MACHINE LEARNING APPROACH FOR POWER MANAGEMENT USING IOT ENABLES EDGE COMPUTING

Abstract

An essential component of smart cities that could be used to effectively manage power was cognitive load forecasting. But a small study was undertaken about this element of Energy Management (EM) in smart cities using the Internet of Things (IoT). Throughout this study, a new Deep Learning (DL) approach was applied to predict how much energy would be used quickly while maintaining good communication between power providers and consumers. The Energy- Net stack contains several stacked spatiotemporal components, most of which are composed of a Temporal Transformer (TT) sub-module and perhaps a Spatial Transformer (ST) comment thread. While the ST approach uses the combination of convolution to retrieve concealed location data, the TT approach denotes temporal associations. In comparison with some other DL approaches that have an RMSE of 0.535 & 0.354, experimental dataset studies show that our technique was preferable. When used in conjunction with an IoT remote server to interface with power buildings and handle electricity activities, Power Net computational cost was appropriate for IoT devices having reliable finite resources.