In this work we employ machine learning (ML) approaches to learn particle interactions using data from quantum mechanics (QM) simulations. The emergence of AI and ML has accelerated research in battery technology and material discovery. The study emphasizes the importance of databases, molecular representation, and AI models in predicting material properties accurately. ML approach can be a quicker and more effective substitute for conventional quantum mechanics calculations. Through the application of these acquired characteristics in molecular dynamics simulations, we can forecast several material properties, assisting scientists in the creation of new materials without requiring experimental data. We learned the Buckingham likelihood, a kind of non-bonded interaction, using our machine learning approach. We then calculated the concentrations of four distinct compounds using these projected values, with an accuracy rate of more than 99%. This illustrates how well our method predicts the properties of materials.