Damage identification of axle-box bearings is essential to ensure the safe operation of railway trains. The envelope spectrums generated by spectral coherence are effective bearing damage identification tools, but the traditional spectral coherence-based envelope spectrums cannot effectively reveal the bearing damage features under strong interference noise or cannot fully extract the damage information distributed in multiple spectral frequency bands. To solve these problems, a weighted combined envelope spectrum (WCES) based on spectral coherence is proposed as an enhanced bearing damage detector in this paper. First, the frequency domain signal-to-noise ratio (FDSNR) is devised to measure the damage information in each spectral frequency component of spectral coherence. Then, an information threshold is introduced into the estimated FDSNR to construct a weighting function to enhance the informative spectral frequency components and eliminate the interference components. Eventually, the spectral coherence normalized by the weighting function is integrated to generate WCES for bearing damage identification. The simulation and experimental results indicate that the proposed method can effectively excavate the fault-related information and detect railway axle-box bearing damages, and the comparisons with the state-of-the-art methods demonstrate the superiority of the proposed method.

Due to the rapid development of high-speed railways, there is an extensive need for the condition monitoring of the stability of vehicles. Considering the weakness of the current alarm standards, a new index is introduced for describing the periodicity of the state variable series in the nonlinear dynamics system, and a new guideline – based on the proportion distribution of the index – is proposed for detecting the hunting and for evaluating the stability performance in both the space and time domains. Then, the selection of algorithm parameters is conducted and the generalization ability is validated; a number of simulations and tests are applied to further verify its effectiveness. The advantages of this new method, such as an accurate alarm for hunting motions, evaluation of vehicle dynamics, robustness and feasibility of the real time application, are highlighted. This method can further contribute to the prognostic and health management of railway vehicles through early warning and condition assessment.

Bearings are key components of modern mechanical equipment. To address the issue that the limited information contained in the single-source signal of the bearing leads to the limited accuracy of the single-source fault diagnosis method, a multi-sensor fusion fault diagnosis method is proposed to improve the reliability of bearing fault diagnosis. Firstly, the feature extraction process of the convolutional neural network (CNN) is improved based on the theory of variational Bayesian inference, which forms the variational Bayesian inference convolutional neural network (VBICNN). VBICNN is used to obtain preliminary diagnosis results of single-channel signals. Secondly, considering the redundancy of information contained in multi-channel signals, a voting strategy is used to fuse the preliminary diagnosis results of the single-channel model to obtain the final results. Finally, the proposed method is evaluated by an experimental dataset of the axlebox bearing of a high-speed train. The results show that the average diagnosis accuracy of the proposed method can reach more than 99% and has favorable stability.

Railway vehicle dampers undergo degradation and failure after long-term service. Bench tests in this paper indicate a stochastic failure process, in which damping characteristics follow bimodal random distributions between normal and failure; failure probability increases in the ageing process until it becomes deterministic. This failure mode may be neglected due to the probabilistic absence of failure, which poses great risk to reliability. Then, nonlinear suspension dynamics with stochastic failure is modelled; Itô stochastic differential rule and cumulant-neglect closure are adopted to derive moment equations. The scheme of progressive iteration is proposed to obtain fast-convergent and stable solutions for analysing the effect of stochastic failure process on vibration response. Further, numerical simulation of a vehicle with stochastic failure is performed for failure effects analysis on dynamical performance. The results demonstrate that the deterioration of failure probability and damping reduction amplitude leads to larger vibrations. The stochastic failure of secondary lateral damper and anti-yaw damper is harmful for lateral vehicle dynamics, while that of secondary vertical damper has negative effects on vertical ride quality. It is of great importance to pay full attention to the stochastic failure process of dampers for reliability and security of railway vehicles.

Plenty of studies focused on single-direction active suspensions on railway vehicles; at least four actuators per vehicle are needed forconventional active secondary suspensions in both lateral and vertical directions, leading to a sharp increase in costs and reduction inreliability. This paper proposes a novel configuration, lateral-vertical coupled (LVC) active secondary suspension, with only two actuators and two accelerometers per vehicle. To deal with the coupled dynamics, linear quadratic Gaussian (LQG) method and optimal skyhook damping (OSD) method are developed for achieving optimal control. OSD method has similar advantages as LQG method and is easier to implement. Then, linearised vehicle dynamics is modelled to design the controllers and observer, and a more complete nonlinear simulation model is established for verification. Theoretical analysis and simulation results show that LVC active suspension reduces car body vibration in one direction at the cost of negative effects in the other. LQG control fails to provide satisfying performance compared with passive suspensions, whereas OSD control can effectively improve lateral and vertical vibration and ride comfort by applying reasonable actuator forces. Besides, actuator dynamics is important to the effectiveness of LVC active suspension, and OSD controller is robust to measurement noise and model errors to some extent.