A Deep Learning Approach for Intelligent IoT based Energy Management System

The intelligent load forecasting is an important part of smart cities, used for the purpose of energy management in an efficient manner. However, there is not much research done on this aspect of energy management (EM) in smart cities, with the help of Internet of Things (IoT). In this work, a new deep learning (DL) methodology is used for predicting the amount of energy consumed within a short period of time, while maintaining proper communication between the users and energy providers. There are multiple stacked spatiotemporal modules present in the Energy-Net stack such that every module is made up of a Spatial Transformer and a Temporal Transformer (TT) sub-module. The temporal relationships are denoted in the TT model while the ST model uses the integration of convolutional layers to extract the hidden spatial information. Experimental observations on the datasets indicate that this methodology proves to be superior when compared with other existing DL methods that have an RMSE of 0.535 and 0.354. Energy Net computational complexity is suitable for IoT devices with dependable resource constraints, in collaboration with IoT cloud server to communicate with the smart grids to manage energy-related tasks.