Short Term Load Forecasting using feature engineering

This repository provides the code for the paper "A Novel Technique for Short-Term Load Forecasting Using Sequential Models and Feature Engineering" available at paper .

---

Introduction

In the paper tests were performed on five different datasets: Smart Grid Smart City (SGSC), The Almanac of Minutely Power dataset (AMPD), Réseau de Transport d’Électricité (RTE), The Electric Reliability Council of Texas (ERCOT), Pakistan Residential Electricity Consumption (PRECON).In this repository we provide the code for the derived models and derived feature generation. The code provided is generic and needs to be slightly modified to fit the format of each dataset.

Dependencies

Make sure you have Python>=3.6 installed on your machine.

pip install -q requirements.txt

Training

Training and Testing and Graph Generation funcations arfe provided in /src/utils.py, which are imported in Experiments.ipynb. The experiments notebook contains code for agressive training and testing models on a particular dataset.

Set the parameters to select differnet model types, evaluation and loss functions and features.

Example:

experiment_type = "derived"
loss_function = "mape"
windows = [12]
model_names = ['LSTM']

Runs experiment using LSTM model with MAPE loss fucntion and 12 timesteps and derived features.

Datasets

SGSC
@manual{smart_grid,
    title  = "Smart-Grid Smart-City Customer Trial Data",
    author = "",
    note   = "\url{https://data.gov.au/data/dataset/4e21dea3-9b87-4610-94c7-15a8a77907ef}",
    year   = "2019 (Accessed online:10.09.2019)"
}

RTE
@manual{rte-data,
    title  = "RTE, Grid data.",
    author = "",
    note   = "\url{https://data.rte-france.com/}",
    year   = "2019 (Accessed online:27.08.2019)"
}

ERCOT
@manual{ercot-data,
    title  = "ERCOT, Grid data.",
    author = "",
    note   = "\url{https://ercot.com/}",
    year   = "2019 (Accessed online:27.08.2019)"
} 

AMPD
@inproceedings{ampd,
    author={S. {Makonin} and F. {Popowich} and L. {Bartram} and B. {Gill} and I. V. {Bajić}},
    booktitle={2013 IEEE Electrical Power Energy Conference},
    title={{AMPds}: A public dataset for load disaggregation and eco-feedback research},
    year={2013},
    volume={},
    number={},
    pages={1-6},
    keywords={computerised monitoring;decision making;domestic appliances;energy conservation;home automation;load forecasting;natural gas technology;power system measurement;smart meters;eco-feedback research;home-based intelligent energy conservation system;home appliances;intelligent feedback;intelligent decision making;nonintrusive load monitoring;NILM;real power reading;Almanac of Minutely power dataset;submeter;AMPds;natural gas;water consumption;load disaggregation algorithm;Current measurement;Power measurement;Home appliances;Voltage measurement;Natural gas;Data acquisition;Power Meter;Current;Dataset;Load Disaggre-gation;Eco-Feedback;Single-Measurement;Maximum a Posteriori (MAP);Energy Conservation},
    doi={10.1109/EPEC.2013.6802949},
    ISSN={},
    month={Aug}
}

PRECON
@inproceedings{Nadeem:2019:PPR:3307772.3328317,
    author = {Nadeem, Ahmad and Arshad, Naveed},
    title = {{PRECON}: {Pakistan} Residential Electricity Consumption Dataset},
    booktitle = {Proceedings of the Tenth ACM International Conference on Future Energy Systems},
    series = {e-Energy '19},
    year = {2019},
    isbn = {978-1-4503-6671-7},
    url = {http://doi.acm.org/10.1145/3307772.3328317},
    doi = {10.1145/3307772.3328317},
    publisher = {ACM},
    keywords = {Consumption, Dataset, Electricity, PRECON},
}

Citation

@inproceedings{IEEE Access,
    author = {Abdul Wahab, Muhammad Anas Tahir, Naveed Iqbal, Adnan Ul-Hasan, Faisal Shafiat, Syed Muhammad Raza Kazmi},
    title = {A Novel Technique for Short-Term LoadForecasting using Sequential Modelsand Feature Engineering},
    year = {2021},
    keywords = {Load Forecasting, Smart Grids, Deep Learning, Feature Engineering, Sequential Models},
}