Skip to content

Solar radiance prediction using CNN and ConvLSTM models

Notifications You must be signed in to change notification settings

harmanpreet93/solar-irradiance-prediction

Repository files navigation

Solar Irradiance - IFT6759

Team 08

To run the evaluation script:

1. cd scripts/
2. Update submit_evalution.sh 
3. sbatch submit_evalution.sh

OR

1. cd scripts/
2. Update run_evaluatior.sh 
3. Run run_evaluatior.sh

To setup a new local environment:

module load python/3.7
virtualenv ../local_env
source ../local_env/bin/activate
pip install -r requirements_local.txt

To setup a new server node environment:

module load python/3.7
virtualenv ../server_env --no-download
source ../server_env/bin/activate
pip install --no-index -r requirements.txt

OR, if no requirement.txt file is available:

pip install --no-index tensorflow-gpu==2 pandas numpy tqdm

To evaluate results from server locally using tensorboard:

Run the commands to synchronize data from the server and to launch tensorboard:

./rsync_data.sh
./run_tensorboard.sh

Use a web browser to visit: http://localhost:6006/

CNN Architecture

  • CNN followed by dense layers along with the metadata information CNN
    NN

K-Fold Strategy

  • Hold out 1 year of data (year 2015)
  • No use of k-fold until pipeline is optimized

Coding Standards

  • Lint your code as per PEP8 before submitting a pull request
  • Pull requests are required for merging to master for major changes
  • Use git branch
  • No large files allowed in git

License

Apache 2.0

About

Solar radiance prediction using CNN and ConvLSTM models

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages