Analysis of dynamic process models for aviation battery systems

Abstract

The paper analyzes existing modeling approaches, including empirical, physicochemical and statistical methods, as well as machine learning methods. The advantages and limitations of models such as the Shepherd’s model, Butler-Volmer equation-based models, regression analysis-based models, and Long Short-Term Memory (LSTM) neural networks are discussed. Special attention is paid to a promising Method of Mathematical Prototyping of Energy Processes (MMPEP), this approach enables the construction of physically accurate models that conform with the fundamental laws of thermodynamics and electrodynamics. Based on MMPEP, a new voltage dynamics model has been developed specifically for lithium-ion aircraft batteries (LIABs), which take into account polarization processes, temperature changes and nonlinear effects. The model proposed in the paper is derived through numerical-analytical transformation of dynamic processes equations obtained by the method of mathematical prototyping of energy processes. A comparative analysis of existing modeling approaches is carried out and the advantages of the proposed MMPEP method are shown. An example of modeling the dynamics of physical and chemical processes in a lithium-ion battery with some limitations is presented. The research results demonstrate that the MMPEP-based models have high accuracy and versatility, which makes them applicable for charge state prediction, failure diagnostics, and digital twin development. The analytical expression presented in the paper expands the classical Shepherd’s model, providing a description of complex dynamic processes. The methodological potential of MMPEP is supported by the possibility of integration with machine learning methods to refine model parameters. Prospects for further research include extending the model to account for battery degradation, developing simplified models for real-time diagnostics systems, and introducing hybrid modeling approaches.