Calibration and measurement of cotton moisture using real time system with statistical analysis

Abstract

Accurate moisture measurement in cotton is essential for maintaining fibre quality, ensuring safe storage, and supporting efficient processing. Improper moisture levels can result in microbial growth, fibre degradation, or mechanical damage during ginning and spinning operations. This study presents the development of a real-time moisture measurement system for cotton used in the ginning industry. The system operates on the principle of electrical resistance change to detect varying moisture levels. Cotton samples were categorized into four types: wet, new, old, and dry. The system is designed for use on moving or in-process cotton. To evaluate system performance, linear discriminant analysis (LDA), and hierarchical clustering analysis (HCA) were employed for classification. Partial least squares (PLS) regression was used to calibrate the system against the standard oven-drying method (ASTM D2495-07). Further, artificial neural network (ANN) modelling was applied for moisture prediction. The system successfully discriminated between the cotton types, achieving over 85% explained variance in classification. ANN-based prediction aligned closely with the standard reference method. The developed system provides a low-cost, fast, and real-time solution for moisture measurement in cotton, with strong potential for industrial application.