High-speed Train Braking System Fault Detection Dataset
Dataset description: This dataset is derived from a high-speed train braking system fault detection task based on project cooperation with Alstom. The braking system is a crucial mission subsystem for the safe operation of High-Speed Train (HST). The dataset contains monitored variables related to train-level operating conditions, such as speed, operation mode, external power supply, line voltage, and line current, as well as braking system-level conditions, such as internal temperature, battery voltage, detected slip or slide, ED brake state, TCL brake state, and achieved brake effort. After data transformation, the dataset contains 46 numeric features, including 31 binary features and 15 continuous features. The processed HST dataset contains 22,368 samples, including 389 faulty samples, and the imbalance ratio is 56.50:1. Due to commercial agreement constraints, all data were linearly transformed and feature names were desensitized.
Publications making use of this dataset are requested to cite the following paper.
M. Qian and Y. F. Li, "A Novel Adaptive Undersampling Framework for Class-Imbalance Fault Detection," IEEE Transactions on Reliability, vol. 72, no. 3, pp. 1003-1017, 2023.
The dataset can be downloaded by the following link:
https://cloud.tsinghua.edu.cn/d/c8506f091a9c47c2903c/?p=%2FHigh-speed%20train%20data&mode=list
Multivariate RSRP Distribution-valued Time Series Dataset
Dataset description: This dataset comes from a telecommunication network quality-of-service scenario. The data consist of hourly quantiles, computed at 400 points evenly spaced from 0.01 to 0.99, of Reference Signal Received Power (RSRP) records collected from 68 base stations in Hong Kong, with 673 hourly observations for each base station. Each log entry includes the base-station ID, timestamp, end-user ID, and RSRP value. For each base station and each hour, within-hour RSRP measurements are aggregated into an empirical distribution, yielding a multivariate distribution-valued time series across the 68 base stations.
Publications making use of this dataset are requested to cite the following papers.
Z. Li, Y. F. Li, and W. Zhao, "Forecasting Multivariate Distribution-valued Time Series via Penalized Local Distribution Regression," submitted.
Y. F. Li, W. Zhao, C. Zhang, J. Ye, and H. He, "A study on the prediction of service reliability of wireless telecommunication system via distribution regression," Reliability Engineering & System Safety, vol. 250, 110291, 2024.
The dataset can be downloaded by the following link:
https://cloud.tsinghua.edu.cn/d/c8506f091a9c47c2903c/?p=%2FRSRP%20quantile%20data&mode=list
Autonomous Driving Assistance System Performance Testing Dataset
Dataset description: This dataset comes from an autonomous driving assistance system (ADAS) performance testing scenario. The experiment studies how environmental factors affect ADAS measurement performance. The scalar environmental covariates include temperature, humidity, and vibration, and their interaction effects are also considered. The response variable is the lane width error, defined as the difference between the lane width recorded by the ADAS and the actual lane width. The experiment was conducted using 9 identical OnePlus mobile phones as test devices, and 100 time points were collected for each device. Temperature, humidity, and vibration were used to construct environmental settings, and the resulting data form functional response curves for evaluating nonlinear covariate effects and device-specific random effects.
Publications making use of this dataset are requested to cite the following papers.
W. Zhao, Y. F. Li, and W. Zheng, "A nonlinear mixed-effects functional regression model based on variable selection," Journal of Quality Technology, vol. 57, no. 3, pp. 220-235, 2025.
L. Wang, Y. F. Li, and Y. Xue, "Dynamic Assessment of Mission Reliability for Autonomous Vehicles Considering Mission Criticality and Environmental Dependence," IEEE Transactions on Industrial Informatics, vol. 22, no. 2, pp. 1050-1061, 2026.
The dataset can be downloaded by the following link:
https://cloud.tsinghua.edu.cn/d/c8506f091a9c47c2903c/?p=%2Fadas%20data&mode=list