Abstract

Flame retardants are going to be such an important component in future Lithium-based batteries. However, these flame retardants have a major problem: they decrease the conductivity values of electrolytes. During this project we have been trying to use some new optimization techniques (Genetic Algorithms) to get the better conditions that avoid this decrease. We have based this Genetic Algorithms in some computational predictions made by a statistical model created with a Random Forest (a Machine Learning technique). In order to build this statistical model we have used some experimental data collected in the lab, measuring conductivities of some randomly generated samples. These experimental results combined with some other physical properties described in bibliography, such as viscosity and dielectrical constant of the electrolytes, allows us to get some pretty accurate predictions.