Multivariate time series forecasting poses an ongoing challenge across various disciplines. Time series data often exhibit diverse intra-series and inter-series correlations, contributing to intricate and interwoven dependencies that have been the focus of numerous studies. Nevertheless, a significant research gap remains in comprehending the varying inter-series correlations across different time scales among multiple time series, an area that has received limited attention in the literature. To bridge this gap, this paper introduces MSGNet, an advanced deep learning model designed to capture the varying inter-series correlations across multiple time scales using frequency domain analysis and adaptive graph convolution. By leveraging frequency domain analysis, MSGNet effectively extracts salient periodic patterns and decomposes the time series into distinct time scales. The model incorporates a self-attention mechanism to capture intra-series dependencies, while introducing an adaptive mixhop graph convolution layer to autonomously learn diverse inter-series correlations within each time scale. Extensive experiments are conducted on several real-world datasets to showcase the effectiveness of MSGNet. Furthermore, MSGNet possesses the ability to automatically learn explainable multi-scale inter-series correlations, exhibiting strong generalization capabilities even when applied to out-of-distribution samples.

Developing energy management strategies (EMSs) for different types of hybrid electric vehicles (HEVs) is a time-consuming and laborious task for automotive engineers. Experienced engineers can reduce the developing cycle by exploiting the commonalities between different types of HEV EMSs. Aiming at improving the efficiency of HEV EMSs development automatically, this paper proposes a transfer learning based method to achieve the cross-type knowledge transfer between deep reinforcement learning (DRL) based EMSs. Specifically, knowledge transfer among four significantly different types of HEVs is studied. We first use massive driving cycles to train a DRL-based EMS for Prius. Then the parameters of its deep neural networks, wherein the common knowledge of energy management is captured, are transferred into EMSs of a powersplit bus, a series vehicle and a series-parallel bus. Finally, the parameters of 3 different HEV EMSs are fine-tuned in a small dataset. Simulation results indicate that, by incorporating transfer learning (TL) into DRL-based EMS for HEVs, an average 70% gap from the baseline in respect of convergence efficiency has been achieved. Our study also shows that TL can transfer knowledge between two HEVs that have significantly different structures. Overall, TL is conducive to boost the development process for HEV EMS.

Modern automotive systems have been equipped with a highly increasing number of onboard computer vision hardware and software, which are considered to be beneficial for achieving eco-driving. This article combines computer vision and deep reinforcement learning (DRL) to improve the fuel economy of hybrid electric vehicles. The proposed method is capable of autonomously learning the optimal control policy from visual inputs. The state-of-the-art convolutional neural networks-based object detection method is utilized to extract available visual information from onboard cameras. The detected visual information is used as a state input for a continuous DRL model to output energy management strategies. To evaluate the proposed method, we construct 100 km real city and highway driving cycles, in which visual information is incorporated. The results show that the DRL-based system with visual information consumes 4.3-8.8% less fuel compared with the one without visual information, and the proposed method achieves 96.5% fuel economy of the global optimum-dynamic programming.

Train driver fatigue and distraction are the main reasons for railway accidents. One of the new technologies to monitor drivers is by using the EEG signals, which provides vital signs monitoring of fatigue and distraction. However, monitoring systems involving full-head scalp EEG are time-consuming and uncomfortable for the driver. The aim of this study was to evaluate the suitability of recently introduced forehead EEG for train driver fatigue and distraction detection. We first constructed a unique dataset with experienced train drivers driving in a simulated train driving environment. The EEG signals were collected from an EEG recording device placed on the driver’s forehead, and numerous features including energy, entropy, rhythmic energy ratio and frontal asymmetry ratio were extracted from the EEG signals. Therefore, a time-series ensemble learning method was proposed to perform fatigue and distraction detection based on the extracted feature. The proposed method outperforms other popular machine learning algorithms including Support Vector Machine(SVM), K-Nearest Neighbor(KNN), Decision Tree(DT), and Long short-term memory(LSTM). The proposed method is stable and convenient to meet the real-time requirement of train driver monitoring.

Accurate and interpretable delay predictions are vital for decision-making in the aviation industry. However, effectively incorporating spatiotemporal dependencies and external factors related to delay propagation remains a challenge. To address this challenge, we propose the SpatioTemporal Propagation Network (STPN), a novel space-time separable graph convolutional network that models delay propagation by considering both spatial and temporal factors. STPN uses a multi-graph convolution model that considers both geographic proximity and airline schedules from a spatial perspective, while employing a multi-head self-attention mechanism that can be learned end-to-end and explicitly accounts for various types of temporal dependencies in delay time series from a temporal perspective. Experiments on two real-world delay datasets show that STPN outperforms state-of-the-art methods for multi-step ahead arrival and departure delay prediction in large-scale airport networks. Additionally, the counterfactuals generated by STPN provide evidence of its ability to learn explainable delay propagation patterns. Comprehensive experiments also demonstrate that STPN sets a robust benchmark for general spatiotemporal forecasting. The code for STPN is available at https://github.com/Kaimaoge/STPN .

This proof‐of‐concept study couples machine learning and physical modeling paradigms to develop a computationally efficient simulator‐emulator framework for generating super‐resolution (<250 m) urban climate information, that is required by many sectors. To this end, a regional climate model/simulator is applied over the city of Montreal, for the summers of 2019 and 2020, at 2.5 km (LR) and 250 m (HR) resolutions, which are used to train and validate the proposed super‐resolution deep learning (DL) model/emulator. The DL model uses an efficient sub‐pixel convolution layer to generate HR information from LR data, with adversarial training applied to improve physical consistency. The DL model reduces temperature errors significantly over urbanized areas present in the LR simulation, while also demonstrating considerable skill in capturing the magnitude and location of heat stress indicators. These results portray the value of the innovative simulator‐emulator framework, that can be extended to other seasons/periods, variables and regions.