WEC Power AI/ML

Anthony Truelove MASc, P.Eng.
email:   gears1763@tutanota.com
github:  gears1763-2

See license terms

This is a project which experiments with applying artificial intelligence and machine learning (AI/ML) to the modelling of a point absorber wave energy converter (WEC).

---

Contents

In the directory for this project, you should find this README, a LICENSE file, a main report main.pdf, and the following sub-directories:

animations/     to hold some WEC animations

Maple/          to hold supporting Maple worksheets
    mw/         to hold the supporting Maple worksheets in native `.mw` format
    pdf/        to hold the supporting Maple worksheets in `.pdf` format

ProteusDS/      to hold an example Proteus DS simulation (used in generating WEC performance data)

Python/         to hold supporting Python scripts and C extensions
    data/       to hold the data table, `numpy` binaries, etc.
    pyx/        to hold supporting Cython files, etc.

---

Key Features

• Reduced dynamics for a heave constrained point absorber are solved exactly (with respect to expected power output).
• Time domain modelling software is used to generate a data set of more complex dynamics.
• A perturbation machine (dense neural network) is trained to correct from the reduced to the complex dynamics.
• Example applications to WEC design optimization and WEC performance mapping are presented.

---

Setup

If you run into a wave_utils or wave_energy_converter module not found error, chances are you need to build the C extensions for these. See below.

Note that the provided setup.py has the following dependencies (listed in pipreqs.txt):

Cython==3.0.4
setuptools==59.6.0

so you may need to pip install these before building the C extensions. Additionally, you need to make sure that you are set up with a compatible C compiler.

However, if you are all set up, then building the bindings (from Python/) is as simple as

python(3) setup.py build_ext --inplace

depending on your setup (either python or python3).

--- Linux (Debian/Ubuntu) Notes ---

You should already be set up with a Python 3 interpreter, so just doing the pip install steps should be sufficient. Additionally, you may also be set up with a compatible C compiler already. However, if this is not the case, then you can simply

sudo apt-get install build-essential

--- Windows Notes ---

On Windows, there are a few things to set up first before building the C extensions; namely

• You need to be using Python for Windows, as available from https://www.python.org/downloads/. Version 3.10+ should do fine.
• You need to pip install the dependencies listed above (and in pipreqs.txt).
• You need to be using the MSVC compiler (cl.exe). This can be obtained by way of installing Visual Studio (as available from https://visualstudio.microsoft.com/downloads/; the free community edition is fine) and then selecting the "Desktop development with C++" workload.

--- Testing Notes ---

To test that the C extensions have been built and are functioning correctly, a testing script is provided in test.py, so simply

python(3) test.py

---