Cylindrical Li-ion batteries (cells) typically have safety vents in the positive terminal to enable the release of gases that build up inside the battery and thus help reduce the effects of thermal runaway, including fire and explosion. However, the vents are not always effective, and it is critical to understand why. This paper overviews various vent designs and presents two case studies of vents in cylindrical Li-ion batteries that failed to operate properly. The first case study concerns a Sony VCT5 18650 Li-ion battery in which thermal runaway caused the cylindrical casing (can) to rupture. The second case study involves an Ampking 20700 Li-ion battery in which thermal runaway caused side-wall rupture of the casing. In both cases, the vents did not function as intended. Computed tomography (CT) scanning was used to investigate the internal structure of the battery to assess the failure mechanisms of the failed vents. These cases raise concerns about the efficiency, reliability, and safety of some current vent designs.

Lithium-ion (Li-ion) battery fires and explosions in various battery-operated products, have raised safety concerns. While external abuse of batteries can cause fires and explosions, most of the reported problems arose from internal battery defects, which are often difficult to detect. One particular and significant defect pertains to burrs on the tab used to connect the anode and cathode layers to the external terminals of the battery. This paper investigates burr-related issues, presents a case wherein a burr most likely caused thermal runaway in a battery, and overviews the standard associated with burr control and the application of computed tomography (CT) scanning to assess the risks.

Discriminative feature extraction is a challenge for data-driven fault diagnosis. Although deep learning algorithms can automatically learn a good set of features without manual intervention, the lack of domain knowledge greatly limits the performance improvement, especially for nonstationary and nonlinear signals. This paper develops a multiscale information fusion-based stacked sparse autoencoder fault diagnosis method. The autoencoder takes advantage of the multiscale normalized frequency spectrum information obtained by dual-tree complex wavelet transform as input. Accordingly, the multiscale normalized features guarantee the translational invariance for signal characteristics, and the stacked sparse autoencoder benefits the unsupervised feature learning and ensures accurate and stable diagnosis performance. The developed method is performed on motor bearing vibration signals and worm gearbox vibration signals, respectively. The results confirm that the developed method can accommodate changing working conditions, be free of manual feature extraction, and perform better than the existing intelligent diagnosis methods.

State of health (SOH) prediction for lithium-ion batteries (LIBs) can evaluate battery reliability and determine the aging state to discover a battery’s lifetime. Health indicators (HIs) are a powerful approach to predict battery SOH. The existing HIs SOH prediction methods for LIBs are more inclined to consider the temporal dependence of HIs. It is more conducive to SOH prediction to use data-driven methods to mine more potential interdependence between HIs in limited data. This paper proposes a novel method for determining LIB SOH using graph neural network, which takes into full consideration the HIs’ temporal feature and the correlation features between different HIs. Firstly, the Pearson’s correlation coefficients between HIs and SOH are calculated to extract the first six strongly correlated HIs subsets to build a graph. Secondly, the temporal features are extracted by a convolutional neural network and long short-term memory neural network. Finally, in order to mine the correlation among HIs, the graph sample aggregate (GraphSAGE) is employed to propagate messages on pre-defined graph structure with HIs as nodes to extract spatial features. Applications on the MIT and NASA datasets show the effectiveness of the proposed approach in predicting battery SOH.