Energy storage is recognized as a key technology for enabling the transition to a low-carbon, sustainable future. Energy storage requires careful management, and capacity prediction of a lithium-ion battery (LIB) is an essential indicator in a battery management system for Electric Vehicles and Electricity Grid Management. However, present techniques for capacity prediction rely mainly on the quality of the features extracted from measured signals under strict operating conditions. To improve flexibility and accuracy, this paper introduces a new paradigm based on a multi-domain features time-frequency image (TFI) analysis and transfer deep learning algorithm, in order to extract diagnostic characteristics on the degradation inside the LIB. Continuous wavelet transform (CWT) is used to transfer the one-dimensional (1D) terminal voltage signals of the battery into 2D images (i.e., wavelet energy concentration). The generated TFIs are fed into the 2D deep learning algorithms to extract the features from the battery voltage images. The extracted features are then used to predict the capacity of the LIB. To validate the proposed technique, experimental data on LIB cells from the experimental datasets published by the Prognostics Center of Excellence (PCoE) NASA were used. The results show that the TFI analysis clearly visualised the degradation process of the battery due to its capability to extract different information on electrochemical features from the non-stationary and non-linear nature of the battery signal in both the time and frequency domains. AlexNet and VGG-16 transfer deep learning neural networks combined with stochastic gradient descent with momentum (SGDM) and adaptive data momentum (ADAM) optimization algorithms are examined to classify the obtained TFIs at different capacity values. The results reveal that the proposed scheme achieves 95.60% prediction accuracy, indicating good potential for the design of improved battery management systems.

No abstract

Lithium-ion batteries are recognised as a key technology to power electric vehicles and integrate gridconnected renewable energy resources. The economic viability of these applications is affected by the battery degradation during its lifetime. This study presents an extensive experimental degradation data for lithium-ion battery cells from three different manufactures (Sony, BYD and Samsung). The Sony and BYD cells are of LFP chemistry while the Samsung cell is of NMC. The capacity fade and resistance increase of the battery cells are quantified due to calendar and cycle aging. The charge level and the temperature are considered as the main parameters to affect calendar aging while the depth of discharge, current rate and temperature for cycle aging. It is found that the Sony and BYD cells with LFP chemistry has calendar capacity loss of nearly 5% and 8% after 30 months respectively. Moreover, the Samsung NMC cell reached 80% state of health after 3000 cycles at 35C and 75% discharge depth suggesting a better cycle life compared to the other two battery cells with the same conditions.

Accurate prediction of the remaining useful life (RUL) in Lithium-ion batteries (LiBs) is a key aspect of managing its health, in order to promote reliable and secure systems, and to reduce the need for unscheduled maintenance and costs. Recent work on RUL prediction has largely focused on refining the accuracy and reliability of the RUL prediction. The author introduces a new online RUL prediction for LiB using smooth particle filter (SPF)-based likelihood approximation method. The proposed algorithm can accurately estimate the unknown degradation model parameters and predict the degradation state by solving the optimisation problem at each iteration, rather than only taking a gradient step, that tends to lead to rapid convergence, avoids instability issues and improves predictive accuracy. From the experimental datasets published by Prognostics Centre of Excellence (PCoE) of NASA, a second order degradation model was created to explore the degradation of LiB, utilising non-linear characteristics and non-Gaussian capacity degradation. RUL prediction was tested with various predicted starting points to assess whether the amount of data and parameters' uncertainty influenced the accuracy of the prediction. Results show that the proposed prediction approach gives improved prediction accuracy and improves the convergence rate in comparison with the particle filter (PF) and other methods such as unscented particle filter (UPF). Since the maximum error of the SPF predicting approach is relatively small, RUL prediction in the best case at the prediction starting point consisting of 80 cycles is 127 cycles. The prediction relative error was approximately 0.024, and the absolute error of the proposed algorithm is around 2 cycles, which is lower than the PF (around 16 cycles). RUL prediction is close to 108 cycles and relative error is around 0.136, while the absolute error prediction is approximately 16. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

No abstract