M. Razghandi, H. Zhou, M. Erol-Kantarci, and D. Turgut

Smart Home Energy Management: VAE-GAN synthetic dataset generator and Q-learning


Cite as:

M. Razghandi, H. Zhou, M. Erol-Kantarci, and D. Turgut. Smart Home Energy Management: VAE-GAN synthetic dataset generator and Q-learning. Accepted to appear at IEEE Transactions on Smart Grid, May 2023. DOI: 10.1109/TSG.2023.3288824

Download:

Download 

Abstract:

In recent years, there has been a growing interest in academia and industry in the analysis of electrical consumption in residential buildings and the implementation of smart home energy management systems (HEMS) to reduce household energy usage and costs. HEMS have been designed to emulate the statistical and functional characteristics of real smart grids. However, a major challenge in this research area is the limited availability of publicly accessible datasets. To address this challenge and further leverage the potential of artificial HEMS applications, it is crucial to develop time series that accurately represent diverse operating conditions of synthetic systems. This paper introduces a novel approach based on the combination of variational auto-encoder-generative adversarial network (VAE-GAN) techniques to generate time-series data of energy consumption in smart homes. Additionally, we investigate the performance of the generative model when integrated with a Q-learning based HEMS. The effectiveness of the Q-learning based HEMS is assessed through online experiments using real-world smart home data. To evaluate the quality of the generated dataset, we employ various metrics including KullbackÔÇôLeibler (KL) divergence, maximum mean discrepancy (MMD), and the Wasserstein distance, which quantify the disparities between probability distributions of the real and synthetic data. Our experimental results demonstrate that the synthetic data generated by VAE-GAN closely aligns with the distribution of real data. Furthermore, we demonstrate that the utilization of the generated data facilitates the training of a more efficient Q-learning based HEMS, surpassing the performance achieved with datasets generated using baseline approaches.

BibTeX:

@article{Razghandi-2023-TSG,
	author = "M. Razghandi and H. Zhou and M. Erol-Kantarci and D. Turgut",
	title = "Smart Home Energy Management: VAE-GAN synthetic dataset generator and Q-learning",
	journal = "Accepted to appear at IEEE Transactions  on Smart Grid",
   year = "2023",
   month = "May",
   note = "DOI: 10.1109/TSG.2023.3288824",
	abstract = {
      In recent years, there has been a growing interest in academia and industry in the analysis of electrical consumption in residential buildings and the implementation of smart home energy management systems (HEMS) to reduce household energy usage and costs. HEMS have been designed to emulate the statistical and functional characteristics of real smart grids. However, a major challenge in this research area is the limited availability of publicly accessible datasets. To address this challenge and further leverage the potential of artificial HEMS applications, it is crucial to develop time series that accurately represent diverse operating conditions of synthetic systems.
      This paper introduces a novel approach based on the combination of variational auto-encoder-generative adversarial network (VAE-GAN) techniques to generate time-series data of energy consumption in smart homes. Additionally, we investigate the performance of the generative model when integrated with a Q-learning based HEMS. The effectiveness of the Q-learning based HEMS is assessed through online experiments using real-world smart home data. To evaluate the quality of the generated dataset, we employ various metrics including Kullback–Leibler (KL) divergence, maximum mean discrepancy (MMD), and the Wasserstein distance, which quantify the disparities between probability distributions of the real and synthetic data. Our experimental results demonstrate that the synthetic data generated by VAE-GAN closely aligns with the distribution of real data. Furthermore, we demonstrate that the utilization of the generated data facilitates the training of a more efficient Q-learning based HEMS, surpassing the performance achieved with datasets generated using baseline approaches.},
}

Generated by bib2html.pl (written by Patrick Riley, Lotzi Boloni ) on Mon Nov 13, 2023 12:30:29