Machine learning models for material selection: Framework for predicting flatwise compressive strength using ANN

The Polyurethane foam cores (PUR) are synthesized by combining reactive chemicals namely Isocyanates and Polyols in appropriate proportions. The paper presents the physical and ANN approach of forming the PUR foam cores with change in chemical compositions. Five varied proportions with five densities of foams were synthesized and tested. The paper reports the physical properties in comparison with flatwise compression strength. It is observed that the physical test data is in absolute compatibility with the ANN results. The regression values are in good agreement with the 50:50 chemical compositions and higher density foams are observed to be brittle and lower density foams were flexible and elastic.