Adaptive model-based state monitoring and prognostics for lithium-ion batteries

Abstract

Lithium-ion batteries feature high energy density, high power density, and long lifetime, making them preferable for the vehicle market. The battery management system (BMS) is the main control unit for battery systems. Among different functions, it monitors the key parameters including battery state of charge (SOC), state of health (SOH), and predicts the battery future conditions such as battery power capability, residual energy, and remaining useful life (RUL).

The main objective of this thesis is to develop advanced model-based algorithms for battery state monitoring and prognostics. After a comprehensive literature review, specific requirements of monitoring algorithms are elaborated. Owing to the nonlinear charging and discharging processes, the methods take various battery characteristics, uncertainties in the cell measurement as well as in the battery modeling into account. In case of large battery systems, further challenges arise from the limited computational resources of a BMS. Instead of multiplicating the existing techniques for individual cells, novel methods are proposed in this work, which aim at determining the pack conditions effectively and reliably without sacrificing the safety of the battery system.

In order to fulfill the task of verification, different approaches such as simulation with modeled cell variation, offline validation with measured data, and online test with a battery module and a state-of-the-art BMS are conducted. The proposed methods show promising results in the entire operating range, while the computational complexity can be significantly reduced.