AI - Based Smart Energy Consumption Prediction for IoT - Connected Homes

Omkar Reddy Polu

Abstract: As smart homes are becoming more and more adopted by IoT technology, so are energy consumption levels increasing, which requires for intelligent and practical energy management solutions. The Smart Energy Consumption Prediction System is an AI powered, IoT, machine learning (ML), blockchain and edge computing integrated system to predict dynamically the consumption of energy. In the proposed system, the deep learning models including LSTMs, Transformers, and hybrid AI models are applied to real time energy consumption forecasting. Furthermore, optimization based on reinforcement learning is used for automation of energy scheduling aiming for the input of minimum cost with minimal power wastage. This is supplemented with a blockchain - based peer to peer (P2P) energy trading system that will allow for decentralized and secure energy transaction between smart home users. Federated learning is used to scale the system and achieve privacy, in order to train decentralized AI to multiple households without data leakage. It is integrated with edge computing in order to process the streaming sensor data with minimum latency and do away with cloud computing. Real world datasets are evaluated on the system, with, more accurate predictions, lower energy demand and cost savings. Ultimately, this research lays the groundwork for future AI enabled energy management solutions with respect to smart city infrastructure, and the smart city infrastructure of the future, as it aligns with the sustainable development goals.

Keywords: AI - driven energy management, IoT, deep learning, smart homes, energy prediction, blockchain, edge computing, federated learning, reinforcement learning, peer - to - peer energy trading

How to Cite?: Omkar Reddy Polu, "AI - Based Smart Energy Consumption Prediction for IoT - Connected Homes", Volume 13 Issue 12, December 2024, International Journal of Science and Research (IJSR), Pages: 1836-1841, https://www.ijsr.net/getabstract.php?paperid=SR24128114158, DOI: https://dx.doi.org/10.21275/SR24128114158