Synthetic Time-Series Load Data via Conditional Generative Adversarial Networks

TL;DR

This paper introduces a framework using conditional GANs to generate realistic synthetic transmission-level load data, enabling improved data availability for power system analysis and machine learning tasks.

Contribution

It presents a novel application of conditional GANs for generating season- and load type-specific synthetic load profiles from real data.

Findings

Synthetic data closely matches real load characteristics

Generated profiles are useful for power system simulations

Model effectively captures seasonal and load-type variations

Abstract

A framework for the generation of synthetic time-series transmission-level load data is presented. Conditional generative adversarial networks are used to learn the patterns of a real dataset of hourly-sampled week-long load profiles and generate unique synthetic profiles on demand, based on the season and type of load required. Extensive testing of the generative model is performed to verify that the synthetic data fully captures the characteristics of real loads and that it can be used for downstream power system and/or machine learning applications.