Lithium-ion batteries are essential for a wide range of applications due to their high energy density and rechargeability. However, their production and performance improvement often rely on time-consuming and expensive experiments. Typically, simulation analyzes is used to build process models to optimize battery performance and lifetime. However, simulations cannot always take into account the limitations of the manufacturing process. As a result, the process parameters determined by such a model remain largely theoretical. For the efficient production of lithium-ion batteries, an understanding of the relationships between physical quantities is essential to meet optimized performance and safety standards. However, these relationships are quite complex, making it difficult for traditional methods to determine the physical model. Alternatively, artificial intelligence methods can be beneficial. This work uses a novel gray-box modeling technique that incorporates physical knowledge and empirical data. To achieve this goal, we combine a deep neural network with a genetic algorithm to determine the existing physical relationships within the data and then estimate the final model for the system.