Smart control of HVAC systems using data-driven methods.

Data documentation

Source repository

GitHub mirror

This repository contains code for analysis and control of various HVAC systems at Engineering Science Building. It contains:

1. Controller scripts which are run in production.

2. Jupyter notebooks used for data analysis and machine learning.

3. Python packages for data-preprocessing and defining various functions for the above 2.

This code runs in a virtual environment. Environments are created using Miniconda for python 3. The following steps assume that the miniconda executable conda is on the system PATH.

This repository can be downloaded via git clone or directly as a zip file here.

Refer to the controller documentation.

For development, install the environment described in dev.yml:

conda env create -f dev.yml
conda activate esb

There is a minority of code that may depend on pyTorch. Depending on GPU availability, install the appropriate version.