A deep learning architecture for power management in smart cities

Energy Reports

Mamoun Alazab¹  Rubén González Crespo²  Qin Xin³  Vicente García Díaz⁴  Carlos Enrique Montenegro-Marin⁵ 

[1] Corresponding author.;Facultad de ingeniería, Universidad Distrital Francisco José de Caldas, Colombia;Faculty of Science and Technology, University of the Faroe Islands, Vestarabryggja 15, FO 100 Torshavn, Faroe Islands, Denmark;IT and Environment, Charles Darwin University, Australia;University of Oviedo, Spain;

关键词: Deep learning;    Internet of Things;    Power management;    Wireless communication;   

DOI  :  

来源: DOAJ

【 摘 要 】

Sustainable energy management is an inexpensive approach for improved energy use. However, the research used does not focus on cutting-edge technology possibilities in an Internet of things (IoT). This paper includes the needs for today’s distributed generation, households, and industries in proposing smart resource management deep learning model. A deep learning architecture of power management (DLA-PM) is presented in this article. It predicts future power consumption for a short period and provides effective communication between power distributors and customers. To keep power consumption and supply constant, mobile devices are linked to a universal IoT cloud server connected to the intelligent grids in the proposed design. An effective brief forecast decision-making method is followed by various preprocessing strategies to deal with electrical data. It conducts extensive tests with RMSE reduced by 0.08 for both residential and business data sources. Significant strengths include refined device-based, real-time energy administration via a shared cloud-based server data monitoring system, optimized selection of standardization technology, a new energy prediction framework, a learning process with decreased time, and lower error rates. In the proposed architecture, mobile devices link to a universal IoT cloud server communicating with the corresponding intelligent grids such that the power consumption and supply phenomena continually continue. It utilizes many preprocessing strategies to cope with the diversity of electrical data, follows an effective short prediction decision-making method, and executes it using resources. For residential and business data sources, it runs comprehensive trials with RMSE lowered by 0.08.

【 授权许可】

Unknown