Logo

e-ISSN: 2455-3743 | Published by Insightive Reaserch Private Limited

Home Aim & Scope Abstracting and Indexing Call For Papers Instructions For Authors Editorial Board Archives Conferences Download Ethics and Policies Publication Ethics and Publication Malpractice Statement Article withdrawal Policy Plagiarism Policy Peer Review Policy Copyright, Grants and Ownership Declaration Open Access Policy Advertising Policy Join Us

Archives of International Journal of Research in Computer & Information Technology(IJRCIT)

Volume 11 Issue 2 March 2026



1. A Stacking-Based Hybrid ANN–Ensemble Framework for Accurate Heat Transfer Prediction in Complex Systems

AUTHOR NAME : Niraj A. Dakhore, Ankit A. Jiwarkar, Ashish V. Kadu

ABSTRACT : Accurate prediction of heat transfer characteristics in complex thermal systems is essential for optimizing engineering design and improving energy efficiency. Traditional empirical and numerical methods often struggle to capture the nonlinear and multivariate nature of heat transfer phenomena, leading to reduced accuracy and high computational cost. This paper proposes a hybrid artificial intelligence-based framework that integrates Artificial Neural Networks with ensemble learning techniques, including Random Forest and Extreme Gradient Boosting, to enhance prediction performance. A dataset comprising approximately 5000 samples is utilized, incorporating key thermophysical parameters such as Reynolds number, Prandtl number, temperature, fluid velocity, and material properties. A systematic preprocessing pipeline involving data cleaning, normalization, outlier removal, and feature selection is applied to ensure data quality and model efficiency. The proposed model employs a stacking-based architecture to combine the strengths of individual learners, enabling robust modeling of complex nonlinear relationships. Performance evaluation is conducted using standard regression metrics, including the coefficient of determination, Root Mean Square Error, Mean Absolute Error, and Mean Absolute Percentage Error. Experimental results demonstrate that the proposed hybrid model outperforms conventional approaches, achieving an R² value of 0.999, RMSE of 0.003, MAE of 0.002, and MAPE below 2%. The findings indicate that the proposed framework offers high accuracy, strong generalization capability, and computational efficiency, making it suitable for real-world heat transfer applications such as heat exchanger design and thermal system optimization.

Download





© 2020 @ Global Advanced Research Publication House