Yang Gao

This study focuses on the safety and reliability issues of lithium-ion batteries, proposing a fault diagnosis strategy that leverages dual-feature extraction from both the time and frequency domains. Additionally, by modifying the traditional autoencoder, the study proposes a feature-guided autoencoder as an unsupervised model for extracting features in the time domain. Initially, wavelet packet decomposition and its energy-denoising treatment are employed to refine fault information within battery voltage signals. Subsequently, the reconstruction error outputted by the Feature-Guided Autoencoder is utilized as the time-domain fault feature, while the cosine similarity of the energy of signals in various frequency bands obtained after wavelet packet decomposition serves as the frequency-domain fault feature. Ultimately, this article selects the Isolation Forest algorithm for two-dimensional outlier detection of time and frequency features. Experimental results demonstrate that the feature-guided autoencoder proposed in this study not only enhances the sensitivity of time-domain fault features compared to traditional autoencoders and their variants but also optimizes issues related to training time and computational load. The effectiveness of the proposed dual-feature fault diagnosis method in both the time and frequency domains is validated through data from two actual vehicles, showing superior early fault detection capability relative to single-feature fault diagnosis methods.

With the development of electric vehicles, the demand for lithium-ion batteries has been increasing annually. Accurately estimating the State of Health (SOH) of lithium-ion batteries is crucial for their efficient and reliable use. Most of the existing research on SOH estimation is based on parameters such as current, voltage, and temperature, which are prone to fluctuations. Estimating the SOH of lithium-ion batteries based on Electrochemical Impedance Spectroscopy (EIS) and data-driven approaches has been proven effective. In this paper, we explore a novel SOH estimation model for lithium batteries based on EIS and Convolutional Neural Network (CNN)-Vision Transformer (VIT). The EIS data is treated as a grayscale image, eliminating the need for manual feature extraction and simultaneously capturing both local and global features in the data. To validate the effectiveness of the proposed model, a series of simulation experiments are conducted, comparing it with various traditional machine learning models in terms of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R²). The simulation results demonstrate that the proposed model performs best overall in the testing dataset at three different temperatures. This confirms that the model can accurately and stably estimate the SOH of lithium-ion batteries without requiring manual feature extraction and knowledge of battery aging temperature.

In order to ensure the driving safety of electric vehicles and avoid potential failures, it is important to properly estimate the state of health (SOH) of lithium-ion batteries. In this paper, a method of lithium-ion battery SOH estimation based on electrochemical impedance spectroscopy (EIS) and an algorithm fused by Elman neural network and cuckoo search (CS-Elman) is proposed. First, by extracting 19 features of EIS and using principal component analysis to reduce dimension, we obtain four principal components as model inputs. Second, CS algorithm optimizes the weights and thresholds of Elman algorithm. Next, we use the CS-Elman model to estimate the battery SOH and verify the model with the remaining battery data. In addition, we propose a variable temperature estimation model and verify the feasibility of the model between 25 °C and 45 °C. Finally, the experimental results show that the mean absolute error of the method is less than 1.36%.