Predicting air permeability and porosity of nonwovens with image processing and artificial intelligence methods

Abstract

This study was conducted to investigate the relationship between the fiber distribution and the permeability properties of the hydroentangled nonwoven fabrics in terms of air permeability and porosity. The fiber distribution and web properties of the hydroentangled nonwoven fabrics have significant influence on mechanical and physical performance of the finished products. Control of these properties during production, without physical testing, plays an important role in reducing response time to change the process parameters as well as production costs and material waste. In this study, an artificial intelligence method has been developed to predict the porosity and air permeability properties of hydroentangled nonwoven fabrics from their texture features. For this aim, two image processing algorithms were developed in order to measure the fabric porosity and to extract texture statistical features. For the prediction of the investigated fabric properties, an Artificial Neural Network (ANN) model was built. The investigated samples were composed of Polyester (PES) and Viscose (CV) fiber with different areal weights. High regression values were obtained between predicted and actual values for both porosity and air-permeability properties. According to ANN results, it was revealed that the air permeability and porosity properties of hydroentangled nonwovens can be predicted with high accuracy from their texture images.