Advanced machine learning framework for hole quality prediction and optimization in polycarbonate AWJD
Abstract
Abrasive waterjet drilling of polymers is difficult to optimize owing to nonlinear interactions and conflicting performance requirements. This study suggests combining experimental, ML models, multi-objective optimization, and statistical validation to improve polycarbonate drilling performance. A full factorial design was used to vary water pressure (250–350 MPa), standoff distance (1.5–2.5 mm), and traversal rate (300–500 mm/min) in 125 tests. DE and drilling rate were performance responses. Analysis of variance showed water pressure and standoff distance were the principal variables affecting the hole diameter error (DE) and drilling rate (DR). Four machine learning models, namely XGBoost, Decision Tree, Random Forest, and AdaBoost Regressor were built in the Anaconda-based Python environment. The Random Forest model demonstrated high reliability in capturing complex process behavior, with R² = 0.92263 for DE and 0.90471 for DR. Random Forest was utilized as a stand-in paradigm for multi-objective optimization using the Grey Wolf optimizer, Whale optimization algorithm, and Arithmetic optimization algorithm. With Whale optimization method, Water pressure = 350 MPa, Standoff distance = 2.5 mm, and Traverse rate = 300 mm/min produced the highest performance with decreased DE (0.1238) and improved DR (2.0450). The Friedman test ( p < 0.001) and post-hoc analysis show that Whale optimization method outperforms Grey Wolf and Arithmetic optimization algorithms. The results of confirmation tests with the optimized parameters for the WOA showed good agreement with the predicted results, within a range of ± 3%.
Related
- Clinical-Anchored Incremental Survival Learning: A Stability-Constrained Framework for Molecularly Enhanced Survival Prediction in Lung Adenocarcinoma
- sORF-Trans2MS: a two-module deep learning framework for sORF translation and MS-supported microprotein prediction
- MarkerScout: A Disease-Agnostic Machine Learning Framework for Biomarker Prediction from Multi-Scale Mechanistic Models
- scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning
- Physically aligned forest fire risk prediction: A deep learning framework coupling fuel and climate multivariate factors
BibTeX
@article{chandar2026advanced,
title = {{Advanced machine learning framework for hole quality prediction and optimization in polycarbonate AWJD}},
author = {J. Bharani Chandar and C. Rathinasuriyan and N. Lenin and M. Sivakumar and Ravi Prakash Singh and Sachin Salunkhe and Poongavanam GaneshKumar and R. Palanisamy and Emad Abouel Nasr},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-71583-2},
url = {https://www.nature.com/articles/s41598-026-71583-2},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Advanced machine learning framework for hole quality prediction and optimization in polycarbonate AWJD
AU - J. Bharani Chandar
AU - C. Rathinasuriyan
AU - N. Lenin
AU - M. Sivakumar
AU - Ravi Prakash Singh
AU - Sachin Salunkhe
AU - Poongavanam GaneshKumar
AU - R. Palanisamy
AU - Emad Abouel Nasr
PY - 2026
DA - 2026/10/01
JO - Scientific Reports
DO - 10.1038/s41598-026-71583-2
UR - https://www.nature.com/articles/s41598-026-71583-2
AB - Abrasive waterjet drilling of polymers is difficult to optimize owing to nonlinear interactions and conflicting performance requirements. This study suggests combining experimental, ML models, multi-objective optimization, and statistical validation to improve polycarbonate drilling performance. A full factorial design was used to vary water pressure (250–350 MPa), standoff distance (1.5–2.5 mm), and traversal rate (300–500 mm/min) in 125 tests. DE and drilling rate were performance responses. Analysis of variance showed water pressure and standoff distance were the principal variables affecting the hole diameter error (DE) and drilling rate (DR). Four machine learning models, namely XGBoost, Decision Tree, Random Forest, and AdaBoost Regressor were built in the Anaconda-based Python environment. The Random Forest model demonstrated high reliability in capturing complex process behavior, with R² = 0.92263 for DE and 0.90471 for DR. Random Forest was utilized as a stand-in paradigm for multi-objective optimization using the Grey Wolf optimizer, Whale optimization algorithm, and Arithmetic optimization algorithm. With Whale optimization method, Water pressure = 350 MPa, Standoff distance = 2.5 mm, and Traverse rate = 300 mm/min produced the highest performance with decreased DE (0.1238) and improved DR (2.0450). The Friedman test ( p < 0.001) and post-hoc analysis show that Whale optimization method outperforms Grey Wolf and Arithmetic optimization algorithms. The results of confirmation tests with the optimized parameters for the WOA showed good agreement with the predicted results, within a range of ± 3%.
SN - 2045-2322
ER -