Electricity Demand Regression with Machine Learning

Table of Contents

• Overview
• Requirements
• Setup
• File Descriptions

Overview

The goal of this project is to predict hourly, balancing authority-scale demand from local meteorological conditions (i.e. hourly temperature).

Our modeling approach is to break the regression into two steps. First, we predict daily peak demand from daily peak temperature using an artificial neural network. Next, we downscale daily values to hourly based on historical demand profiles (method TBD).

Requirements

The following libraries are used in this project:

• keras-tuner version: 1.1.2
• pandas version: 1.4.2
• tensorflow version: 2.8.0
• sklearn version:0.0

Setup

To install the required libraries in a virtual environment:

	python -m venv env
	source env/bin/activate
	python -m pip install -r requirements.txt

To run the ANN pipeline on the University of Michigan ARC Great Lakes computing cluster, the project directory should be changed in the default_config.json to a reasonable location (e.g. a directory in Turbo Research Storage). The configuration file contains a parameter template for future models. It defines data filepaths, the hyperparameter search space, and other pipeline inputs.

File Descriptions

User Scripts

ann_pipeline_gl.sbat: Executes ann_pipeline.sh, invoking SLURM on Great Lakes.

ann_pipeline.sh: Builds config file and file structure, then executes ann_pipeline.py.

compile_results.py: Collects results from top-level project directory to single folder.

make_plots.py: Statically generates figures for several balancing authorities.

Data & Model (src)

ann_pipeline.py: Downloads and processes data, builds ANN, conducts a hyperparameter search, and saves results.

download_data.py: Module to download temperature and demand data from their respective git repositories.

clean_data.py: Module to reformat raw downloaded data to full, consistent data files.

process_data.py: Module to feature engineer and split data into training, validation, and test datasets.

model.py: Module for building ANN with TensorFlow.

hyperparameter_search.py: Module to conduct hyperparameter search with Keras Tuner.

evaluate.py: Module to compile evaluation metrics into a DataFrame.