PREDICTING THE THERMAL INSULATION PROPERTIES OF TWILL WOVEN COTTON FABRIC BY USING ANN AND ANFIS

This study analyzes two machine learning models, artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS), to predict thermal insulation of cotton fabric woven with twill. The input parameters include fabric thickness, ends per inch (EPI), and picks per inch (PPI). The ANN model has a 3-8-1 network structure, with output and hidden layers having sigmoid and linear activation functions. The ANFIS model employs sugeno-type fuzzy logic, while the network is trained using the feedforward backpropagation Levenberg-Marquardt technique. The weighted average approach was used in the defuzzification process. MATLAB was used to create both models. The ANN model performs well in predictions, as evidenced by its R2 value of 0.9942, which indicates a significant correlation between the target and prediction values. The ANN model's exceptional performance metrics, such as a mean absolute percentage error (MAPE) of 1.31401 and a root mean squared error (RMSE) of 0.00176, demonstrate its precision and reliability. However, the ANFIS model has considerably lower accuracy metrics, with an R2 value of 0.9570. The ANN offers more accuracy and precision than the ANFIS model, which has an RMSE of 0.00489 and a MAPE of 2.07495. This study will improve the textile engineering prediction model by revealing the intricate connection between fabric characteristics and the thermal insulation of clothing composed of cotton fabric's twill structure.