Universal Battery Performance and Degradation Model for Electric Aircraft

Abstract

<jats:p>In this work, we generate a battery performance and thermal dataset specific to eVTOL use-cases and develop a fast and accurate performance and degradation model around that dataset. We use a machine-learning based physics-informed battery performance model to break the typically observed accuracy-computing cost trade-off. We fit the aging parameters for each cycle in a given cell's lifetime, and then model the evolution of those parameters using a new approach that combines traditional physics-based models, consisting of SEI film growth, charge loss, and Li Plating, along with a neural network in a universal ordinary differential equations (u-ODEs) framework.</jats:p>

Prognostics for Lithium-ion batteries for electric Vertical Take-off and Landing aircraft using data-driven machine learning

“…Although in [28] a nominal battery capacity of 3.0Ah is indicated in the dataset available at [24], all mission profiles, except VAH23, have a battery capacity of more than 3.0Ah during the first capacity test. For example, for mission profile VAH01, at the first capacity test, the battery has a capacity of 3.03Ah.…”

“…For mission profile VAH23, the battery has a capacity of 2.71Ah during the first capacity test. Thus, a capacity of 3.0Ah does not seem to be the nominal capacity for all battery cells considered in [28]. Dataset [24] contains 𝑄𝑐ℎ𝑎𝑟𝑔𝑒 𝑚,𝑐 𝑖 (mAh), the amount of charge supplied to the cell during charging.…”

A review of safety considerations for batteries in aircraft with electric propulsion

“…52 Empirical and data-driven models typically have a lower computational cost than physics-based models, but often have difficulty extrapolating to conditions outside of the data on which they were trained. 53 There are also data-driven approaches that directly predict the safety envelope of battery packs in electric vehicles. 54 The coupling of these safety envelopes with performance envelopes would be an important contribution in battery performance and safety modeling.…”

Neural Ordinary Differential Equations for Grey-Box Modelling of Lithium-Ion Batteries on the Basis of an Equivalent Circuit Model

“…The authors of Ref. [16] focus on physical battery modelling in combination with NODEs. They consider ageing effects such as solid electrolyte interface formation, lithium plating and active material isolation as well as the increase in the internal resistance.…”