Prediction of thermo-physiological properties of plated knits by different neural network architectures

Abstract

Thermo-physiological properties of polyester-cotton plated knits have been predicted using two different network architectures (NA1 & NA2). NA1 consists of four individual networks working in tandem with common set of inputs and NA2 consists of one network giving four outputs. It is found that network architecture NA1 is able to predict the thermo-physiological properties of plated fabrics better as compared to NA2 network architecture. Sensitivity analysis is performed to judge the sensitivity or the importance of each input parameter in determining thermo-physiological properties of plated fabrics. The most sensitive parameter in prediction of thermal resistance is total yarn linear density, filament fineness for thermal absorptivity, loop length for air permeability and moisture vapour transmission rate.