Virtual agent organizations for user behaviour pattern extraction in energy optimization processes: A new perspective

NEUROCOMPUTING

卷:452

Article

Gonzalez-Briones, Alfonso^1,2,3  Prieto, Javier²  De la Prieta, Fernando²  Demazeau, Yves⁴  Corchado, Juan M.^2,3,5,6 

[1] Univ Complutense Madrid, Res Grp Agent Based Social & Interdisciplinary Ap, Madrid 28040, Spain

[2] Univ Salamanca, BISITE Res Grp, Edificio IDi,Calle Espejo 2, Salamanca 37007, Spain

[3] IoT Digital Innovat Hub, Air Inst, Salamanca 37188, Spain

[4] Univ Grenoble Alps, CNRS LIG, Grenoble, France

[5] Osaka Inst Technol, Dept Elect Informat & Commun, Fac Engn, Osaka 5358585, Japan

[6] Univ Malaysia Kelantan, Pusat Komputeran & Informat, Karung Berkunci 36, Kota Baharu 16100, Kelantan, Malaysia

关键词: Energy savings;    Virtual organization;    CBR system;    Sensor-based monitoring;    Ambient intelligent;   

DOI  :  10.1016/j.neucom.2020.05.117

来源: Elsevier

【 摘 要 】

The optimization of energy use in family homes and public buildings is an ongoing topic of discussion. State-of-the-art research has almost always focused on reducing the consumption of heating systems, air-conditioning or lighting. Despite their importance, user-related variables, such as comfort, are normally not included in the optimization process. These aspects should be considered to be able to effectively minimize energy consumption. Thus, there is a need for a comprehensive energy optimization approach, one that will consider both climatological factors and user behaviour. Learning about user behaviour is key to effective optimization. In this work, the proposed architecture's capacity to organize Virtual Agent Organizations (VAO) allows it to adapt to highly variable user behavior and preferences. This agent methodology has the ability to manage Wireless Sensor Networks (WSNs), Artificial Neural Networks (ANN) and Case-Based Reasoning (CBR) to obtain user preferences and predict their behaviour in the home or building. The proposed approach has been tested in two different buildings, a traditional-construction house and a modular home, obtaining savings of 30.16% and 13.43%, respectively. These results validate the proposed mixed approach of temperature adjustment algorithms together with the extraction of user behavior patterns for the establishment of a threshold based on preferences. (C) 2020 Elsevier B.V. All rights reserved.

【 授权许可】

Free