The application of machine learning-based state-of-health (SOH) prediction is hindered by the large demand for training data. To conquer this defect, a flexible and easily transferred SOH prediction scheme for lithium-ion battery packs is developed. First, the charging duration for a predefined voltage range is hired as the health feature to quantify capacity degradation. Then, the long short-term memory (LSTM) network and transfer learning (TL) with fine-tuning strategy are incorporated to constitute the cell mean model (CMM) for SOH prediction with partial training data. Next, to evaluate the SOH inconsistencies among cells, the LSTM model is employed as the cell difference model (CDM), and the minimum estimation value of CDM is identified to determine pack SOH. The experimental results reveal that, even when the first 360 cycle data, occupying only 40% in the whole 904 cycle data, are chosen and constituted to the data set for model training, the obtained estimation algorithm can still predict SOH precisely with the error of less than 3%, thus remarkably reducing the training data amount and mitigating the computation burden during model training. In addition, the preferable validation results on different types of lithium-ion batteries further manifest the extendibility of the proposed strategy.

Accurate state of health (SOH) prediction is significant to guarantee operation safety and avoid latent failures of lithium-ion batteries. With the development of communication and artificial intelligence technologies, a body of researches have been performed toward precise and reliable SOH prediction method based on machine learning (ML) techniques. In this paper, the conception of SOH is defined, and the state-of-the-art prediction methods are classified based on their primary implementation procedure. As an essential step in ML-based SOH algorithms, the health feature extraction methods reported in the literature are comprehensively surveyed. Next, an exhausted comparison is conducted to elaborate the development of ML-based SOH prediction techniques. Not only their advantages and disadvantages of the application in SOH prediction are reviewed but also their accuracy and execution process are fully discussed. Finally, pivotal challenges and corresponding research directions are provided for more reliable and high-fidelity SOH prediction.

In this paper, a novel state of health (SOH) estimation method based on partial charge voltage and current data is proposed. The extraction of feature variables, which are energy signal, the Ah-throughput, and the charge duration, is discussed and analyzed. The support vector machine (SVM) with radial basis function (RBF) as kernel function is applied for the SOH estimation. The predictive performance of the SOH by the SVM are performed with full and partial charging data. Experiment results show that the addressed approach enables estimating the SOH accurately for practical application.

Accurate estimation of state of charge (SOC) of lithium-ion battery packs remains challenging due to inconsistencies among battery cells. To achieve precise SOC estimation of battery packs, firstly, a long short-term memory (LSTM) recurrent neural network (RNN)-based model is constructed to characterize the battery electrical performance, and a rolling learning method is proposed to update the model parameters for improving the model accuracy. Then, an improved square root-cubature Kalman filter (SRCKF) is designed together with the multi-innovation technique to estimate battery cell’s SOC. Next, to cope with inconsistencies among battery cells, the SOC estimation value from the maximum and minimum cells are combined with a smoothing method to estimate the pack SOC. The robustness and accuracy of the proposed battery model and cell SOC estimation method are verified by exerting the experimental validation under time-varying temperature conditions. Finally, real operation data are collected from an electric-scooter (ES) monitoring platform to further validate the generalization of the designed pack SOC estimation algorithm. The experimental results manifest that the SOC estimation error can be limited within 2% after convergence.