Skip to content

Improving HVAC Control with Transfer Learning: Using Padding Techniques for Cross-Building Pre-training and Fine-tuning

License

Notifications You must be signed in to change notification settings

kad99kev/PTCBPF

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PTCBPF

Improving HVAC Control with Transfer Learning: Using Padding Techniques for Cross-Building Pre-training and Fine-tuning

Link to Research - https://www.sciencedirect.com/science/article/pii/S2666546825000631

Setup

This repository runs in a Docker container configured by Sinergym.

Follow the instructions on how to install Sinergym via Docker and then follow the steps below.

Installation

In a conda or virtual environment, run the following code.

git clone https://github.com/kad99kev/PTCBPF.git
pip install -e .

Running an experiment.

Once the Docker container is built, there are different options available:

  1. controller - Will run an experiment using a rule-based controller agent.
  2. pretrain - Will train an agent with imitation learning.
  3. scratch - Will train a Deep RL agent from scratch (no fine-tuning).
  4. finetune - Will finetune a Deep RL agent using pre-trained weights.
  5. test- Will test any agent (trained via imitate, scratch or finetune).

The commands can be run as follows:

ptcbpf scratch --algo ppo --run_name scratch_experiment --experiment scratch_experiment -env Eplus-5zone-hot-continuous-stochastic-v1 -c /home/PTCBPF/config.yaml -s 0

Run ptcbpf --help for more information.

Example run scripts are given in run.sh.

Example configuration file is given in config.yaml.

Dataset for Behavioural Cloning.

The dataset for the pre-training (behavioural cloning) stage can be found from HVACIRL.

About

Improving HVAC Control with Transfer Learning: Using Padding Techniques for Cross-Building Pre-training and Fine-tuning

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published