Physics-informed data-driven model of dehydration reaction stage in the sintering process of ternary cathode materials

The sintering process is the main process that affects the performance of ternary cathode materials. Due to the closed, continuous, and long-term characteristics of the sintering process carried out in a roller kiln, the reaction state including reaction degree, particle size, and etc., of the raw materials cannot be directly obtained through sampling detection, resulting in ineffective temperature control. This article mainly focuses on modeling the dehydration reaction stage of the sintering process to obtain the reaction state of the raw materials. Firstly, a two-sphere model is established based on Newton's second law and the particle volume conservation equation to describe the relationship between the particle radius and sintering temperature of lithium hydroxide and ternary precursor; Secondly, the chemical kinetic parameters of the dehydration reaction in the two-sphere model are obtained by combining the single heating rate scanning method with the multiple heating rate scanning method; Then, in order to improve the accuracy of the two-sphere model and reduce the impact of external environmental temperature changes, a physics-informed neural network model is constructed by combining the two-sphere model with the neural ordinary differential equation model. Finally, the effectiveness of the modeling method is verified through numerical simulation and empirical data.