A deep neural network for multi-fault diagnosis of battery packs based on an incremental voltage measurement topology

Energy, Год журнала: 2025, Номер 316, С. 134590 - 134590

Опубликована: Янв. 22, 2025

Язык: Английский

---

Concurrent multi-fault diagnosis of lithium-ion battery packs using random convolution kernel transformation and Gaussian process classifier

Energy, Год журнала: 2024, Номер 306, С. 132467 - 132467

Опубликована: Июль 22, 2024

Язык: Английский

---

Data-driven strategy: A robust battery anomaly detection method for short circuit fault based on mixed features and autoencoder

Applied Energy, Год журнала: 2025, Номер 382, С. 125267 - 125267

Опубликована: Янв. 8, 2025

Язык: Английский

---

Self-Supervised Time-Series Preprocessing Framework for Maritime Applications

Electronics, Год журнала: 2025, Номер 14(4), С. 765 - 765

Опубликована: Фев. 16, 2025

This study proposes a novel self-supervised data-preprocessing framework for time-series forecasting in complex ship systems. The integrates an improved Learnable Wavelet Packet Transform (L-WPT) adaptive denoising and correlation-based Uniform Manifold Approximation Projection (UMAP) approach dimensionality reduction. enhanced L-WPT incorporates Reversible Instance Normalization to improve training efficiency while preserving performance, especially low-frequency sporadic noise. UMAP reduction, combined with modified K-means clustering using correlation coefficients, enhances the computational interpretability of reduced data. Experimental results validate that state-of-the-art models can effectively forecast data processed by this framework, achieving promising MSE MAE metrics.

Язык: Английский

---

Fault mitigation and diagnosis for lithium-ion batteries: a review

Frontiers in Energy Research, Год журнала: 2025, Номер 13

Опубликована: Фев. 19, 2025

Due to their high energy density, long life cycle, minimal self-discharge (SD), and environmental benefits, lithium-ion batteries (LIBs) have become increasingly prevalent in electronics, electric vehicles (EVs), grid support systems. However, usage also brings about heightened safety concerns potential hazards. Therefore, it is crucial promptly identify diagnose any issues arising within these mitigate risks. Early detection diagnosis of faults such as Battery Management Systems (BMS) malfunctions, internal short circuits (ISC), overcharging, over-discharging, aging effects, thermal runaway (TR) are essential for mitigating risks preventing accidents. This study aims provide a comprehensive overview fault by meticulously examining prior research the field. It begins with an introduction significance LIBs, followed discussions on concerns, diagnosis, benefits diagnostic approaches. Subsequently, each thoroughly examined, along methods including both model-based non-model-based Additionally, elevates role cloud-based technologies real-time monitoring enhancing mitigation strategies. The results show how well approaches work increase LIB systems’ safety, dependability, economic feasibility while emphasizing necessity sophisticated growing use variety applications.

Язык: Английский

---

Temperature prediction of lithium-ion battery based on adaptive GRU transfer learning framework considering thermal effects decomposition characteristics

Energy, Год журнала: 2025, Номер unknown, С. 135504 - 135504

Опубликована: Март 1, 2025

Язык: Английский

---

Comparative Study of Multiple-Sensor-Fault-Detection Based Time–Frequency Analysis Methods on Lithium-Ion Batteries

Processes, Год журнала: 2025, Номер 13(4), С. 929 - 929

Опубликована: Март 21, 2025

Rapid multi-sensor fault detection is crucial for the battery management system (BMS). Almost all existing diagnosis methods current sensors are model-based, and complexity of models poses a huge challenge to their application in engineering. Firstly, this paper conducts detailed analysis physical meanings six forms sensor faults, these types faults modeled using mathematical methods. To better compare ability each method different standardized during modeling. Then, characteristics five time–frequency analyzed. Finally, multi-window short-time Fourier transform (MW-STFT) lithium-ion proposed. The experimental results show that proposed MW-STFT can detect faults.

Язык: Английский

---

Thermal fault detection of lithium-ion battery packs through an integrated physics and deep neural network based model

Communications Engineering, Год журнала: 2025, Номер 4(1)

Опубликована: Апрель 28, 2025

Battery packs develop faults over time, many of which are difficult to detect early. For instance, cooling system blockages raises temperatures but may not trigger alerts until protection limits exceeded. This work presents a model-based method for early thermal fault detection and identification in battery packs. By comparing measured estimated temperatures, the identifies including failed sensors, coolant pump malfunctions, flow blockages. The core is high-accuracy temperature estimation model, integrating physics-based model with neural network, achieves root mean square error 0.39 °C maximum 1 under US06 discharge 6C charge at 15 °C. Tested on 72-cell air-cooled pack, detects using only eight sensors within 13 45 minutes, zero false detections 11 testing cycles. approach enables alerts, enhancing reliability safety electric vehicles.

Язык: Английский

---

Identification of combined sensor faults in structural health monitoring systems

Smart Materials and Structures, Год журнала: 2024, Номер 33(8), С. 085026 - 085026

Опубликована: Июль 10, 2024

Abstract Fault diagnosis (FD), comprising fault detection, isolation, identification and accommodation, enables structural health monitoring (SHM) systems to operate reliably by allowing timely rectification of sensor faults that may cause data corruption or loss. Although is scarce in FD SHM systems, recent methods have included assuming one at a time. However, real-world include combined simultaneously affect individual sensors. This paper presents methodology for identifying occurring To improve the quality comprehend causes leading faults, (ICSF) based on formal classification types faults. Specifically, ICSF builds upon long short-term memory (LSTM) networks, i.e. type recurrent neural used classifying ‘sequences’, such as sets acceleration measurements. The validated using measurements from an system installed bridge, demonstrating capability LSTM networks thus improving systems. Future research aims decentralize reformulate models mathematical form with explanation interface, explainable artificial intelligence, increased transparency.

Язык: Английский

---

Multi-scenario failure diagnosis for lithium-ion battery based on coupling PSO-SA-DBSCAN algorithm

Journal of Energy Storage, Год журнала: 2024, Номер 99, С. 113393 - 113393

Опубликована: Авг. 20, 2024

Язык: Английский

---