AI-Enhanced System to Monitor Real-Time Energy and to Identify Home Appliances

• M. Khan Department of Mechatronics, Nelson Mandela University, South Africa https://orcid.org/0009-0000-3184-8723
• F. Smith Department of Mechatronics, Nelson Mandela University, South Africa https://orcid.org/0000-0003-4417-7388
• S. van Aardt Department of Mechatronics, Nelson Mandela University, South Africa https://orcid.org/0000-0001-6731-8257
• S. Hatefi Department of Mechatronics, Nelson Mandela University, South Africa https://orcid.org/0000-0001-5103-1198

DOI:

https://doi.org/10.7166/36-3-3350

Abstract

An AI-powered energy monitoring system is presented to identify real-time household appliances and to track power use. The proposed system combines low-cost, non-intrusive clip-on current and voltage sensors with a Raspberry Pi. A multi-layer perceptron neural network processes electrical measurements in near real-time (every two seconds) to disaggregate the total load and to identify active appliances. In a prototype with four appliances – a kettle, toaster, heater, and fan –  the system achieved 100% classification accuracy under controlled single-appliance test conditions. The system provides users with a web-based dashboard displaying each appliance’s power draw, energy consumption, and estimated cost in real time. The key contributions are: (1) a hardware design integrating a Raspberry Pi with an alternating current/voltage sensor for plug-and-play household monitoring; (2) a lightweight AI model deployed on edge computing for immediate appliance recognition; and (3) a user interface that delivers actionable feedback to encourage energy-efficient behaviour. The experimental results demonstrate an accurate identification of appliance signatures and responsive performance, although future work is needed to generalise the model to a larger variety of devices and to simultaneous appliance operation. This approach lays a foundation for smarter home energy management by combining IoT sensing with artificial intelligence, thereby empowering users with fine-grained, real-time insight into their electricity usage.