Fault diagnosis of high-speed train (HST) bogie is essential in guaranteeing the normal daily operation of HST. In prior works, feature extraction from multi-sensor vibration signals mainly relies on signal processing methods, which is independent of the classification process. Based on convolutional neural networks (CNNs), this paper presents a novel fault diagnosis system using the residual-squeeze net (RSNet), which is directly applicable to raw data (time sequences) and does not require any signal transformation or post-processing. In this network, information fusion is achieved by using the convolutional layer. More specifically, via the squeeze operation, an optimal combination of channels is learnt by training the network. Experimental results obtained by using SIMPACK simulation data demonstrate the effectiveness of the proposed approach in both complete failure case and single failure case, with diagnosis accuracy near 100%. The proposed approach also shows good performance in identifying the locations of faulty components. Comparisons between RSNet and competitive methods shows the advantages of RSNet for fault classification.

Intelligent fault diagnosis of bogie is fundamental for the reliability, stability and security of high-speed train (HST). However, thorny issues, including complicated structure of bogie and complex motion states of HST, aggravate the difficulty of research, where traditional approaches for fault diagnosis might fail. In this paper, CapsNet, a contemporary novel neural network architecture, in which a single unit called capsule is unique for its multi-dimension and abundant spatial information, is adopted to accomplish the recognition and classification of 7 working conditions of HST bogie, comprising both single and compound faults. The experimental accuracy achieves 96.65%, which proves the efficiency and potency of CapsNet in this term. The ability of CapsNet to diagnose faults of bogie in this paper exceeds that of Convolutional Neural Network (CNN), especially for compound ones. Moreover, the method employed is capable of extracting features from raw data automatically and independent of expert experience or knowledge about signal processing.

Realizing the accurate fault diagnosis of high-speed train (HST) bogie is of great significance for ensuring the safe operation of HSTs. This article proposes a novel fault diagnosis method to identify the fault states of HST bogie and localize the positions of faulty components simultaneously by virtue of the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and 1-D convolutional neural network (1-D CNN). First, the raw vibration signal is decomposed into multiple intrinsic mode functions (IMFs) via the ICEEMDAN. Then, the high-frequency components of IMFs obtained by ICEEMDAN are selected as the fault features of the HST bogie. Afterward, the 1-D CNN model is adopted to learn the deeper features from the high-frequency components to conduct fault classification and localization. The prediction accuracy of the proposed method is 99.3% and 98.7% on two bogie data sets, respectively. Meanwhile, experimental results demonstrate the effectiveness of the proposed method in identifying both the categories and the locations of faulty components, whose superiority has been verified by the comprehensive comparison analysis with traditional deep learning methods and state-of-art methods, including residual-squeeze net, convolutional recurrent neural network, and DenseNet.

As an important part of high-speed train (HST), the mechanical performance of bogies imposes a direct impact on the safety and reliability of HST. It is a fact that, regardless of the potential mechanical performance degradation status, most existing fault diagnosis methods focus only on the identification of bogie fault types. However, for application scenarios such as auxiliary maintenance, identifying the performance degradation of bogie is critical in determining a particular maintenance strategy. In this article, by considering the intrinsic link between fault type and performance degradation of bogie, a novel multiple convolutional recurrent neural network (M-CRNN) that consists of two CRNN frameworks is proposed for simultaneous diagnosis of fault type and performance degradation state. Specifically, the CRNN framework 1 is designed to detect the fault types of the bogie. Meanwhile, CRNN framework 2, which is formed by CRNN Framework 1 and an RNN module, is adopted to further extract the features of fault performance degradation. It is worth highlighting that M-CRNN extends the structure of traditional neural networks and makes full use of the temporal correlation of performance degradation and model fault types. The effectiveness of the proposed M-CRNN algorithm is tested via the HST model CRH380A at different running speeds, including 160, 200, and 220 km/h. The overall accuracy of M-CRNN, i.e., the product of the accuracies for identifying the fault types and evaluating the fault performance degradation, is beyond 94.6% in all cases. This clearly demonstrates the potential applicability of the proposed method for multiple fault diagnosis tasks of HST bogie system.

This article proposes a new spatial iterative learning control (ILC) algorithm that can track the desired trajectory for operation control of high-speed train (HST) in the presence of time-varying adhesion dynamics between wheel and track. Although the adhesion dynamics will affect the overall performance and might lead to an unsafe scenario, not much work has been done to handle them in the field of operation control of HST. As the operation process of HST is repetitive in spatial domain, the conventional ILC algorithms, in which finite time domain is always considered, cannot be directly applied to such kind of system. In order to address spatial learning, the train’s operation dynamics, including the adhesion dynamics, are first developed. By revealing the link between the temporal gradient and the spatial gradient, the operation dynamics can be converted from the time domain into the spatial domain, making the ILC design feasible. A novel ILC algorithm is then proposed to ensure the convergence with the help of a new composite energy function (CEF). It is highlighted that in this CEF analysis, dynamic systems are not required to be globally Lipschitz, which is one of the standing assumptions needed in the convergence analysis of ILC. The effectiveness of the proposed spatial ILC algorithm is demonstrated from a simulation example.

In this article, an adaptive iterative learning control scheme is presented for a class of nonlinear parametric strict-feedback systems with unknown state delays, aiming to achieve the point-wise tracking of desired trajectory in a finite interval. The appropriate Lyapunov-Krasovskii functions are established to compensate the influence of time-delay uncertainties on the control systems. As the main features, the proposed approach integrates the command filter into the backstepping procedure to avoid the differential explosion problem that may occur with the increase of system order, and introduces the hyperbolic tangent functions into the learning controller to handle the singularity problem thus maintaining the continuity of input signal. The results of theoretical analysis and numerical simulation demonstrate that the tracking errors at the entire period will converge to a compact set along the iteration axis. Compared with the existing works, the proposed control scheme is promising to manifest the better performance and practicability owing to the learning mechanism, the dynamic model, as well as the implementation of controller.