Efficient Fault Diagnosis in Lithium-Ion Battery Packs: A Structural Approach with Moving Horizon Estimation

Amir Farakhor, Di Wu, Yebin Wang, Huazhen Fang · 2025

Read Efficient Fault Diagnosis in Lithium-Ion Battery Packs: A Structural Approach with Moving Horizon Estimation by Amir Farakhor, Di Wu, Yebin Wang, Huazhen Fang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-00305.txt.

Introduction

Amir Farakhor, Di Wu, Yebin Wang, and Huazhen Fang design in 2025 a leaner way to spot faults inside lithium-ion battery packs. Finding shorts and sensor faults early keeps electric vehicles and storage systems safe, but standard model-based diagnosis is heavy to compute and hungry for sensors.

Their method leans on two facts about real packs: the cells are nearly uniform, and faults are rare. They feed that structure into a moving horizon estimator, which reads recent measurements and infers the fault signals. The result aims at accurate diagnosis with less hardware and less computing than current practice demands.

Catalog record

CREATORAmir Farakhor, Di Wu, Yebin Wang, Huazhen Fang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-00305

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Efficient Fault Diagnosis in Lithium-Ion Battery Packs: A Structural Approach with Moving Horizon Estimation Amir Farakhor, Di Wu, Yebin Wang, Huazhen Fang Abstract Safe and reliable operation of lithium-ion battery packs depends on effective fault diagnosis. However, model-based approaches often encounter two major challenges: high computational complexity and extensive sensor requirements. To address these bottlenecks, this paper introduces a novel approach that harnesses the structural properties of battery packs, including cell uniformity and the sparsity of fault occurrences. We integrate this approach into a Moving Horizon Estimation (MHE) framework and estimate fault signals such as internal and external short circuits and faults in voltage and current sensors. To mitigate computational demands, we propose a hierarchical solution to the MHE problem. The proposed solution breaks up the pack-level MHE problem into smaller problems and solves them efficiently. Finally, we perform extensive simulations across various battery pack configurations and fault types to demonstrate the effectiveness of the proposed approach. The results highlight that the proposed approach simultaneously reduces the computational demands and sensor requirements of fault diagnosis. I INTRODUCTION Lithium-ion battery systems are pivotal to the electrification of transportation and the transition towards energy decarbonization. Their success can be attributed to key characteristics such as high energy density, long cycle life, and low self-discharge rates [ 1 ] . However, lithium-ion batteries also present significant safety risks due to the possibility of thermal runaway, which can lead to fires or even explosions [ 2 ] . Several well-publicized incidents have highlighted these safety concerns, particularly as lithium-ion batteries are increasingly used in safety-critical applications, such as electric aircraft [ 3 , 4 ] . To mitigate these risks, battery management systems (BMS) are equipped with advanced fault diagnosis capabilities. A primary function of BMS is to monitor cell behavior via sensor measurements and take appropriate actions in the event of an anomaly [ 5 ] . Given the importance of this role, fault diagnosis of lithium-ion battery packs has become a focus of extensive research. However, current approaches face two significant challenges: 1) they rely heavily on short-term, real-time sensor measurements, overlooking the long-term behavior of cells; and 2) many advanced fault diagnosis methods are computationally intensive, making them difficult to implement on the embedded processors typically found in BMS [ 6 ] . This paper addresses these challenges by proposing a novel approach that leverages the structural properties of battery packs to enhance fault diagnosis. Next, we will review the literature on fault diagnosis of lithium-ion battery packs. I-A Literature Review Existing studies on fault diagnosis can be broadly classified into two categories: model-based [ 7 , 8 , 9 ] and data-driven approaches [ 10 , 11 , 12 ] . Model-based approaches construct a physical model of the battery pack and compare sensor measurements with the model’s predictions, identifying any significant discrepancies as potential faults. In contrast, data-driven approaches utilize collected data to train models capable of distinguishing between normal and abnormal battery pack operation. While both approaches have their respective advantages and limitations, this paper adopts a model-based approach. Accordingly, this literature review focuses on the model-based fault diagnosis methods. Recent research on model-based fault diagnosis for lithium-ion battery packs has grown significantly [ 13 , 7 ] . One major group targets internal and external short circuits in lithium-ion cells [ 14 , 9 , 15 , 16 ] . Another group focuses on diagnosing voltage and current sensor faults [ 17 , 18 ] , while some address thermal anomalies in pouch cells [ 19 , 20 ] . However, running multiple fault detection schemes simultaneously in a BMS is impractical. Ideally, a single algorithm should address both cell and sensor faults. Some studies have attempted to achieve this by using structural analysis for multi-fault diagnosis [ 21 , 22 , 23 ] , but these methods often require numerous sensors, adding cost and complexity. Others propose active fault diagnosis [ 8 ] , though its applicability is limited to reconfigurable battery packs. In hindsight, existing approaches fall short in providing an efficient and practical fault diagnosis solution for entire battery packs. Key challenges include: 1) the need for solutions that apply to full battery packs, not just individual cells; 2) minimizing the number of required sensors; and 3) developing algorithms capable of handling multiple fault types, including both cell and sensor faults. This paper aims to address these limitations and advance the state of the art in lithium-ion battery fault diagnosis. (a) (b) Fig. 1: The considered battery pack configurations. (a) Series-parallel ( n S m P). (b) Parallel-series ( m P n S). I-B Statement of Contributions Despite recent advances, model-based fault diagnosis approaches have yet to reach practical maturity. Existing studies typically rely solely on sensor measurements, overlooking the valuable information provided by the structural properties of the battery pack. This paper challenges this conventional approach by demonstrating how these properties can be exploited to improve fault diagnosis. The key contributions are as follows: 1) We propose leveraging the structural properties of battery packs—specifically the uniformity of cells and the sparsity of fault occurrences—to enhance fault diagnosis. To achieve this, we formulate the fault diagnosis problem within the Moving Horizon Estimation (MHE) framework. The MHE problem estimates fault signals, including internal and external short circuits and current and voltage sensor faults, across various battery pack configurations (series, parallel, parallel-series, and series-parallel). This contribution enables fault diagnosis with significantly fewer sensors compared to existing studies in the literature. 2) To address the computational complexity of MHE, we present a hierarchical solution that decomposes the large-scale, pack-level problem into smaller, module-level problems that can be solved in parallel. This approach significantly reduces computational demands while maintaining diagnostic accuracy. A preliminary version of this work appeared in [ 24 ] , but the current paper substantially extends the study by covering all battery pack configurations and introducing a computationally efficient hierarchical solution to the MHE problem. II PROBLEM FORMULATION This section first presents the battery pack configurations under consideration, along with their corresponding sensor placements. Next, we introduce an electro-thermal model, and finally, we formulate the fault diagnosis problem within the MHE framework. II-A Battery Pack Structure Fig. 1 illustrates the two battery pack configurations considered for fault diagnosis: series-parallel ( n S m P) and parallel-series ( m P n S). In the n S m P configuration (Fig. 1 (a)), m parallel modules are formed, each consisting of n serially connected cells. We equip the n S m P configuration with n\times m voltage sensors, m temperature sensors and a single current sensor. The voltage sensors measure the terminal voltages of all the cells and the temperature sensors measure the aggregate temperature of the cells within a module. The m P n S configuration, on the other hand, consists of n serially connected modules, each containing m parallel cells (Fig. 1 (b)). For this configuration, a single current sensor measures the pack current, while each module is provided with one voltage sensor and one temperature sensor. Therefore, an m P n S configuration requires n voltage and temperature sensors, along with one current sensor for the entire pack. In both battery pack configurations, the

Read the complete text