Clustering Appliance Operation Modes With Unsupervised Deep Learning Techniques

Castangia, Marco; Barletta, Nicola; Camarda, Christian; Quer, Stefano; Macii, Enrico; Patti, Edoardo

doi:10.1109/tii.2022.3217495

Cited by 3 publications

(2 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both of these cycles are part of the same operation mode, but there is no way to recognize this when clustering the raw data. Furthermore, methods based on raw data are also susceptible to noisy points and outliers; thus, they often lead to poor clustering results [34]. For this reason, the first step of our work was to identify a method for operation mode recognition.…”

Section: A Hypothetical Smart Homementioning

confidence: 99%

“…The recognition of the operation modes of appliances is necessary to estimate the consumption value and the duration of the appliances' modes, thus keeping track of the status of smart devices at any given time. To address this problem, an approach similar to the one presented by Castangia et al [34] is used. Specifically, unsupervised deep learning clustering techniques are applied to a learned, latent-state representation of the raw data to produce the list of appliances' modes of operation.…”

Section: Recognition Of Appliance Operation Modesmentioning

confidence: 99%

See 1 more Smart Citation

Enabling End-User Development in Smart Homes: A Machine Learning-Powered Digital Twin for Energy Efficient Management

Cotti,

Guizzardi,

Barricelli

et al. 2024

Future Internet

View full text Add to dashboard Cite

End-User Development has been proposed over the years to allow end users to control and manage their Internet of Things-based environments, such as smart homes. With End-User Development, end users are able to create trigger-action rules or routines to tailor the behavior of their smart homes. However, the scientific research proposed to date does not encompass methods that evaluate the suitability of user-created routines in terms of energy consumption. This paper proposes using Machine Learning to build a Digital Twin of a smart home that can predict the energy consumption of smart appliances. The Digital Twin will allow end users to simulate possible scenarios related to the creation of routines. Simulations will be used to assess the effects of the activation of appliances involved in the routines under creation and possibly modify them to save energy consumption according to the Digital Twin’s suggestions.

show abstract

Section: A Hypothetical Smart Homementioning

confidence: 99%

Section: Recognition Of Appliance Operation Modesmentioning

confidence: 99%