Article
Thermodynamics
Xinghua Liu, Siqi Li, Jiaqiang Tian, Zhongbao Wei, Peng Wang
Summary: This work proposes a battery state of health (SOH) estimation method for Lithium-ion batteries in electric vehicles based on voltage reconstruction and fusion models. The proposed method reconstructs voltage curves using importance sampling, analyzes the correlation between extracted feature factors and SOH, and establishes a SOH estimation fusion model based on improved Support Vector Regression (SVR) and Convolutional Neural Network (CNN). Experimental results show that the proposed method outperforms other methods such as Gauss Process Regression (GPR), CNN, whale optimization algorithm-SVR (WOA-SVR), and Long Short Term Memory (LSTM) neural network, demonstrating its good performance.
Article
Automation & Control Systems
Qi Dai, Jian-wei Liu, Jia-Peng Yang
Summary: For class-imbalance problems, traditional supervised learning algorithms struggle with accurately identifying minority instances. Ensemble learning, by building multiple classifiers on the training dataset, can improve recognition accuracy for minority instances. Few researchers have used sliding windows to select majority instances and construct ensemble learning models. Therefore, this paper proposes a novel sliding window-based selective ensemble learning method (SWSEL) that utilizes similarity mapping and distance alignment to address the class-imbalance problem.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Environmental Sciences
Jiansi Ren, Ruoxiang Wang, Gang Liu, Yuanni Wang, Wei Wu
Summary: This paper proposes a Nested Sliding Window (NSW) method based on the correlation between pixel vectors to extract spatial information from hyperspectral images (HSI) and reconstruct original data. The NSW-PCA-SVM model combines NSW with Principal Component Analysis (PCA) and Support Vector Machine (SVM) to achieve high classification accuracy, with the advantage of easily adjustable parameters for better performance.
Article
Energy & Fuels
YongFang Guo, Kai Huang, XiaoYa Hu
Summary: This paper proposes a method for estimating the state of health of lithium-ion batteries, including a feature extraction framework, health indicator generation, and support vector regression model, which can improve accuracy and prediction precision.
JOURNAL OF ENERGY STORAGE
(2021)
Article
Energy & Fuels
Kang Liu, Longyun Kang, Lei Wan, Di Xie, Jie Li
Summary: This article presents a battery remaining useful life (RUL) prediction method that combines the sliding window technique and Box-Cox transformation. The method achieves accurate online RUL prediction with low computational burden. Experimental results demonstrate that the optimized method obtains lower prediction errors during the last 20% of battery lifetime and accurately predicts the RUL.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Energy & Fuels
Jinjin Shi, Haisheng Guo, Dewang Chen
Summary: This paper introduces a recursive least square parameter identification method using variable forgetting factor and the difference between open circuit and terminal voltages in sliding window mode, aiming to accurately identify lithium-ion battery model parameters. The approach adapts sliding window size according to working conditions, enhances information utilization by considering mean square value of voltage difference, and achieves accurate parameter identification with relatively narrow terminal voltage error range.
JOURNAL OF ENERGY STORAGE
(2021)
Article
Computer Science, Artificial Intelligence
Jie Xie, Kai Hu, Guofa Li, Ya Guo
Summary: This study presents a CNN-based method for driving behavior classification using multi-sliding window fusion. By constructing multiple sliding windows of different sizes to extract features and utilizing CNN for classification, the proposed method achieves a macro F1-score of up to 80.25% on the UAH-DriveSet dataset, outperforming other fusion methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Energy & Fuels
Lingfeng Fan, Ping Wang, Ze Cheng
Summary: This paper introduces a remaining capacity prediction technique for lithium-ion batteries based on partial charging curve and health feature fusion, with a battery aging model established by Gaussian process regression. The reliability and accuracy of the proposed method are validated on six battery data sets from NASA and the University of Oxford.
JOURNAL OF ENERGY STORAGE
(2021)
Article
Chemistry, Analytical
Shuang Qiu, Guangzhe Zhao, Xiao Li, Xueping Wang
Summary: This paper proposes a local Sliding Window Attention Network (SWA-Net) for facial expression recognition (FER), which uses a sliding window strategy for feature-level cropping to extract fine-grained features without complex preprocessing. It also introduces a local feature enhancement module and an adaptive local feature selection module to mine essential local features. Extensive experiments show that the SWA-Net model achieves comparable performance to state-of-the-art methods with scores of 90.03% on RAF-DB, 89.22% on FERPlus, and 63.97% on AffectNet.
Article
Energy & Fuels
Zhe Wang, Fangfang Yang, Qiang Xu, Yongjian Wang, Hong Yan, Min Xie
Summary: In this study, battery measurements are organized as a graph structure and utilized using a graph neural network. Specific data aggregation and feature fusion operations are selected using neural architecture search, which improves adaptability. The proposed scheme is validated using two public datasets and additional discussions emphasize the capability of the graph neural network and the necessity of architecture searching. Comparison analysis and performance under noisy environment demonstrates the superiority of the proposed scheme.
Article
Chemistry, Physical
Quanqing Yu, Yuwei Nie, Shizhuo Liu, Junfu Li, Aihua Tang
Summary: In this study, a lithium-ion battery state of health estimation method with dynamic operating conditions generalization is proposed. The method utilizes an equivalent circuit model and feature fusion to achieve accurate estimation, while having low computational requirements and potential applicability.
JOURNAL OF POWER SOURCES
(2023)
Review
Computer Science, Artificial Intelligence
Aditya Kumar, Jainath Yadav
Summary: This study presents and analyzes various feature set partitioning (FSP) methods, comparing their performance for the multi-view ensemble learning (MEL) framework. Different classification models are applied to the partitioned views and their predictions are combined. The classification performance of recent FSP methods is compared using nonparametric statistical analysis.
INFORMATION FUSION
(2023)
Article
Engineering, Environmental
Huixing Meng, Qiaoqiao Yang, Enrico Zio, Jinduo Xing
Summary: In this paper, a methodology for dynamic risk prediction of thermal runaway in lithium-ion batteries is proposed, which integrates fault tree, dynamic Bayesian network, and support vector regression. The proposed methodology can be used for risk early warning of thermal runaway in lithium-ion batteries.
PROCESS SAFETY AND ENVIRONMENTAL PROTECTION
(2023)
Article
Energy & Fuels
Junqi Bai, Jiayin Huang, Kai Luo, Fan Yang, Yanhua Xian
Summary: This paper proposes a feature reuse-based multi-model fusion method for precise estimation of lithium-ion batteries' state of health (SOH). Four features are extracted from multiple perspectives and used to generate preliminary SOH predictions, which are then fused with input features using Bayesian linear regression. Experimental results show that the proposed method achieves better accuracy and robustness compared to other methods.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Energy & Fuels
Shaheer Ansari, Afida Ayob, M. S. Hossain Lipu, Aini Hussain, Maher G. M. Abdolrasol, Muhammad Ammirrul Atiqi Mohd Zainuri, Mohamad Hanif Md. Saad
Summary: This article presents a novel framework for predicting the remaining useful life (RUL) of lithium-ion batteries (LIB) using a hybrid optimized data-driven approach. The framework combines a cascaded forward neural network (CFNN) with the innovative jellyfish optimization (JFO) technique. It utilizes a mathematical systematic sampling (SS) method to select relevant data features, and employs the overlapping sliding window (OSW) technique to mitigate the capacity regeneration effect. The performance of the proposed hybrid JFO-CFNN model is validated by comparing with other models and using another battery dataset, showing high prediction accuracy and applicability.
JOURNAL OF ENERGY STORAGE
(2023)