MOO_TREES

This repository contains scripts for the multi-objective extension of ENTMOOT featured in: .

Please cite this work as:

@article{thebelt2021mootrees,
  title={{Multi-objective constrained optimization for energy applications via tree ensembles}},
  author={Thebelt, Alexander and Tsay, Calvin and Lee, Robert M and Sudermann-Merx, Nathan and Walz, David and Tranter, Tom and Misener, Ruth},
  journal={Applied Energy},
  volume={306},
  pages={118061},
  year={2022},
  publisher={Elsevier}
}

Dependencies

• python >= 3.7.4
• numpy >= 1.20.3
• scipy >= 1.6.3
• gurobipy >= 9.1.2
• pyaml >= 20.4.0
• scikit-learn >= 0.24.2
• lightgbm >= 3.2.1
• pybamm >= 0.4.0

For PyBaMM please install this branch https://github.com/pybamm-team/PyBaMM/tree/issue-1575-discharged_energy, which allows direct access to the discarged_energy variable. The following command will install the right branch:

pip install git+https://github.com/pybamm-team/[email protected]_energy

Installing Gurobi

The solver software Gurobi is required to run the examples. Gurobi is a commercial mathematical optimization solver and free of charge for academic research. It is available on Linux, Windows and Mac OS.

Please follow the instructions to obtain a free academic license. Once Gurobi is installed on your system, follow the steps to setup the Python interface gurobipy.

Running Experiments

This repo includes the two benchmark problems: (i) windfarm layout optimization which was adapted from here, and (ii) battery optimization which uses PyBaMM to simulate different configurations.

To run experiments please first execute create_init to generate all initial points for 25 different random seeds for both benchmarks which will be stored in moo_results/bb_init.json. A directory moo_results will be created if it doesn't exist already.

Afterwards, you can call main.py to run experiments:

e.g. python main.py Windfarm 101 10 runs the windfarm benchmark for random seed 101 and evaluation budget 10.

Authors

• Alexander Thebelt (ThebTron) - Imperial College London
• Calvin Tsay (tsaycal) - Imperial College London
• Robert M. Lee - BASF SE
• Nathan Sudermann-Merx (spiralulam) - Cooperative State University Mannheim
• David Walz (DavidWalz) - BASF SE
• Tom Tranter (TomTranter) - Electrochemical Innovation Lab UCL
• Ruth Misener (rmisener) - Imperial College London

License

This repository is released under the BSD 3-Clause License. Please refer to the LICENSE file for details.

Acknowledgements

This work was supported by BASF SE, Ludwigshafen am Rhein, EPSRC Research Fellowships to RM (EP/P016871/1) and CT (EP/T001577/1), and an Imperial College Research Fellowship to CT. TT acknowledges funding from the EPSRC Faraday Institution Multiscale Modelling Project (EP/S003053/1, FIRG003).