Forecasting Dielectric Behavior of Nano-Epoxy Materials through AI-based Electronic Properties

Authors Vaibhav D. Dabhade1, Bhagyashree Ashok Tingare2, Sandip R. Thorat3, R.A. Kapgate4, Tarun Dhar Diwan5, Laxmikant S Dhamande3, P. William6
Affiliations

1MET Institute of Engineering, Nashik, MH, India

2Department of Artificial Intelligence and Data Science, D Y Patil College of Engineering, Akurdi, Pune, India

3Department of Mechanical Engineering, Sanjivani College of Engineering, Kopargaon, MH, India

4Department of Mechatronics Engineering, Sanjivani College of Engineering, Kopargaon, MH, India

5Controller of Examination (COE), Atal Bihari Vajpayee University, Bilaspur, India

6Department of Information Technology, Sanjivani College of Engineering, Kopargaon, MH, India

Е-mail vaibhavd_ioe@bkc.met.edu
Issue Volume 17, Year 2025, Number 3
Dates Received 08 April 2025; revised manuscript received 20 June 2025; published online 27 June 2025
Citation Vaibhav D. Dabhade, Bhagyashree Ashok Tingare, et al., J. Nano- Electron. Phys. 17 No 3, 03023 (2025)
DOI https://doi.org/10.21272/jnep.17(3).03023
PACS Number(s) 07.05.Mh, 77.55. + f
Keywords Dielectric behavior, Nano-epoxy materials, Artificial intelligence, Machine learning models, FSS-MAdaBoost.
Annotation

To predict the dielectric behavior of nano-epoxy composites, sophisticated machine learning algorithms were suggested. Dielectric characteristics were precisely estimated to maximize the use of the nano-epoxy composite in electronics. Using AI models and data on their electrical properties, the objective is to predict the dielectric behavior of nano-epoxy materials. To achieve consistent feature contributions, the dataset was preprocessed using min-max normalization, which normalized the range of input characteristics. Therefore, present the Fine-tuned Squirrel Search Algorithm-driven Malleable AdaBoost model (FSS-MAdaBoost), which combines MAdaBoost with the FSS. This hybrid model may overcome the typical drawbacks of improved prediction accuracy and the successful handling of complicated and nonlinear connections between features. The suggested model is compared to an existing model. The performance was evaluated using RMSE (0.018) and MAE (0.01) measures. According to the foregoing results, the FSS-MAdaBoost-based model outperforms previous approaches with much lower values of RMSE and MAE, indicating superior predictions and dependability. The results indicated promising directions for dielectric property forecasting using the FSS-MAdaBoost model for nano-epoxy materials, providing valuable insights that material scientists and engineers can use to optimize material design, thereby improving electronic applications.

List of References