comprehensive review on fiber-reinforced polymer composites: raw materials to applications, recycling, and waste management. Prog Mater Sci2024; 146: 101326.
4.
SaeedifarMZarouchasD. Damage characterization of laminated composites using acoustic emission: a review. Compos Part B Eng2020; 195: 108039.
5.
DuJZengJChenC, et al. Acoustic emission monitoring for damage diagnosis in composite laminates based on deep learning with attention mechanism. Mech Syst Signal Process2025; 222: 111770.
6.
SuYZhangTSunH, et al. Cryogenic damage behavior of carbon fiber reinforced polymer composite laminates via fiber-optic acoustic emission. Compos Part A Appl Sci Manuf2024; 186: 108435.
7.
XuDLiuPFLiJG, et al. Damage mode identification of adhesive composite joints under hygrothermal environment using acoustic emission and machine learning. Compos Struct2019; 211: 351–363.
8.
BrunnerAJ. Identification of damage mechanisms in fiber-reinforced polymer-matrix composites with acoustic emission and the challenge of assessing structural integrity and service-life. Constr Build Mater2018; 173: 629–637.
9.
LyuPYaoLMaX, et al. Correlation between failure mechanism and rupture lifetime of 2D-C/SiC under stress oxidation condition based on acoustic emission pattern recognition. J Eur Ceram Soc2020; 40: 5094–5102.
10.
JungDLeeB-SYuW-R, et al. Effect of propagation distance on acoustic emission of carbon fiber/epoxy composites. Struct Health Monit2021; 20: 3342–3353.
11.
SaeedifarMSalehMNNijhuisP, et al. Damage assessment of a titanium skin adhesively bonded to carbon fiber-reinforced plastic omega stringers using acoustic emission. Struct Health Monit2022; 21: 407–423.
12.
WangXXieHTongY, et al. Three-point bending properties of 3D_C/C_TiC_Cu composites based on acoustic emission technology. Mech Syst Signal Process2023; 184: 109693.
13.
LiDMaXGongM. Joint learning of feature extraction and clustering for large-scale temporal networks. IEEE Trans Cybern2023; 53: 1653–1666.
14.
NieFDongXTianL, et al. Unsupervised feature selection with constrained ℓ2,0-norm and optimized graph. IEEE Trans Neural Netw Learn Syst2022; 33: 1702–1713.
15.
MartarelliNJNaganoMS. Unsupervised feature selection based on bio-inspired approaches. Swarm Evol Comput2020; 52: 100618.
16.
ZhangRZhangYLiX. Unsupervised feature selection via adaptive graph learning and constraint. IEEE Trans Neural Netw Learn Syst2022; 33: 1355–1362.
17.
ShangRGaoLChiH, et al. Non-convex feature selection based on feature correlation representation and dual manifold optimization. Expert Syst Appl2024; 250: 123867.
18.
XuDLiuPChenZ, et al. Dynamic feature evaluation on streaming acoustic emission data for adhesively bonded joints for composite wind turbine blade. Struct Health Monit2022; 21: 387–406.
19.
GulsenAKolukisaBCaliskanU, et al. Ensemble feature selection for clustering damage modes in carbon fiber-reinforced polymer sandwich composites using acoustic emission. Adv Eng Mater2024; 26: 2400317.
20.
NieFMaZWangJ, et al. Fast sparse discriminative K-means for unsupervised feature selection. IEEE Trans Neural Netw Learn Syst2023; 35: 9943–9957.
21.
HeXCaiDNiyogiP. Laplacian score for feature selection. In: Advances in neural information processing systems. MIT Press: Cambridge, UK, 2005; pp:507-514.
22.
SongDXuFMaT. Crack damage monitoring for compressor blades based on acoustic emission with novel feature and hybridized feature selection. Struct Health Monit2022; 21: 2641–2656.
23.
OzFEAhmadvashaghbashSErsoyN. Damage mode identification in transverse crack tension specimens using acoustic emission and correlation with finite element progressive damage model. Compos Part B Eng2019; 165: 84–95.
24.
Rubio-GonzálezCde Urquijo-VenturaMdelPRodríguez-GonzálezJA. Damage progression monitoring using self-sensing capability and acoustic emission on glass fiber/epoxy composites and damage classification through principal component analysis. Compos Part B Eng2023; 254: 110608.
25.
MichałSSzymonDPawełZ, et al. Damage characterisation of GFRP composites based on clustering acoustic emission events utilizing single-failure-cause tests as reference. Compos Struct2025; 351: 118596.
26.
RomhányGCzigányTKarger-KocsisJ. Failure assessment and evaluation of damage development and crack growth in polymer composites via localization of acoustic emission events: a review. Polym Rev2017; 57: 397–439.
27.
LiLLomovSVYanX, et al. Cluster analysis of acoustic emission signals for 2D and 3D woven glass/epoxy composites. Compos Struct2014; 116: 286–299.
28.
AgrawalURohatgiVKataryaR. Normalized mutual information-based equilibrium optimizer with chaotic maps for wrapper-filter feature selection. Expert Syst Appl2022; 207: 118107.
29.
GongHLiYZhangJ, et al. A new filter feature selection algorithm for classification task by ensembling Pearson correlation coefficient and mutual information. Eng Appl Artif Intel2024; 131: 107865.
30.
WeiYChenZYeZ-S, et al. High-dimensional process monitoring under time-varying operating conditions via covariate-regulated principal component analysis. Reliab Eng Syst Safe2024; 252: 110440.
31.
LeeIYRohHDParkHW, et al. Advanced non-destructive evaluation of impact damage growth in carbon-fiber-reinforced plastic by electromechanical analysis and machine learning clustering. Compos Sci Technol2022; 218: 109094.
32.
BagirovAMAliguliyevRMSultanovaN. Finding compact and well-separated clusters: clustering using silhouette coefficients. Pattern Recogn2023; 135: 109144.
33.
RosFRiadRGuillaumeS. PDBI: a partitioning Davies-Bouldin index for clustering evaluation. Neurocomputing2023; 528: 178–199.
34.
GagolewskiMBartoszukMCenaA. Are cluster validity measures (in) valid?Inform Sci2021; 581: 620–636.
35.
ZhouWZhaoWZhangY, et al. Cluster analysis of acoustic emission signals and deformation measurement for delaminated glass fiber epoxy composites. Compos Struct2018; 195: 349–358.
36.
BohmannTSchlampMEhrlichI. Acoustic emission of material damages in glass fibre-reinforced plastics. Compos Part B Eng2018; 155: 444–451.
37.
JiaRZhaoLCurtiR, et al. Determination of pure mode-I fracture toughness of multidirectional composite DCB specimens. Eng Fract Mech2021; 252: 107776.
