STATISTICAL ANALYSIS OF TELEMETRY DATA OF SHUNTING LOCOMOTIVES
DOI:
https://doi.org/10.34185/1991-7848.itmm.2026.01.075Keywords:
telemetry data, shunting locomotive, statistical analysis, time series, energy efficiency, cross-correlation, data filteringAbstract
The study is devoted to the statistical analysis of large arrays of telemetry data of shunting locomotives (over 4.89 million measurements) to monitor their energy efficiency. Discrete multidimensional time series including fuel level, speed, and power were analyzed. It has been established that the continuous data stream contains hardware and operational interference (1501 hardware spikes were detected), which complicates the direct application of standard algorithms. Global statistical characteristics have been calculated, indicating high variance and asymmetry of the distribution. A significant proportion of the time shunting locomotives spend in idle state (57.7–98.8%) was revealed, during which the diesel generator set continues to consume fuel. Cross-correlation analysis showed a low overall correlation between fuel level and power (0.055–0.341), which increases to 0.485 in active load modes. The necessity of mandatory preliminary filtering and segmentation of data for the correct algorithmic detection of operational anomalies is proved.
References
Navixy. Fuel level sensors: resistive, capacitive, ultrasonic methods. URL: https://www.navixy.com/blog/fuel-level-sensors/ (date of access: 20.03.2026).
Box, G. E. P., Jenkins, G. M., Reinsel, G. C. Time Series Analysis: Forecasting and Control. 5th ed. Hoboken, NJ: John Wiley & Sons, 2015. 712 p.
Ivashchev D., Herasymov V. Alhorytm sehmentatsii danykh vymiriuvan rivnia palyva na transporti. Visnyk Khmelnytskoho natsionalnoho universytetu. Tekhnichni nauky. 2025. № 2 (332). S. 60–65. DOI: 10.31891/2219-9365-2025-82-60.
