Smart Meters Time Series Clustering for Demand Response Applications in the Context of High Penetration of Renewable Energy Resources

Santiago Bañales,Raquel Dormido,Natividad Duro

doi:10.3390/en14123458

Santiago Bañales, Raquel Dormido + Show 1 more

Open Access

https://doi.org/10.3390/en14123458

Copy DOI

Abstract

The variability in generation introduced in the electrical system by an increasing share of renewable technologies must be addressed by balancing mechanisms, demand response being a prominent one. In parallel, the massive introduction of smart meters allows for the use of high frequency energy use time series data to segment electricity customers according to their demand response potential. This paper proposes a smart meter time series clustering methodology based on a two-stage k-medoids clustering of normalized load-shape time series organized around the day divided into 48 time points. Time complexity is drastically reduced by first applying the k-medoids on each customer separately, and second on the total set of customer representatives. Further time complexity reduction is achieved using time series representation with low computational needs. Customer segmentation is undertaken with only four easy-to-interpret features: average energy use, energy–temperature correlation, entropy of the load-shape representative vector, and distance to wind generation patterns. This last feature is computed using the dynamic time warping distance between load and expected wind generation shape representative medoids. The two-stage clustering proves to be computationally effective, scalable and performant according to both internal validity metrics, based on average silhouette, and external validation, based on the ground truth embedded in customer surveys.

Highlights

Demand response (DR), or the capability of electrical loads to adapt their shape at specific points in time given the right incentives, is receiving increasing attention from policy makers and energy system designers [1]
Notwithstanding the large uncertainties introduced by the COVID-19 pandemic in the energy system [3], a sustainable economic recovery is set to be based on channeling new investments in clean energy and further digitalization [4], which would further promote the automation of demand response programs [5], in particular for residential customers [6] in the context of higher penetration of renewable resources
The widely used Irish CER dataset has been chosen to illustrate the application of the proposed methodology [42]. This dataset has three main advantages: (1) it is publicly available and results can be replicated; (2) the data are of high quality and significant size; and (3) it contains a detailed customer survey that can be used as ground truth

Summary

Introduction

Demand response (DR), or the capability of electrical loads to adapt their shape at specific points in time given the right incentives, is receiving increasing attention from policy makers and energy system designers [1]. Time series clustering approaches have been largely applied to smart meters load profiles datasets [11,12,13]. The challenges associated with time series clustering are well recognized, and they include high dimensionality and the definition of similarity taking the time dimension into account, from which three key research areas are derived: dimensionality reduction; clustering approach, which includes the choice of distance measurement, clustering prototypes and clustering algorithm; and clustering performance evaluation [14,15]

Objectives

Results

Discussion

Conclusion