Welcome to the Documentation of HAI-CPS

The Hamburg AI Benchmark for Cyber-Physical Systems (HAI-CPS) is a comprehensive dataset designed for evaluating AI models in the domains of anomaly detection, diagnosis, and reconfiguration for Cyber-Physical Systems (CPS).

HAI-CPS is structured as a benchmark comprising ten different scenarios of a modular process plant, each demonstrating various functionalities and increasing complexities. This enables you to comprehensively test your algorithms not only on a single use-case but systematically across increasingly complex examples within the same domain. Each scenario can feature multiple anomalies occurring within a single module or across multiple modules of the CPS.

banner

HAI-CPS includes:

  • OpenModelica simulation models

  • Pre-simulated datasets for benchmarking

  • Docker integration for easy execution

In addition to the provided setups, you can also create and simulate your own system constellations using the OpenModelica models and the HAI-CPS Python API.

Note

HAI-CPS extends the Benchmark for Diagnosis, Reconfiguration, and Planning (BeRfiPl). You can access the previous version here: BeRfiPl Benchmark

Contents:

When using HAI-CPS, please cite

@misc{Ehrhardt2025,
author = {Ehrhardt, Jonas and Moddemann, Lukas and Niggemann, Oliver},
year = {2025},
title = {The Hamburg Artificial Intelligence Benchmark for Cyber-Pyhsical Sytems - HAI-CPS},
howpublished = {\url=https://github.com/j-ehrhardt/hai-cps-benchmark},
doi = {},
}

HAI-CPS was developed at the Chair of Informatics in Mechanical Engineering at Helmut-Schmidt-Universtiy.

hsu imb