Skip to main content
Top

2024 | Book

A Data-Driven Fleet Service: State of Health Forecasting of Lithium-Ion Batteries

insite
SEARCH

About this book

Given the limitations of state-of-the-art methods, this book presents a state of health (SOH) forecasting method that is suitable for lithium-ion battery (LIB) systems in real-world battery electric vehicle operation. Its histogram-based features can capture the higher operational variability compared to constant and controlled laboratory operation. Also, the transferability of a trained machine learning model to new LIB cell types and new operational domains is investigated. The presented SOH forecasting method can be provided as a cloud service via a web or smartphone app to fleet managers. Forecasting the SOH enables fleet managers of battery electric vehicle fleets to forecast and plan vehicle replacements.

Table of Contents

Frontmatter
Chapter 1. Introduction
Abstract
In battery electric vehicles (BEVs), lithium-ion batteries (LIBs) are currently the most important battery technology. Over the lifetime of a LIB, the battery’s energy content, i.e., capacity, which determines the vehicle’s range decreases due to battery aging depending on its usage and environmental conditions. This chapter describes limitations of State of the Art methods for LIB state of heath (SOH) forecasting and derives four research questions.
Friedrich von Bülow
Chapter 2. Theoretical Background
Abstract
To answer the research questions considering the current limitations of state of heath (SOH) forecasting methods, background from two fields is required: Given the importance of lithium-ion batteries (LIBs) as power source of battery electric vehicles (BEVs), first domain knowledge on LIB cells, their aging, and their composition to battery modules and battery packs is provided. Second, relevant basics of supervised machine learning (ML) models for regression are presented.
Friedrich von Bülow
Chapter 3. Towards State of Health Forecasting of Lithium-Ion Batteries
Abstract
The terms state of heath (SOH) estimation, SOH prediction, SOH forecasting, and remaining useful life (RUL) prediction are crucial to distinguish. However, these terms are used inconsistently in the literature and their conceptualization is complex and vague until now. Thus, this chapter introduces a formal concept and definitions mainly for SOH estimation and SOH forecasting. Furthermore, the relevance of model transfer for SOH forecasting to a new battery type is motivated. This chapter also delimits the object of investigation and motivates key criteria SOH forecasting models shall fulfill.
Friedrich von Bülow
Chapter 4. Related Work
Abstract
Given the delimitation of the object of investigation and the key criteria required for a suitable state of heath (SOH) forecasting model, building upon these key criteria in this chapter a structured literature survey for SOH forecasting models is conducted. The related work is structured and discussed regarding the limitations of applicability and comparability with existing models. Also, the limitations of transferability in existing literature on battery aging models are shown. This chapter concludes with an analysis of the existing research gap.
Friedrich von Bülow
Chapter 5. Data
Abstract
The training data suitable for state of heath (SOH) forecasting should be obtained from several batteries which have aged under a wide range of operational scenarios. This enables the model to forecast the SOH of batteries given an aging scenario encoded in the stressors. Here different data sets are used: Five public battery data sets from laboratory operation and one non-public battery data set from battery electric vehicle (BEV) fleet operation are introduced in this chapter.
Friedrich von Bülow
Chapter 6. Battery Cell State of Health Forecasting
Abstract
State of heath (SOH) forecasting is implemented using a public lithium-ion battery (LIB) cell data set from laboratory operation (Research Question.1). Battery electric vehicle (BEV) fleet managers can improve the operation and replacement of their fleet members by applying the proposed SOH forecasting model.
Friedrich von Bülow
Chapter 7. Transfer of Battery Cell State of Health Forecasting
Abstract
A model obtained in the previous chapter is used as base for a model transfer to other laboratory lithium-ion battery (LIB) cell data sets (Research Question.2). To enable quick applicability on new batteries two aspects are examined: "When to transfer” concerns the data availability in the target domain. "How to transfer" concerns the transfer method.
Friedrich von Bülow
Chapter 8. Battery System State of Health Forecasting
Abstract
The state of heath (SOH) forecasting method from the previous chapter is applied to a real-world lithium-ion battery (LIB) pack data set from mobility on demand (MOD) battery electric vehicle (BEV) fleet operation (Research Question.3).
Friedrich von Bülow
Chapter 9. Concept for a Technical Implementation
Abstract
This chapter discusses Research Question.4: "How can state of health (SOH) forecasting of lithium-ion batteries (LIBs) in automotive applications technically be realized?" To answer this question also necessary infrastructure and systems are discussed especially from the perspective of the battery electric vehicle (BEV) manufacturer. Also, BEV fleets, their stakeholder roles and a brief overview on the state of the art of fleet management is given.
Friedrich von Bülow
Chapter 10. Limitations & Outlook
Abstract
The answers to the research questions (RQs) developed and presented in the previous chapters are critically reflected. Based on their limitations possibilities for future research are presented. These are structured by the following categories: Method, future data, battery systems, and further learning paradigms.
Friedrich von Bülow
Chapter 11. Conclusion
Abstract
State of health (SOH) forecasting of lithium-ion batteries (LIBs) concerns the influence of battery usage on their SOH which is especially relevant for battery electric vehicle (BEV) fleets. Until now, applicability of SOH forecasting methods to BEV operation has been limited due to a) smaller operational variability from laboratory operation compared to real-world vehicle operation, b) assuming the same load during whole battery life, c) a focus of methods only on cells, but not on battery systems assembled in BEV, and d) a lack of transferability to different battery domains, e.g., new battery types or new operational domains. These limitations have been tackled by the SOH forecasting method presented.
Friedrich von Bülow
Backmatter
Metadata
Title
A Data-Driven Fleet Service: State of Health Forecasting of Lithium-Ion Batteries
Author
Friedrich von Bülow
Copyright Year
2024
Electronic ISBN
978-3-658-43188-4
Print ISBN
978-3-658-43187-7
DOI
https://doi.org/10.1007/978-3-658-43188-4

Premium Partner