A stretch of railway across China’s Inner Mongolia has become the testing ground for a breakthrough in heavy freight transport, as engineers demonstrated a record-breaking convoy of coal trains operating as a single, digitally coordinated unit, Xinhua reports.
In early December, seven heavy-haul freight trains, each carrying 5,000 tonnes of coal, moved in tightly synchronised formation along the Baotou–Shenmu line, a key corridor linking Inner Mongolia with neighbouring Shaanxi province. Together, the convoy transported 35,000 tonnes, Xinhua says - the largest load ever handled in this way, without the trains being physically coupled.
The trial replaces traditional steel couplers with a high-speed wireless control system that allows trains to operate in close formation while remaining mechanically independent. Using real-time data links, the system synchronises speed, braking and acceleration, enabling multiple trains to behave as a single entity before separating automatically at their destination.
The project, led by CHN Energy Baoshen Railway in partnership with the China Railway Signal & Communication Research & Design Institute, is designed to push beyond the long-accepted physical limits of rail freight. Conventional heavy-haul operations face constraints from mechanical stress and safety spacing, which cap the length and frequency of trains on busy routes. By contrast, the digitally linked convoy sharply reduces the required distance between trains and avoids the damaging forces generated in ultra-long, mechanically coupled formations.
Engineers estimate that the technology could lift carrying capacity on the Baotou–Shenmu line by more than 50% without laying new track or extending platforms. The railway already handles about 180mn tonnes of freight a year, much of it coal from nearby mines, making efficiency gains particularly valuable.
AI-based prediction of train-induced environmental vibration with limited measurements
KeAi Communications Co., Ltd.
image:
Framework of the transfer learning–based train-induced environmental vibration prediction method.
view moreCredit: Ruihua Liang
Rapid and accurate prediction of train-induced environmental vibration is key in railway engineering, as it directly supports route planning and the design of vibration mitigation measures. Such predictions help prevent excessive vibration from affecting nearby buildings, sensitive equipment, and residents’ comfort. Conventional rapid prediction methods, however, are mainly based on statistical or empirical models calibrated using field measurements. Hence, their performance depends strongly on the availability of sufficient data, which are often scarce, costly, and difficult to obtain.
A new study published in the Journal of Railway Science and Technology demonstrates that reliable vibration prediction can be achieved with limited measurement data. Using a transfer learning strategy, the proposed model first learns general vibration patterns from physics-based numerical simulations and is then fine-tuned using a small number of measurements to account for discrepancies between simulations and real-world responses. This improves existing rapid prediction workflows that would otherwise rely heavily on field data.
“Our work shows that physically meaningful information from numerical simulations can be effectively transferred into measurement-based machine learning models, enabling accurate predictions even when measurement data are limited,” shares Dr. Ruihua Liang, lead author of the study and a Research Fellow at the School of Civil and Environmental Engineering, Nanyang Technological University.
The main innovation of this study lies in the use of data fusion within a neural network, which integrates complementary information from physics-based simulations and field measurements. “A case study using vibration data from Beijing metro lines shows that the proposed method outperforms conventional machine-learning models trained solely on measurements, particularly under data-scarce conditions,” adds Liang. “By reducing the dependence on expensive field measurements, our method offers engineers and planners a faster and more cost-effective way to evaluate environmental vibration risks.”
###
Contact the author: Ruihua Liang (ruihua.liang@ntu.edu.sg), School of Civil and Environmental Engineering, Nanyang Technological University, Singapore.
The publisher KeAi was established by Elsevier and China Science Publishing & Media Ltd to unfold quality research globally. In 2013, our focus shifted to open access publishing. We now proudly publish more than 200 world-class, open access, English language journals, spanning all scientific disciplines. Many of these are titles we publish in partnership with prestigious societies and academic institutions, such as the National Natural Science Foundation of China (NSFC).
Article Title
Transfer learning-based scoping prediction for train induced ground vibration using both simulated and experimental data
No comments:
Post a Comment