yyystg / hem-deeprl-v2-1

Home Energy Management based on Deep Reinforcement Learning Approach.

Home Energy Management Model.

This file provides instructions to run the model.

Description

This model is an implementation of Deep Q learning for the purpose of Home Energy Management.

Requirements

Python

You need Python to run the model.

In Ubuntu, Mint and Debian you can install Python 3 like this:

$ sudo apt-get install python3 python3-pip

Dependencies

Some of the dependencies required to run the code, and some info about the environment (e.g., reward, actions, etc.).

• tensorflow (>= 2.3)
• numpy
• Pandas
• matplotlib

Use the package manager pip to install them.

C:\Project_PATH>pip install tensorflow
C:\Project_PATH>pip install numpy
C:\Project_PATH>pip install Pandas
C:\Project_PATH>pip install matplotlib

Usage

There are 3 simple command line interface to run the code for training, testing and ploting:

- Train a new model (this will save the model only at the end of the training):

C:\Project_PATH>python main.py -train_model <number_of_Episodes>

note that 300 Episodes should be enough.

- Test a trained model (I've already provided a trained model in the "trained_models" folder)

C:\Project_PATH>python main.py -test_model

- Data plotting of the trained model:

You can plot one of the graphs: "money_spent", "battery_charge", "reward_function"

1- Output ( Money spent):

C:\Project_PATH>python main.py -plot_graph money_spent

2- Output ( Battery charge):

C:\Project_PATH>python main.py -plot_graph battery_charge

3- Output ( Reward function):

C:\Project_PATH>python main.py -plot_graph reward_function

4- Output ( Demans):

C:\Project_PATH>python main.py -plot_graph demands

5- Output ( Generation):

C:\Project_PATH>python main.py -plot_graph generation

License