EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Performance Evaluation of Mirror Coatings with Limited Data Using a Transfer Learning Approach

Ernesto Bolaños-Rodríguez, Asdrúbal López-Chau, Juan-Carlos Gonzalez-Islas, Eduardo Alvarado-Santos, Evangelina Lezama-León, Gaby Yolanda Vega-Cano, Alonso Ernesto Solis-Galindo

Applied Sciences · 14.05.2026 · 16 (10) · 4900

DOI: 10.3390/app16104900 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 14.05.2026. Verfasst von Ernesto Bolaños-Rodríguez und 6 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Autonomous University of the State of Hidalgo.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The deterioration of mirror coatings in aggressive environments is one of the main causes of staining, which is a manifestation of corrosion. Electrochemical Impedance Spectroscopy (EIS) allows the electrochemical behavior of these coatings to be evaluated. However, an issue with ANNs is that to perform predictions with high accuracy, it is necessary to adjust their parameters using a large amount of samples. Depending on the ANN architecture, the requirement can range from hundreds to thousands of data points. This is a problem in many real cases, since measurements are expensive in terms of time and resources. In this study, we use a Transfer Learning approach. First, we generate and use synthetic data to train a neural network; then, real data are used to fine-tune the model. The results show that the ANN can identify patterns of coating deterioration with high accuracy and provides an effective mechanism for early performance evaluation. A high accuracy of 0.98 is achieved in the advanced stage, which means that the ANN detects severely damaged protective coatings. For the initial stage, an accuracy of 0.71 and a recall of 0.56 are obtained, indicating that the model has significant difficulty detecting initial damage due to very subtle changes, in the low signal versus noise ratio and the behavior of the protective coatings’ properties when they are close to their intact state.

Förderung

  • Autonomous University of the State of Hidalgo

Zitieren

APA

Bolaños-Rodríguez, E., López-Chau, A., Gonzalez-Islas, J., Alvarado-Santos, E., Lezama-León, E., Vega-Cano, G. Y., & Solis-Galindo, A. E. (2026). Performance Evaluation of Mirror Coatings with Limited Data Using a Transfer Learning Approach. Applied Sciences, 16(10), 4900. https://doi.org/10.3390/app16104900

Vancouver

Bolaños-Rodríguez E, López-Chau A, Gonzalez-Islas J, Alvarado-Santos E, Lezama-León E, Vega-Cano GY, et al. Performance Evaluation of Mirror Coatings with Limited Data Using a Transfer Learning Approach. Applied Sciences. 2026;16(10):4900. doi:10.3390/app16104900

BibTeX
@article{BolaosRodrguez2026,
  author = {Bolaños-Rodríguez, Ernesto and López-Chau, Asdrúbal and Gonzalez-Islas, Juan-Carlos and Alvarado-Santos, Eduardo and Lezama-León, Evangelina and Vega-Cano, Gaby Yolanda and Solis-Galindo, Alonso Ernesto},
  title = {Performance Evaluation of Mirror Coatings with Limited Data Using a Transfer Learning Approach},
  journal = {Applied Sciences},
  year = {2026},
  volume = {16},
  number = {10},
  pages = {4900},
  doi = {10.3390/app16104900}
}
Kostenlos starten →