Search

This paper investigates the modal parameter identification technique utilizing a drop impact load. A 12-DOF mathematical model of a Railway Vehicle System (RWVS) was constructed, followed by theoretical calculations, simulation analysis, and tests with four different drop impact loads. The responses to these loads were inspected through FFT, and the modal frequencies of the RWVS were determined by the Peak Picking (P-P) method. The simulation demonstrated that four types of drop impact loads can activate the bounce, pitch, and roll modes of the RWVS. The theoretical calculation and simulation analysis revealed that the maximum error of modal frequency identification is no greater than 6.5%. The experimental results verified that the drop impact excitation method can be used to identify the modal parameters of a complex railway vehicle system, which is highly beneficial for the design and verification of the vehicle structure.

Driving fatigue impairs the driver's hand-foot coordination, a significant factor in traffic accidents. However, the functional brain connectivity underlying driver coordination remains unclear. In this study, we conducted a monotonous long-distance driving experiment (approximately 2 hours) with thirty participants, who rated their fatigue level every 6 minutes using the Karolinska sleepiness scale (KSS). We then categorized the ratings into three fatigue levels: alertness or slight fatigue (AF), moderate fatigue (MF), and severe fatigue (SF). Emergency avoidance scenarios occurred randomly during the experiment, requiring the driver to brake and steer simultaneously to prevent collisions. The multi-band brain functional connectivity network and the critical connections were utilized for global-regional brain functional connectivity analysis and explainable artificial intelligence (XAI) analyses. We found that inter-region connectivity between cognitive and motor regions in the frontal-central and frontal-parietal regions decreased as the driver's fatigue increased to MF. Simultaneously, intra-region connectivity within the central region decreased, impairing intra-motor region connectivity. These diminished the driver's coordination ability. However, profound fatigue triggered an inter-regional and intra-regional reorganization of the brain functional network, reflecting anti-fatigue characteristics but insufficient to enhance driving performance. The XAI results indicated that the beta-band brain functional connection network dataset achieved the highest accuracy (0.941) on the Resnet18 model and revealed fatigue differences in the cognitive and motor regions of the dataset. Our study demonstrated that the driver's hand-foot coordination had distinct brain functional connectivity characteristics under different fatigue states, which could inform the design of neural network detection algorithms for driving fatigue.

The driver's behavior plays a crucial role in transportation safety. It is widely acknowledged that driver vigilance is a major contributor to traffic accidents. However, the quantitative impact of driver vigilance on driving risk has yet to be fully explored. This study aims to investigate the relationship between driver vigilance and driving risk, using data recorded from 28 drivers who maintain a speed of 80 km/h on a monotonous highway for 2 hours. The k-means and linear fitting methods are used to analyze the driving risk distribution under different driver vigilance states. Additionally, this study proposes a research framework for analyzing driving risk and develops three classification models (KNN, SVM, and DNN) to recognize the driving risk status. The results show that the frequency of low-risk incidents is negatively correlated with the driver's vigilance level, whereas the frequency of moderate-risk and high-risk incidents is positively correlated with the driver's vigilance level. The DNN model performs the best, achieving an accuracy of 0.972, recall of 0.972, precision of 0.973, and f1-score of 0.972, compared to KNN and SVM. This research could serve as a valuable reference for the design of warning systems and intelligent vehicles.

In recent years, with the development of new energy vehicle industry, the development potential of hybrid electric vehicles (HEVs) is increasing. As one of the key technologies, energy management strategy (EMS) has always been a hot research area for HEVs. This paper proposed a deep Q-network (DQN) based EMS for a parallel HEV. Simulation results after training show that, compared with the EMS based on dynamic programming (DP), the DQN-based EMS can achieve 8.38% of the fuel consumption gap while the calculation time is only 12.5%. By the computational advantage of neural network, the average output time of an action in each state is 1ms, which has the potential for real-time applications. Since the final EMS is parameterised and fitted by deep neural networks (DNNs) of deep learning, it is necessary to find further methods for the actual experimental scheme instead of simulation in the future.

Real-time applications of energy management strategies (EMSs) in hybrid electric vehicles (HEVs) are the harshest requirements for researchers and engineers. Inspired by the excellent problem-solving capabilities of deep reinforcement learning (DRL), this paper proposes a real-time EMS via incorporating the DRL method and transfer learning (TL). The related EMSs are derived from and evaluated on the real-world collected driving cycle dataset from Transportation Secure Data Center (TSDC). The concrete DRL algorithm is proximal policy optimization (PPO) belonging to the policy gradient (PG) techniques. For specification, many source driving cycles are utilized for training the parameters of deep networks based on PPO. The learned parameters are transformed into the target driving cycles under the TL framework. The EMSs related to the target driving cycles are estimated and compared in different training conditions. Simulation results indicate that the presented transfer DRL-based EMS could effectively reduce the time consumption and guarantee the control performance.

Real-time applications of energy management strategies (EMSs) in hybrid electric vehicles (HEVs) are the harshest requirements for researchers and engineers. Inspired by the excellent problem-solving capabilities of deep reinforcement learning (DRL), this paper proposes a real-time EMS via incorporating the DRL method and transfer learning (TL). The related EMSs are derived from and evaluated on the real-world collected driving cycle dataset from Transportation Secure Data Center (TSDC). The concrete DRL algorithm is proximal policy optimization (PPO) belonging to the policy gradient (PG) techniques. For specification, many source driving cycles are utilized for training the parameters of deep networks based on PPO. The learned parameters are transformed into the target driving cycles under the TL framework. The EMSs related to the target driving cycles are estimated and compared in different training conditions. Simulation results indicate that the presented transfer DRL-based EMS could effectively reduce the time consumption and guarantee the control performance.