Improving User Interaction in Smart Metering Systems Through AI-Powered Conversational Interfaces

Authors

  • Caio Pasqualon Departamento de Energia Elétrica, Universidade Federal de Juiz de Fora, MG, Brazil
  • Yan F. Coutinho Departamento de Energia Elétrica, Universidade Federal de Juiz de Fora, MG, Brazil
  • Túlio F. Moreira Departamento de Energia Elétrica, Universidade Federal de Juiz de Fora, MG, Brazil
  • Moisés V. Ribeiro Departamento de Energia Elétrica, Universidade Federal de Juiz de Fora, MG, Brazil

Keywords:

smart metering, artificial intelligence, user interface, large language model

Abstract

This work advances conversational interfaces powered by artificial intelligence (AI) models into smart metering systems to make them more user-friendly. By overcoming the complexity of human language and the limitations of current large language models, such as their tendency to produce false or undesirable responses and poor performance in interpreting numeric data or large volumes of data, we introduce AI-powered Data Analyzer and User Interface into smart metering systems. Numerical results obtained from a prototype of a single-user AI-powered smart metering system show recognition rates close to 100% for commands without spelling errors and 99% for commands with spelling errors, even without additional training in the large language models (LLMs). Moreover, these results highlight distinct trade-offs between accuracy and cost when using different pre-trained large language models. Overall, AI-based models are useful for making smart metering systems user-friendly.

Downloads

Published

2024-10-18

Issue

Vol. 4 No. 1 (2024): CBA2024

Section

Articles