As one of the critical state parameters of the battery management system, the state of charge (SOC) of lithium batteries can provide an essential reference for battery safety management, charge/discharge control, and the energy management of electric vehicles (EVs). To analyze the application of deep learning in electric vehicles’ power battery SOC estimation, this study reviewed the technical process, common public datasets, and the neural networks used, as well as the structural characteristics and advantages and disadvantages of lithium battery SOC estimation in deep learning methods. First, the specific technical processes of the deep learning method for SOC estimation were analyzed, including data collection, data preprocessing, feature engineering, model training, and model evaluation. Second, the current commonly and publicly used lithium battery dataset was summarized. Then, the input variables, data sets, errors, and advantages and disadvantages of three types of deep learning methods were obtained using the structure of the neural network used for training as the classification criterion; further, the selection of the deep learning structure for SOC estimation was discussed. Finally, the challenges and future development directions of lithium battery SOC estimation using the deep learning method were explained. Over all, this review provides insights into deep learning for EVs’ Li-ion battery SOC estimation in the future.

It is known that it is critical for train rescheduling problem to address some uncertain disturbances to keep the normal condition of railway traffic. This paper is keen on a mathematical model to reschedule high-speed trains controlled by the quasi-moving blocking signalling system impacted by multidisturbances (i.e., primary delay, speed limitation, and siding line blockage). To be specific, a mixed-integer linear programming is formulated based on an improved alternative graph theory, by the means of rerouting, reordering, retiming, and train control. In order to adjust the train speed and find the best routes for trains, the set of alternative arcs and alternative arrival/departure paths are considered in the constraints, respectively. Due to this complex NP-hard problem, a two-step algorithm with three scheduling rules based on a commercial optimizer is applied to solve the problem efficiently in a real-word case, and the efficiency, validity, and feasibility of this method are demonstrated by a series of experimental tests. Finally, the graphical timetables rescheduled are analysed in terms of free conflicts of the solution. Consequently, the proposed mathematical model enriches the existing theory about train rescheduling, and it can also assist train dispatchers to figure out disturbances efficiently.

In order to study the influence of an unsupported sleeper on the vertical bearing characteristics of heavy-haul railway ballast, a three-dimensional discrete element model (DEM) was established for a ballasted track, by removing ballast particles that come into contact with the bottom of the sleeper from the model to simulate the unsupported sleeper. Vertical bearing characteristics for ballast on different types of unsupported sleepers were studied. The results showed that an unsupported sleeper could reduce the bearing area of the ballast below the sleeper and reduce the number of ballast particles that were in contact. It could also lead to an increase in the maximum contact force between the particles, accelerating the deterioration of the particles (thus affecting the overall performance of the ballast) and reducing the vertical stiffness of the ballast. As the unsupported length and width increased, vertical stiffness gradually decreased. The vertical ballast stiffness for an unsupported sleeper was then used in a dynamic coupled vehicle/track model, and the effect of the unsupported sleeper on wheel/rail interaction was analyzed. Results showed that increasing the unsupported length and width leads to a decrease in the supporting force on the unsupported sleeper and to an increase in the supporting force on the adjacent sleepers.

In engineering practice, the diagnosis of rotating machinery faults often faces numerous challenges, including noise interference and changes in operating conditions, which pose new difficulties for deep learning methods lacking prior knowledge. In response to this issue, this paper proposes a rotating machinery fault diagnosis method (OCML) based on octave, convolutional neural networks, and MOGRIFIER LSTM. This method can simply and effectively achieve noise reduction, feature extraction, and classification. Firstly, through octave analysis, noise and redundant information can be conveniently and effectively filtered out, enhancing the signal feature representation. Secondly, the designed CNN-MOGRIFIER LSTM model can effectively capture local features and temporal dependencies in the data and has good information interaction capabilities. Experiments on the CWRU dataset and the permanent magnet synchronous motor fault dataset demonstrate that the proposed method exhibits good diagnostic performance across different fault scenarios and different operating conditions. Furthermore, compared to other rotating machinery fault diagnosis methods, the proposed OCML method performs better in terms of diagnostic accuracy and stability. These results collectively confirm the good generalization of the proposed method.

As one of the critical state parameters of the battery management system, lithium battery state of charge (SOC) can provide an essential reference for battery safety management, charge/discharge control, and energy management of electric vehicles. To analyze the application of deep learning in electric vehicle power battery SOC estimation, this study reviewed the technical process, common public datasets, and the neural networks used, structural characteristics, advantages and disadvantages of lithium battery SOC estimation in deep learning method. First, the specific technical processes of the deep learning method for SOC estimation were analyzed, including data collection, data preprocessing, feature engineering, model training, and model evaluation. Secondly, the current commonly and publicly used lithium battery dataset was summarized. Then, the input variables, data sets, errors, and advantages and disadvantages of four types of deep learning methods, were concluded using the structure of neural network used for training as the classification criterion. Finally, the challenges and future development directions of lithium battery SOC estimation in deep learning method were explained.