An augmented multi-objective particle swarm optimizer for building cluster operation decisions

Mengqi Hu, Jeffery D. Weir, Teresa Wu

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

It is envisioned that other than the grid-building communication, the smart buildings could potentially treat connected neighborhood buildings as a local buffer thus forming a local area energy network through the smart grid. As the hardware technology is in place, what is needed is an intelligent algorithm that coordinates a cluster of buildings to obtain Pareto decisions on short time scale operations. Research has proposed a memetic algorithm (MA) based framework for building cluster operation decisions and it demonstrated the framework is capable of deriving the Pareto solutions on an 8-h operation horizon and reducing overall energy costs. While successful, the memetic algorithm is computational expensive which limits its application to building operation decisions on an hourly time scale. To address this challenge, we propose a particle swarm framework, termed augmented multi-objective particle swarm optimization (AMOPSO). The performance of the proposed AMOPSO in terms of solution quality and convergence speed is improved via the fusion of multiple search methods. Extensive experiments are conducted to compare the proposed AMOPSO with nine multi-objective PSO algorithms (MOPSOs) and multi-objective evolutionary algorithms (MOEAs) collected from the literature. Results demonstrate that AMOPSO outperforms the nine state-of-the-art MOPSOs and MOEAs in terms of epsilon, spread, and hypervolume indicator. A building cluster case is then studied to show that the AMOPSO based decision framework is able to make hourly based operation decisions which could significantly improve energy efficiency and achieve more energy cost savings for the smart buildings.

Original languageEnglish (US)
Pages (from-to)347-359
Number of pages13
JournalApplied Soft Computing Journal
Volume25
DOIs
StatePublished - 2014

Fingerprint

Particle swarm optimization (PSO)
Intelligent buildings
Evolutionary algorithms
Energy efficiency
Costs
Fusion reactions
Hardware
Communication
Experiments

Keywords

  • Multi-objective optimization
  • Pareto optimality
  • Particle swarm optimization
  • Smart building
  • Smart grid

ASJC Scopus subject areas

  • Software

Cite this

An augmented multi-objective particle swarm optimizer for building cluster operation decisions. / Hu, Mengqi; Weir, Jeffery D.; Wu, Teresa.

In: Applied Soft Computing Journal, Vol. 25, 2014, p. 347-359.

Research output: Contribution to journalArticle

@article{cc20cc3c532e47ef9542706bb9fb3a9e,
title = "An augmented multi-objective particle swarm optimizer for building cluster operation decisions",
abstract = "It is envisioned that other than the grid-building communication, the smart buildings could potentially treat connected neighborhood buildings as a local buffer thus forming a local area energy network through the smart grid. As the hardware technology is in place, what is needed is an intelligent algorithm that coordinates a cluster of buildings to obtain Pareto decisions on short time scale operations. Research has proposed a memetic algorithm (MA) based framework for building cluster operation decisions and it demonstrated the framework is capable of deriving the Pareto solutions on an 8-h operation horizon and reducing overall energy costs. While successful, the memetic algorithm is computational expensive which limits its application to building operation decisions on an hourly time scale. To address this challenge, we propose a particle swarm framework, termed augmented multi-objective particle swarm optimization (AMOPSO). The performance of the proposed AMOPSO in terms of solution quality and convergence speed is improved via the fusion of multiple search methods. Extensive experiments are conducted to compare the proposed AMOPSO with nine multi-objective PSO algorithms (MOPSOs) and multi-objective evolutionary algorithms (MOEAs) collected from the literature. Results demonstrate that AMOPSO outperforms the nine state-of-the-art MOPSOs and MOEAs in terms of epsilon, spread, and hypervolume indicator. A building cluster case is then studied to show that the AMOPSO based decision framework is able to make hourly based operation decisions which could significantly improve energy efficiency and achieve more energy cost savings for the smart buildings.",
keywords = "Multi-objective optimization, Pareto optimality, Particle swarm optimization, Smart building, Smart grid",
author = "Mengqi Hu and Weir, {Jeffery D.} and Teresa Wu",
year = "2014",
doi = "10.1016/j.asoc.2014.08.069",
language = "English (US)",
volume = "25",
pages = "347--359",
journal = "Applied Soft Computing",
issn = "1568-4946",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - An augmented multi-objective particle swarm optimizer for building cluster operation decisions

AU - Hu, Mengqi

AU - Weir, Jeffery D.

AU - Wu, Teresa

PY - 2014

Y1 - 2014

N2 - It is envisioned that other than the grid-building communication, the smart buildings could potentially treat connected neighborhood buildings as a local buffer thus forming a local area energy network through the smart grid. As the hardware technology is in place, what is needed is an intelligent algorithm that coordinates a cluster of buildings to obtain Pareto decisions on short time scale operations. Research has proposed a memetic algorithm (MA) based framework for building cluster operation decisions and it demonstrated the framework is capable of deriving the Pareto solutions on an 8-h operation horizon and reducing overall energy costs. While successful, the memetic algorithm is computational expensive which limits its application to building operation decisions on an hourly time scale. To address this challenge, we propose a particle swarm framework, termed augmented multi-objective particle swarm optimization (AMOPSO). The performance of the proposed AMOPSO in terms of solution quality and convergence speed is improved via the fusion of multiple search methods. Extensive experiments are conducted to compare the proposed AMOPSO with nine multi-objective PSO algorithms (MOPSOs) and multi-objective evolutionary algorithms (MOEAs) collected from the literature. Results demonstrate that AMOPSO outperforms the nine state-of-the-art MOPSOs and MOEAs in terms of epsilon, spread, and hypervolume indicator. A building cluster case is then studied to show that the AMOPSO based decision framework is able to make hourly based operation decisions which could significantly improve energy efficiency and achieve more energy cost savings for the smart buildings.

AB - It is envisioned that other than the grid-building communication, the smart buildings could potentially treat connected neighborhood buildings as a local buffer thus forming a local area energy network through the smart grid. As the hardware technology is in place, what is needed is an intelligent algorithm that coordinates a cluster of buildings to obtain Pareto decisions on short time scale operations. Research has proposed a memetic algorithm (MA) based framework for building cluster operation decisions and it demonstrated the framework is capable of deriving the Pareto solutions on an 8-h operation horizon and reducing overall energy costs. While successful, the memetic algorithm is computational expensive which limits its application to building operation decisions on an hourly time scale. To address this challenge, we propose a particle swarm framework, termed augmented multi-objective particle swarm optimization (AMOPSO). The performance of the proposed AMOPSO in terms of solution quality and convergence speed is improved via the fusion of multiple search methods. Extensive experiments are conducted to compare the proposed AMOPSO with nine multi-objective PSO algorithms (MOPSOs) and multi-objective evolutionary algorithms (MOEAs) collected from the literature. Results demonstrate that AMOPSO outperforms the nine state-of-the-art MOPSOs and MOEAs in terms of epsilon, spread, and hypervolume indicator. A building cluster case is then studied to show that the AMOPSO based decision framework is able to make hourly based operation decisions which could significantly improve energy efficiency and achieve more energy cost savings for the smart buildings.

KW - Multi-objective optimization

KW - Pareto optimality

KW - Particle swarm optimization

KW - Smart building

KW - Smart grid

UR - http://www.scopus.com/inward/record.url?scp=84908317216&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908317216&partnerID=8YFLogxK

U2 - 10.1016/j.asoc.2014.08.069

DO - 10.1016/j.asoc.2014.08.069

M3 - Article

AN - SCOPUS:84908317216

VL - 25

SP - 347

EP - 359

JO - Applied Soft Computing

JF - Applied Soft Computing

SN - 1568-4946

ER -