Imputation

This imputation algorithm creates a dataset to train on provided OpenDSS model and then the data is used for training and testing, this can later be used for predicting imputed data.

Setup

This neural network architecture represents a sophisticated deep learning approach specifically designed for power system injection prediction tasks.!This feed-forward neural network employs a progressive compression methodology combined with comprehensive regularization techniques to achieve robust predictive performance while maintaining computational efficiency.

git clone https://github.com/pnnl/oedisi_dopf.git
cd oedisi_dopf/imputation
poetry update

Run

Prepare Training Data

The first step is to extract and develop a training dataset. Run the ./src/data_for_imputation_new.py command to prepapre the training dataset.

poetry run python src/model.py --model=ieee123 --input=opendss/

Train Model

Once the input data has been extracted, the InjPred_Train.py code is run to train the model

poetry run python src/train.py --model=ieee123

Imputation Prediction

To predict the imputed data, the Injection_Prediction.py is run.

### Evaluate 

```shell
poetry run python src/train.py --model=ieee123 --ouput=output/