Utilization of Ensemble Techniques in Machine Learning to Predict the Porosity and Hardness of Plasma-Sprayed Ceramic Coating
Ceramic coatings play a vital role in protecting steel components by significantly enhancing their corrosion and wear resistance, thereby extending service life. The performance of these coatings critically depends on surface characteristics, such as porosity and hardness, which are traditionally as...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11106496/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Ceramic coatings play a vital role in protecting steel components by significantly enhancing their corrosion and wear resistance, thereby extending service life. The performance of these coatings critically depends on surface characteristics, such as porosity and hardness, which are traditionally assessed through time-consuming and labour-intensive experimental methods. To address this challenge, the present study employs advanced machine learning ensemble techniques, including bagging, boosting, stacking, weighted averaging, voting, and hybrid methods, to accurately predict the porosity and hardness of plasma-sprayed ceramic coatings based on key process parameters. Feature engineering identifies input power and spray distance as the most influential factors affecting coating properties. Hyperparameter optimization is performed using the random search technique and is compared with the conventional grid search method. The hybrid ensemble technique demonstrates exceptional predictive performance, achieving a coefficient of determination (R2) of 0.92 with a Root Mean Square Error (RMSE) of 2.01 % and a Mean Absolute Error (MAE) of 1.62 % for porosity prediction, and R2 of 0.94 with the RMSE of 82.03 HV and MAE of 66.42 HV for hardness prediction. Experimental validation confirms the model’s reliability, through minimal error deviation between predicted and actual values for porosity and hardness. This ML approach provides a robust framework for optimizing coating processes and ensuring superior steel protection through data-driven decision-making. |
|---|---|
| ISSN: | 2169-3536 |