•We propose a deep learning-based capacity estimation method.•Our method incorporates the concepts of transfer learning and ensemble learning.•Transfer learning reduces the data-collection efforts by ...improving the learning performance.•Ensemble learning improves the accuracy and robustness in capacity estimation.•Cycling data from implantable and 18,650 cells are used to demonstrate the improvements.
It is often difficult for a machine learning model trained based on a small size of charge/discharge cycling data to produce satisfactory accuracy in the capacity estimation of lithium-ion (Li-ion) rechargeable batteries. However, in real-world applications, collecting long-term cycling data is a costly and time-consuming process. To overcome this difficulty, we propose a deep learning-based capacity estimation method that incorporates the concepts of transfer learning and ensemble learning. We target the applications where only a relatively small set of training data is available. Transfer learning is a knowledge learning method that leverages the knowledge learned from a source task to improve learning in a related but different target task. Ensemble learning can reduce the risk of choosing a learning algorithm with poor performance by combining prediction results from multiple learning algorithms. In this study, 10-year daily cycling data from eight implantable Li-ion cells is first used as the source dataset to pre-train eight deep convolutional neural network (DCNN) models. The learned parameters of the pre-trained DCNN models are then transferred from the source task to the target task, resulting in eight DCNN with transfer learning (DCNN-TL) models, respectively. These DCNN-TL models are then integrated to build an ensemble model called the DCNN with ensemble learning and transfer learning (DCNN-ETL). The effectiveness of the DCNN-ETL model is verified using a target dataset consisting of 20 commercial 18650 Li-ion cells, and the performance of the model on the target dataset is compared with that of five other data-driven methods including random forest regression, Gaussian process regression, DCNN, DCNN-TL, and DCNN-EL. The verification and comparison results demonstrate that the proposed DCNN-ETL method can produce a higher accuracy and robustness than these other data-driven methods in estimating the capacities of the Li-ion cells in the target task.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Accurate estimation of lithium-ion battery capacity is crucial for ensuring its safety and reliability. While data-driven modelling is a common approach for capacity estimation, obtaining cycling ...data during charging/discharging processes can be challenging. Collecting cycling data under various charging/discharging protocols is often unrealistic, and the collected data can be fragmented due to the random nature of working conditions in practice. To address these issues, we propose a deep transfer learning method that uses partial segments of charging/discharging data for battery capacity estimation. The proposed method utilizes capacity increment features of partial charging/discharging segments that is designed to satisfy practical scenarios. A deep transfer convolutional neural network (DTCNN) is trained with both source and target data, and a fine-tuning strategy is employed to effectively eliminate distribution discrepancies between different battery types or charging/discharging protocols, leading the improved estimation accuracy. Experimental results demonstrate that the proposed method accurately estimates the lithium-ion battery capacity, with values of RMSE, MAPE, and MD-MAPE of only 0.0220, 0.0247, and 0.0194, respectively, when using partial segments. These results highlight the promising prospects of the proposed method for lithium-ion battery capacity estimation.
•Present a novel deep transfer learning for lithium-ion battery capacity estimation.•Capacity increment features are utilized to train the deep learning model.•The effectiveness and feasibility of partial charge/discharge data are investigated.•Proposed method outperforms other state-of-the-art machine learning models.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Online battery capacity estimation is a critical task for battery management system to maintain the battery performance and cycling life in electric vehicles and grid energy storage applications. ...Convolutional Neural Networks, which have shown great potentials in battery capacity estimation, have thousands of parameters to be optimized and demand a substantial number of battery aging data for training. However, these parameters require massive memory storage while collecting a large volume of aging data is time-consuming and costly in real-world applications. To tackle these challenges, this paper proposes a novel framework incorporating the concepts of transfer learning and network pruning to build compact Convolutional Neural Network models on a relatively small dataset with improved estimation performance. First, through the transfer learning technique, the Convolutional Neural Network model pre-trained on a large battery dataset is transferred to a small dataset of the targeted battery to improve the estimation accuracy. Then a contribution-based neuron selection method is proposed to prune the transferred model using a fast recursive algorithm, which reduces the size and computational complexity of the model while maintaining its performance. The proposed model is capable of achieving fast online capacity estimation at any time, and its effectiveness is verified on a target dataset collected from four Lithium iron phosphate battery cells, and the performance is compared with other Convolutional Neural Network models. The test results confirm that the proposed model outperforms other models in terms of accuracy and computational efficiency, achieving up to 68.34% model size reduction and 80.97% computation savings.
•A convolutional neural network-based battery capacity estimation method is proposed•A framework incorporating transfer learning and network pruning is developed.•Network pruning reduces model size by 68.34% and saves computational cost by 80.97%.•Transfer learning improves model performance on small battery degradation datasets.•The model can achieve fast online capacity estimation using partial charging data.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Prognostics and health management (PHM) are developed to accurately estimate the state of health (SOH) of lithium-ion batteries, which are crucial parts for planning the employment strategy in energy ...storage systems. Numerous studies about the data-driven batteries prognostics mostly assume complete and stable charging/discharging data. The on-board prognostics with random charging/discharging behaviors remains a challenging problem. This paper proposes a novel batteries prognostics method using random segments of charging curves, aiming at improving the flexibility and applicability in practical usage. Firstly, partial incremental capacity analysis is conducted within specific voltage range. And the extracted partial incremental capacity curves are used as features for SOH estimation and prognostics. Second, a long short-term memory network guided by Bayesian optimization is proposed to automatically tune the hyper-parameters and achieve accurate SOH estimation results. The effectiveness and robustness of the partial incremental capacity features acquired from different voltage ranges are investigated to provide guidelines for users. The superiority of the proposed method is validated on lithium-ion battery aging datasets from NASA and CALCE Prognostics Data Repository. The experimental results show that it can accurately predict aging patterns and estimate SOH by solely using small segments of charging curves, showing a promising prospect.
•Propose a novel battery prognostic method with LSTM and partial IC features.•Presented partial IC features avoid the identification of specified IC curve peaks.•Bayesian optimization is adapted into LSTM to automatically tune hyper-parameters.•The effectiveness is comprehensively investigated in two battery aging datasets.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
•A smoothing method for IC curves based on Kalman filter is proposed.•Battery standard and non-standard charging experiments are designed and conducted.•The influence of battery aging and charging ...condition on IC curves is revealed.•The adaptive ICA for capacity estimation takes the charging initial SOC into account.
Incremental capacity analysis (ICA) has been widely employed to investigate the degradation mechanism and perform the capacity estimation of lithium-ion batteries. However, the traditional capacity estimation based on ICA is limited by the computational efficiency and charging condition. In this paper, a rapid acquisition method of the incremental capacity (IC) curve is established, then an adaptive capacity estimation framework based on ICA considering the charging condition is proposed. Aiming at improving the computational efficiency, the Kalman filter is employed to acquire the smooth IC curves expeditiously, which requires small data length to handle the IC value. A considerable number of battery standard and non-standard charging experiments are designed and conducted. The influence of battery aging status, charging initial state of charge (SOC), and temperature on IC curves is revealed. Three features of the IC curve during standard charging are selected to investigate the relationship between battery capacity and the height of features. Furthermore, an adaptive correction method for capacity estimation considering the charging initial SOC is established, and the validation results show the effectiveness of the proposed correction method, which provides high accuracy and robustness.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Lithium-ion batteries have become the main power source of many electronic devices. Accurate capacity estimation and cycle life prediction of lithium-ion batteries are of great significance to ensure ...the reliability of electronic devices. Extreme learning machine(ELM) is a kind of single hidden layer feedforward neural network with fast learning speed and good generalization performance. Considering the shortcomings of deep learning and the increasing size of battery datasets, this paper introduces the idea of Broad Learning(BL) and develops a new ELM model: Broad Learning-Extreme Learning Machine(BL-ELM). First, an ELM network is constructed, and the feature nodes are produced by feature mapping of the input data. Second, the enhancement operation is performed on the mapped features to produce the enhancement nodes. Next, all these two kinds of nodes are merged to become the new input layer of the network, so that the model can quickly and fully obtain effective feature information from the input data. Finally, experiments are performed with different battery datasets. The results show that BL-ELM method can not only ensure the accuracy of estimation and prediction but also save time greatly. Further comparisons with other algorithms show that this novel model is more effective and competitive.
Display omitted
•A new method of battery capacity estimation and cycle life prediction is developed.•The hybrid neural network of extreme learning machine and broad learning is applied.•This model overcomes the complexity of deep learning and is easy to implement.•NASA, CALCE and multi-dimensional MIT&Stanford battery datasets are considered.•The developed method is more competitive in time and precision.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
•Random forest regression is proposed for on-line battery capacity estimation.•The estimation is developed from partial charging voltage-capacity data.•Two features indicative of battery capacity ...fade are extracted from charging curves.•An incremental capacity analysis is used for assisting battery feature selection.
Machine-learning based methods have been widely used for battery health state monitoring. However, the existing studies require sophisticated data processing for feature extraction, thereby complicating the implementation in battery management systems. This paper proposes a machine-learning technique, random forest regression, for battery capacity estimation. The proposed technique is able to learn the dependency of the battery capacity on the features that are extracted from the charging voltage and capacity measurements. The random forest regression is solely based on signals, such as the measured current, voltage and time, that are available onboard during typical battery operation. The collected raw data can be directly fed into the trained model without any pre-processing, leading to a low computational cost. The incremental capacity analysis is employed for the feature selection. The developed method is applied and validated on lithium nickel manganese cobalt oxide batteries with different ageing patterns. Experimental results show that the proposed technique is able to evaluate the health states of different batteries under varied cycling conditions with a root-mean-square error of less than 1.3% and a low computational requirement. Therefore, the proposed method is promising for online battery capacity estimation.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
Real-time capacity estimation of lithium-ion batteries is crucial but challenging in battery management systems (BMSs). Due to the complexity of battery degradation mechanism, data-driven methods are ...prevalent recently. Despite achieved promising results, most of developed approaches still assume that the degradation trajectories of batteries are same between the training and testing domains. However, the inconsistency of batteries and the randomness during degradation process lead to the distribution discrepancy, which further affects the estimation precision of trained model. To overcome this challenge, a novel deep learning framework assisted with domain adaptation is proposed in this paper. First, a deep long short-term memory (LSTM) network is designed to capture the nonlinear mapping from monitored data, specially, terminal voltage and current, to battery capacity. Then, a domain adaptation layer is integrated to the LSTM with the purpose of degradation feature alignment between the source and target batteries. The proposed method is capable of establishing the general capacity estimation model for the discrepant batteries by only using a few cycling data of target batteries. Extensive experiments on two battery datasets from NASA Ames Prognostics Data Repository demonstrate that the proposed method outperforms the state-of-the-art data-driven methods in terms of estimation precision and robustness.
•A battery capacity estimation framework integrating domain adaptation is presented.•The LSTM is enhanced by MMD minimization for battery capacity estimation.•Battery capacity is accurately and quickly predicted with fewer training data.•The proposed method is superior to other popular data-driven methods.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Capacity estimation is essential for battery health management during the whole lifecycle. The data-driven technique has shown advanced performance in battery capacity estimation. However, the strict ...limitations on application scenarios and the long duration for feature determination are still the bottlenecks of existing data-driven estimation methods. This study proposes a data-driven capacity estimation method only using 10-min relaxation voltage data, which is adaptable to the high state of charge (SOC) range. The experiments of commercial batteries are designed to investigate the coupling relationship between relaxation voltage, battery aging, and charging cut-off SOC. Further, a novel dual Gaussian process regression (GPR) framework is put forward, in which one GPR is responsible for the battery open-circuit voltage (OCV) estimation using the sequential relaxation voltage feature, and another GPR estimates battery capacity with the corresponding relaxation voltage feature and the estimated OCV. Quantitative experimental results demonstrate that the proposed approach can achieve accurate OCV estimation with extremely sparse voltage data. When SOC is larger than 90%, the capacity estimation achieves a mean absolute error of 2.493% over the battery lifecycle, showing a noticeable improvement over the traditional estimation method.
•A battery capacity estimation approach using 10-min relaxation voltage is proposed.•The adaptive capacity estimation method can cover the high SOC range.•The sequential voltage features are constructed to realize the multi-state estimation.•The proposed approach contains dual Gaussian process regression models.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP