The user side of sustainability: Modeling behavior and energy usage in the home

https://doi.org/10.1016/j.pmcj.2012.10.004Get rights and content

Abstract

Society is becoming increasingly aware of the impact that our lifestyle choices make on energy usage and the environment. As a result, research attention is being directed toward green technology, environmentally-friendly building designs, and smart grids. This paper looks at the user side of sustainability. In particular, it looks at energy consumption in everyday home environments to examine the relationship between behavioral patterns and energy consumption. It first demonstrates how data mining techniques may be used to find patterns and anomalies in smart home-based energy data. Next, it describes a method to correlate home-based activities with electricity usage. Finally, it describes how this information could inform users about their personal energy consumption and to support activities in a more energy-efficient manner. These approaches are validated by using real energy data collected in a set of smart home testbeds.

Introduction

In 2010, the United States consumed 98,003 quadrillion Btu of power energy. This is a 200% increase from 1949 [1]. The growth of energy usage is not entirely due to manufacturing plants and automobiles, as is often assumed. In fact, the worldwide residential sector is responsible for 16%–50% of energy consumption by all sectors [2]. As a result, there is an urgent need to develop technologies that examine energy usage in homes and to encourage energy efficient behaviors, in addition to energy efficient devices in households.

Although households and buildings are responsible for over 40% of energy usage in most countries [3], many residents still receive little or no detailed feedback about their personal energy usage. A power utility bill traditionally provides information about a month’s total energy consumption and a total price to be paid, leaving homeowners to guess what factors, including external influences and internal behavior, might explain a higher or lower than usual bill. Earlier studies have shown that home residents reduce energy expenditure by 5%–15% just as a response to acquiring and viewing raw energy usage [4]. Residential behavior, which varies widely, can influence energy usage significantly in a given home [5]. Clearly, the typical utility bill provides no information about the relationship between residential behavior and corresponding energy usage. Since behavior-based energy information is capable of encouraging individuals to modify habits in ways that would be beneficial for both the household and community, it would be desirable to develop technologies that could extract information from the home and communicate it to residents. However, occupants’ behavior is difficult to capture accurately. Self-reporting of behavior is error prone [6] and whole-home meter monitoring does not capture the behaviors in the home that influence consumption.

We hypothesize that providing users with knowledge about the relationship between their activities and energy consumption, suggestions for energy reduction, and automation support will result in more substantial decreases in overall consumption. This view is supported by an increasing body of work that links awareness of energy consumption and its impact to behavioral change [7], [8]. In our work we propose using smart homes and pervasive computing techniques to provide these important insights. The long-term vision for this project is to enhance understanding of human resource consumption and to provide resource efficiency in smart homes. We envision this as a three-step process: (1) analyze electricity usage to identify clusters and anomalies, (2) correlate activities with energy usage, and (3) automate energy-efficient activity support. Additionally, we hypothesize that patterns and anomalies may be automatically detected from energy consumption data and that these discoveries can provide insights on behavioral patterns. We further postulate that energy consumption is correlated with the type of activities that are performed and can therefore be predicted based on the automatically-recognized activities that occur in a smart environment. These hypotheses are validated by implementing algorithms to perform these steps and evaluating the algorithms using data collected in the smart apartment testbeds. Finally, a discussion of how the results of this work can be used to give smart home residents feedback on their energy consumption is included. This work represents one of the first projects that utilizes smart home data to investigate the relationship between behavioral patterns and resource consumption in a home environment.

Section snippets

Related work

A smart home environment can be defined as one that acquires and applies knowledge about its residents and their physical surroundings in order to improve their experience in that setting [9]. Such home environments, equipped with sensors for detecting features such as motion, light level, temperature, and energy and water consumption, are ideal testbeds for investigating the relationship between behavior and energy consumption. Using sensor technology combined with data mining and machine

CASAS-Sustain system architecture

In this paper, we describe a prototype system framework for energy data collection, energy data transformation, and energy data analysis, as shown in Fig. 1. The system, called CASAS-Sustain, operates entirely within the structure of the CASAS smart home project [20]. As the diagram indicates, data collected in the smart home is first analyzed to look for patterns and outliers. Next, recognized activities in the home are correlated with energy usage to provide working information on the energy

Energy data analysis outlier detection

Our first step in utilizing smart home technologies for energy efficiency is to better understand the nature of the energy consumption itself. We begin by analyzing normal patterns of usage and identifying abnormal or anomalous situations. We analyze normal patterns by clustering sequences of power usage values. This analysis is useful because the cluster descriptions can provide users with insights on their daily habits and resource usage as well as provide software algorithms with a model of

Activity-based energy prediction

In the second step of our CASAS-Sustain analysis, machine-learning techniques were used to predict energy consumption given information about an activity that residents perform in a smart environment. Because activity recognition techniques are prevalent in the literature [25] and are becoming more robust, this offers a practical approach to automatically correlating activities in the home with energy consumption. The following features were used to describe an activity performed by an

Mobile sustainability intervention

The last component of CASAS-Sustain is behavior-based intervention to promote sustainability in everyday environments. We focus on a pervasive approach to promote sustainability behavior. Fig. 17 shows our CASASviz visualizer, which is web-based and can therefore run on a computer display or a mobile device [33]. CASASviz describes an environment graphically using Scalable Vector Graphics. In one mode (the one shown in Fig. 17(a)), users can look at an interface which displays sensor events

Conclusions

In this article, we consider the role of in-home behaviors upon energy usage. In particular, we analyze patterns of energy usage by monitoring activities as well as collect energy usage data from several smart environments. We analyzed the energy patterns by identifying frequent sequences of energy usage ranges and identifying outliers in the data. We further identify the role of behaviors for energy usage by using machine learning methods to map activities performed in the environment with

References (33)

  • T. Ichinose et al.

    Impact of anthropogenic heat on urban climate in Tokyo

    Atmospheric Environment

    (1999)
  • J.E. Seem

    Using intelligent data analysis to detect abnormal energy consumption in buildings

    Energy and Buildings

    (2007)
  • Multiple, Annual energy review 2010, Technical Report, U.S. Energy Information Administration,...
  • Multiple, Energy Efficiency in Buildings. http://www.wbscd.org/,...
  • S. Darby

    The effectiveness of feedback on energy consumption

  • J. Seryak, K. Kissock, Occupancy and behavioral affects on residential energy use, in: Proceedings of the Solar...
  • S. Szewcyzk et al.

    Annotating smart environment sensor data for activity learning

    Technology and Health Care

    (2009)
  • S. Darby

    Smart metering: what potential for householder engagement?

    Building Research & Information

    (2010)
  • Y. Riche, J. Dodge, R.A. Metoyer, Studying always-on electricity feedback in the home, in: Proceedings of the 28th...
  • D.J. Cook et al.
  • G. Hart

    Nonintrusive appliance load monitoring

    Proceedings of the IEEE

    (1992)
  • M. Berges, E. Goldman, H. Matthews, L. Soibelman, Learning systems for electric consumption of buildings, in: ASCE...
  • S. Gupta, M.S. Reynolds, S.N. Patel, Electrisense: single-point sensing using emi for electrical event detection and...
  • Google, Google Power Meter. http://www.google.com/powermeter/about/,...
  • Microsoft, Microsoft Hohm. http://www.microsoft-hohm.com/,...
  • Opower. http://opower.com/,...
  • Cited by (49)

    • A survey of smart home energy conservation techniques

      2023, Expert Systems with Applications
      Citation Excerpt :

      Energy optimization is a knapsack optimization problem (KP) known as an NP-complete problem. Likewise, Chen et al. (2013) develop a multi-step technique that identifies anomalies in smart home energy usage data using data mining techniques. This technique correlates user activities with appliance energy consumption patterns.

    • How do beliefs and attitudes of people influence energy conservation behavior in Pakistan?

      2022, Heliyon
      Citation Excerpt :

      Literature dictates two approaches to classifying energy-saving behavior by scene and nature. First, Chen et al. (2013) focus on three scenes (bedroom, kitchen, and living room) to analyze energy-saving behavior. However, by scene approach is restricted to include environmental aspects only.

    View all citing articles on Scopus
    View full text