Fine-tuning optimization of poly lactic acid impact strength with variation of plasticizer using simple supervised machine learning methods

Abstract

The use of machine learning to fine-tune the properties of materials is a remarkable achievement in the 21st century. Three machine learning (ML) methods were used to fine-tune and optimize the impact strength of polylactic acid (PLA) with different plasticizers: KNN (K-nearest neighbors), SVR (Support Vector Regression), and ANN (artificial neural net-works). The results demonstrated that, though ANN reached a higher R2 score of 0.901 than the other two ML methods, KNN, with an R2 score of 0.839, showed more stability than ANN. Based on the current research, KNN is recommended for experimentalists to fine-tune the impact strength of variational plasticizers. The experiment study case with polyethylene glycol 1000 (PEG1000) and octyl epoxy stearate (OES) plasticizer showed good agreement and prediction with experiments. It even showed the fine-tuned impact strength as a function of plasticizer content results, which cannot be achieved by only experiments.