.png){kind=link}
The room surrogate model is a machine learning model that predicts the room temperature and the optimum cooling load, based on the weather forecast, the room occupancy and other features. The model is trained on data from energyplus simulations
This repository contains the code for the AgentBasedEnergyModelingFramework. The room surrogate model is a machine learning model that predicts the room temperature and the optimum cooling load, based on the weather forecast, the room occupancy and other features. The model is trained on data from energyplus simulations and is used to predict the office room indoor temperature. The model is based on a 1) Regression model adopted to a 'Nowcast'(copyright Thomas Hirsch). And/Or a 2) LSTM model trained on the same data to predict the room temperature.
Further more, the repository fetches in the Beginning the EnergyPlus weatherfile .epw from the given location (if --location flag used), if it is not already present. The EnergyPlus simulation is run with the eppy package and the results are ploted and used to train the machine learning model. The model is then used to predict the room temperature and optimum cooling load for the next 12-24 hours. It fetches also the weather forecast from the OpenWeatherMap API and uses it to predict the room temperature and cooling load for the next 24 hours.
- Tested on POP!OS (based on Ubuntu 22.04) with Python 3.10
- install EnergyPlus 24.1 from here https://energyplus.net/downloads
- install the required packages with
pip install eppy geopy rapidfuzz argcomplete selenium webdriver_manager mpi4py SALib plotly pyqtgraph pyqt5 - install Repast4py: https://repast.github.io/repast4py.site/index.html
- install MPI for the Agent Based Modeling Module with
sudo apt install mpich env CC=mpicxx pip install repast4py
- install MPI for the Agent Based Modeling Module with
- install the other packages with
pip install -r requirements.txt
+---------------+
| main() |
+---------------+
|
|
v
+-----------------------+
| init_argparse() |
| (parse command line |
| arguments) |
+-----------------------+
|
|
v
+-----------------------+
| LocationHandler |
| (get location from |
| CLI flag, EPW file, |
| or IDF file) |
+-----------------------+
|
|
v
+-------------------------------------------------+
| fetch_epw_file_and_save_it_and_the_location() |
| (fetch EPW file and save it) |
+-------------------------------------------------+
|
|
v
+---------------------------+
| WeatherFetcher |
| (load past and forecast |
| weather data) |
+---------------------------+
|
|
v
+-------------------------------+
| run_energyplus_simulation() |
| (run EnergyPlus simulations) |
+-------------------------------+
|
|
v
+--------------------------+
| ABMSimulationRunner |
| (run ABM simulation) |
+--------------------------+
|
|
v
+------------------------+
| plot_occupancy() |
| (plot occupancy) |
+------------------------+
|
|
v
+-----------------------+
| analyse_results() |
| (analyze results) |
+-----------------------+
|
|
v
+------------------------------------------+
| check_prediction_args_and_apply_them() |
| (check prediction arguments and |
| apply them) |
+------------------------------------------+
|
|
v
+-----------------------+
| plot_temperatures() |
| (plot temperatures) |
+-----------------------+
|
|
v
+-----------------------+
| plot_cooling_load() |
+-----------------------+
|
|
v
+------------------------------------+
| optimize_AC_energy_consumption() |
+------------------------------------+
|
|
v
+-------------------------+
| in_the_end_clean_up() |
| (clean up) |
+-------------------------+
|
|
v
+----------------------+
| print("Done.") |
+----------------------+
- The main file is the
main.pyfile. It can be run withpython main.py --location Burgosor e.g.python3 main.py -i small_office_TUGinff_variated_schedule_medium_occupancy.idf -e AUT_ST_Graz.Univ.112900_TMYx.epw - The
main.pyfile contains the main function that runs the room surrogate model. - flags can be set in the
main.pyfile to change the behavior of the model. See themain.pyfile for more information. - The
main.pyfile can be run with the following flags:-ior--idfto specify the path to the EnergyPlus IDF file-eor--epwto specify the path to the EnergyPlus EPW file-lor--locationto specify the location of the weather forecast-nor--nowcastto specify the nowcast model-lstmor--lstmto specify the LSTM model-oor--outputto specify the output directory-N,--numberto specify the number of simulations-hor--helpto specify the help of the model
- Export Documentation to html when ready for submission, with https://pdoc.dev/docs/pdoc.html
- Correct the 'Nowcast' model for energeryload optimization in the CLI Tool.
- Modul Analysis ausbauen um Ergebnisse zu analysieren
- analysis.plot_cooling_load()
- Seperate the WeatherFetcher from the ABM Module to use it independently.
- Create an API for the Results of the Simulations and Predictions
- LSTM model in LSTM_many_hours.ipynb
- latest file: eppy_multi_run_with_plots_2022_final.ipynb
- simulation results (100 x LHS) in folfder 'simulations_own_schedule_24_final_2' -- derived from final_office_IdealLoad_summer_tiny_hourly_own_schedule_24_final_2.idf
- when appending csvs, shuffle=False is important, otherwise the order of the data is not preserved.