A Deep Learning Model Based on the Introduction of Attention Mechanism is Used to Predict Lithium-Ion Battery SOC

Journal of The Electrochemical Society, Год журнала: 2024, Номер 171(7), С. 070508 - 070508

Опубликована: Июнь 28, 2024

In order to enhance the accuracy of state charge (SOC) prediction for lithium-ion batteries, this paper developed a deep learning model optimized by slime mold algorithm (SMA). This combines convolutional neural networks, bidirectional gated recurrent units, and attention mechanisms. Through SMA optimization critical parameters in model, predictive performance has been significantly improved. experimental phase, collected discharge data from an 18650 battery pack under 8 different operating conditions, totaling 10596 sets data. These were used fully train validate model. The test results show that new demonstrates exceptional SOC prediction, with average absolute error, root mean square percentage error reaching 0.46462%, 0.56406%, 6.8028%, respectively. Moreover, decision coefficient R reaches 0.962. result not only surpasses single models unoptimized but also provides important technical support improving life driving range electric vehicles.

Язык: Английский

---

Review of imbalanced fault diagnosis technology based on generative adversarial networks

Journal of Computational Design and Engineering, Год журнала: 2024, Номер 11(5), С. 99 - 124

Опубликована: Авг. 31, 2024

Abstract In the field of industrial production, machine failures not only negatively affect productivity and product quality, but also lead to safety accidents, so it is crucial accurately diagnose in time take appropriate measures. However, machines cannot operate with faults for extended periods, diversity fault modes results limited data collection, posing challenges building accurate prediction models. Despite recent advancements, intelligent diagnosis methods based on traditional sampling learning have shown notable progress. Nonetheless, these heavily rely human expertise, making challenging extract comprehensive feature information. To address challenges, numerous imbalance generative adversarial networks (GANs) emerged, GANs can generate realistic samples that conform distribution original data, showing promising diagnosing imbalances critical components such as bearings gears, despite their great potential, GAN face including difficulties training generating abnormal samples. whether GAN-based resampling technology or technology, there are fewer reviews noise-containing imbalance, intra- inter-class dual multi-class series other problems small samples, a lack more summary solutions above problems. Therefore, purpose this paper deeply explore under various failure modes, review analyze research basis. By suggesting future directions, aims provide guidance reference production maintenance.

Язык: Английский

---

Exploring the potential of various nanofluids for thermal management of a lithium-ion battery

Applied Thermal Engineering, Год журнала: 2024, Номер 261, С. 125177 - 125177

Опубликована: Дек. 7, 2024

Язык: Английский

---

Uncertainty‐aware nuclear power turbine vibration fault diagnosis method integrating machine learning and heuristic algorithm

IET Collaborative Intelligent Manufacturing, Год журнала: 2024, Номер 6(3)

Опубликована: Июнь 7, 2024

Abstract Nuclear power turbine fault diagnosis is an important issue in the field of nuclear safety. The numerous state parameters operation and maintenance turbines are collected, forming a complex high‐dimensional feature space. These spaces contain redundant information, which increases training cost reduces recognition accuracy efficiency model. To address aforementioned challenges, vibration algorithm proposed. First, long short‐term memory‐based denoising autoencoder (LDAE) designed to enhance capability uncertainty awareness. Then, extraction method integrating variational mode decomposition (VMD), L‐cliffs‐based effective selection, sample entropy devised extract latent features from Furthermore, using extreme gradient boosting (XGBoost) as classifier, LDAE‐VMD‐XGBoost model constructed for turbines. Considering impact multiple hyperparameters on performance, pathfinder used optimise hyperparameter settings improve accuracy. Experimental results demonstrate performance proposed improved accurate diagnosis.

Язык: Английский

---

A Knowledge Transfer Framework Based on Deep Reinforcement Learning for Multi-stage Construction Projects

IEEE Transactions on Engineering Management, Год журнала: 2024, Номер 71, С. 11361 - 11374

Опубликована: Янв. 1, 2024

Язык: Английский

---