Lithium-ion batteries play an essential role in the global trend of vehicle electrification as they can provide a good balance among energy density, power capability, lifespan, environmental impact, and cost. To achieve a safe yet effective utilization of these batteries, it is necessary for us to manage the cells. In vehicle applications where thousands of cells are integrated, the management could be complicated and categorized into different classes, such as safety-, function-, and optimization-levels.

This thesis focuses on the research problems in function-level battery management, which provides a basis for the vehicle’s regular operation. It covers the issues such as the estimation of the battery states and parameters, the prediction of battery remaining useful life, and the con...[ Read more ]

Lithium-ion batteries play an essential role in the global trend of vehicle electrification as they can provide a good balance among energy density, power capability, lifespan, environmental impact, and cost. To achieve a safe yet effective utilization of these batteries, it is necessary for us to manage the cells. In vehicle applications where thousands of cells are integrated, the management could be complicated and categorized into different classes, such as safety-, function-, and optimization-levels.

This thesis focuses on the research problems in function-level battery management, which provides a basis for the vehicle’s regular operation. It covers the issues such as the estimation of the battery states and parameters, the prediction of battery remaining useful life, and the control of the equalization hardware. Specifically, ΣΔ-based and Bayes-Monte-Carlo-based state estimators have been developed to determine the battery state-of-charge (remaining percentage of battery capacity). An enhanced incremental-capacity-analysis-based criterion has been proposed to calculate the state-of-health (degree of aging) with simple linear mappings. For the calculation of the state-of-power (peak power capability), a seeking-based approach has been developed to solve the inverse of a nonlinear, parameter-adaptive battery model. Model-migration-based methods have been developed to predict the remaining useful life of the cells, together with particle filtering or neural networks. And run-to-run control algorithms are also introduced into the field of battery management to handle the battery equalization problems, along with a new balancing-current-ratio-based control scheme.

Notably, potential solutions to the following two open questions in the research field, namely, 1) how to predict the battery’s lifespan before the “knee-point” and 2) how to equalize the cells with voltage plateaus (weakly-observable regions), are proposed. In addition, the developed algorithms are rich in engineering insights and could balance the computational burden and the algorithm accuracy in a satisfactory way.