Dataset of 101 Stochastically Generated Driving Cycles for Powertrain Robustness Evaluation
Year: 2026
Type: dataset
Abstract: This dataset provides 101 stochastically generated velocity time-series profiles (driving cycles) specifically designed for the robustness evaluation of Fuel Cell Hybrid Electric Vehicles (FCHEVs) and their Energy Management Systems (EMS). Standard certification cycles are deterministic and often fail to capture the high-variance dynamics of real-world driving. To address this, these synthetic profiles were generated using a discrete-time Markov chain stochastic approach trained on established driving databases (NEDC and UDDS). A key feature of this dataset is the integration of a continuous aggressiveness parameter (PSI) ranging from 0.00 (mild, eco-driving) to 1.00 (highly aggressive, worst-case scenarios). The 101 individual profiles systematically sweep this PSI spectrum with a step of 0.01, maximizing acceleration variance to provide a controlled environment for sensitivity analysis. This data is highly valuable for researchers and automotive engineers aiming to test predictive control algorithms, and assess fuel cell or battery degradation under unpredictable, transient stress conditions. Dataset Contents: Data_Files (Folder): 101 MATLAB data files (.mat). Each file contains a 1800-second velocity profile sampled at 1 Hz, ready to be imported as a timeseries object. Summary_Table.xlsx: A comprehensive metadata spreadsheet cross-referencing all cycles and detailing their specific PSI level, maximum velocity, and acceleration variance score. Visualize_Cycle.m: A lightweight MATLAB script provided for quick plotting and visualization of the data. ReadMe.txt: Detailed instructions for loading the data into the MATLAB workspace and integrating it with Simulink.
Source: Mendeley Data
DOI: 10.17632/knpktpvytp
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204146146