Enhancing energy efficiency in power looms: utilizing regression machine learning for electrokinetic energy assessment

Abstract

This study focuses on optimizing electro-kinetic energy utilization in textile weaving units, addressing power interruptions, reducing electrical costs, and harnessing latent kinetic energy during the weaving process. Power looms, vital to textile production, often incur inefficiencies and increased operational costs due to underutilized kinetic energy. To address these challenges, three innovative electrokinetic energy harvesting approaches are proposed: the Modified Series-Parallel Piezo Matrix, bi-directional linear generation, and uni-directional non-linear power extraction methods. The research further incorporates the Super Lift and Ultra Lift DC-DC power conversion techniques to enhance energy efficiency. The Modified Series-Parallel Piezo Array integrates Piezoelectric elements to capture and convert discarded kinetic energy efficiently. The bi-directional linear generation method captures mechanical energy from both forward and backward movements in weaving, minimizing energy wastage. The uni-directional non-linear power extraction method optimizes energy recovery by targeting specific weaving cycle phases. Machine learning algorithms, including Gaussian Process Regression, Linear Regression, Neural Network, and Support Vector Machine, enable precise energy estimation for meticulous modeling and optimization. Rigorous validation through MATLAB simulations aims to bolster energy efficiency, trim operational costs, and promote sustainable energy practices in the textile industry, contributing to a more environmentally friendly and sustainable future for the sector.