Information-based Preprocessing

Processing of binary time series data in order to enable fruther analysis and modeling.
In our case, the data originates from a PLC of a cyber-physical system.
This repository covers:

• initial data processing
• feature filtering
• uniformizing event-based data
• cycle (periodicity) detection
• cycle cutting of the data

Run

Poetry install
Execute main.py

Datasets

Information regarding the PLC datasets are contained in the 'meta_data_i.log' files.
Set 'src/configuration/config.py/data_set' with the data name.

PLC_data_1:
Less complex PLC dataset, the states each have a dedicated feature where the positive flank
(0->1 value switch) indicates a state start.
The noise is generated by inducing random features.

PLC_data_2:
More complex PLC dataset, the states each have a unique combination of feature values indicating a state start.
This leads to less precise cycle cutting due to our logic only considering one cyclic signal.
For this reason this part of the code is still in development.
The noise is generated by inducing random values.

Citation

If you find this repository useful in your research, or you make use of our data,
please consider citing the following paper:

@misc{sai2023informationbased,
      title={Information-based Preprocessing of PLC Data for Automatic Behavior Modeling}, 
      author={Brandon K. Sai and Jonas Gram and Thomas Bauernhansl},
      year={2023},
      eprint={2307.15404},
      archivePrefix={arXiv},
      primaryClass={eess.SY}
}

Contact

If you have any questions, feel free to contact Jonas Gram through E-Mail ([email protected]) or Github issues.