Automation Toolbox for Machine learning in water Networks

The Automation Toolbox for Machine learning in water Networks (atmn) offers a selection of easy to use tools for generating and working with synthetic water network data. It builds on the wntr python package to use the EPANET simulator for simulation of leaks and sensor faults in water networks.

If you want a hands-on tour of all features this toolbox has to offer, have a look in the examples/Quickstart.ipynb notebook. To get started you only need to install and run jupyter notebook:

• $ pip install notebook
• $ jupyter notebook
• Open the Quickstart notebook and it will explain everything you need to know.

For implementing atmn, we used the code published for the BattLeDIM 2020 challenge (http://www.ccwi-wdsa2020.com/) as a starting point, please refer to https://github.com/KIOS-Research/BattLeDIM. Besides, to realize the simulation of sensor faults, we adapted the Matlab implementation by https://github.com/eldemet/sensorfaultmodels. We would like to thank our colleagues at KIOS Center of Excellence, University of Cyprus, Cyprus, in particular Stelios G. Vrachimis and Demetrios G. Eliades for providing crucial domain knowledge for developing the atmn package.

Documentation

There is separate documentation on different topics concerning the toolbox:

• If you want to learn more about your options for configuring a Scenario Collection, have a look at the Scenario Configuration page.

• If you seek documentation for loading Scenarios, refer to the Scenario Loader Reference

• If you are interested, how atmn organizes the generated data, have a look at the Folder Structure documentation.

• atmn currently offers three tools:

  • atmn-generate to generate a dataset from a Scenario Configuration.
  • atmn-visualize to visualize either a water network file *.inp or a specific Configuration from a Collection.
  • atmn-export to export a specific Configuration from a Collection to Excel.

  Use the -h flag to get more information on how to use these tools. For example usages, you can have a look in the Quickstart notebook. If your package manager did not create the atmn wrappers, you can also use python -m atmn to use the tools.

Version History

1.1

• Added support for saving measurements with binary datatypes

1.0

• Initial public release

Contributors

Jonas Vaquet, Kathrin Lammers, André Artelt, Fabian Hinder, Valerie Vaquet

Bielefeld University, Germany