Skip to main content
Menu

EngineBench

EngineBench

About EngineBench

A database for machine learning research in thermal propulsion systems

Licensing

This is the EngineBench database, a collection of datasets collated by the Oxford TPSRG specially for machine learning research in thermal propulsion systems. EngineBench is comprised of Particle image Velocimetry (PIV) data from different experiments previously run on the transparent combustion chamber (TCC-III) optical engine by General Motors and the University of Michigan. The data, which were released under the Creative Commons Attribution 4.0 International License, have been formatted into HDF5 files for built-in modularity and ease of use with matrix operations. This provides flexibility in separating train/test splits by crank angle or test point.

Our license: CC BY-NC-SA 4.0

https://creativecommons.org/licenses/by-nc-sa/4.0/

Acknowledgement

Publications arising from the use of EngineBench should cite the EngineBench database including the current URL and include the following acknowledgment to the original data source

"The TCC engine work has been funded by General Motors through the General Motors University of
Michigan Automotive Cooperative Research Laboratory, Engine Systems Division"

Data

EngineBench contains data from the TCC-III CFD Input Dataset. Three test points are included, with up to four PIV planes imaged depending on the test point. For more information see https://deepblue.lib.umich.edu/handle/2027.42/108382

EngineBench_LSP_small is a subset of the full EngineBench data containing data from five crank angles (90, 135, 180, 225, 270) from the lower swirl plane (LSP), at 1300 rpm and 40 kPa.

Kaggle (full): www.kaggle.com/datasets/samueljbaker/enginebench

Kaggle (small): www.kaggle.com/datasets/samueljbaker/enginebench-lsp-small

Zenodo (link): www.kaggle.com/datasets/samueljbaker/enginebench