Predicting the Mechanical Strength of Fire Protective Fabrics after Thermal Aging using Machine Learning

Abstract

Thermal aging leads to a reduction in the tensile strength of fire protective fabrics, which increases the skin burn risks of the wearer. Standardized test methods are generally destructive. In this study, machine learning was applied to predict the tensile strength after heat exposure. Training data was obtained from published articles, and seven features that affect the tensile strength of the fabric were determined. The results indicated that the average R2 and RMSE of machine learning models was 0.83 and 135.40, respectively, which was better than the traditional statistical model (R2 = 0.45, RMSE = 238.41). Among all the models, GBR produced the best prediction result (R2 = 0.95, RMSE = 77.42). Five features (fiber, weight, testing direction, exposure time, and heat flux density) were sufficient to achieve a better prediction.

Predicting the thermal protective performance of flame-retardant fabric based on machine learning

“…Liu et al (2020) used local skin temperature to predict outdoor thermal comfort based on a machine learning approach by building a support vector machine model (SVM) to predict cold discomfort, comfort and thermal discomfort in outdoor environments. Liu et al (2021) predicted the tensile strength of flame-retardant fabrics after thermal ageing based on machine learning algorithms. Five features were selected from the original seven features to enable the model to predict mechanical strength.…”