Analysis and Prediction of Electrical Quantities for Energy Management in Public Buildings
Power quality. Forecasting of electrical variables. Neural networks. Energy management system.
This work proposes a systematic methodology for the analysis and prediction of electrical quanti ties in buildings, focusing on optimizing energy management in the public sector. The approach integrates the use of digital multi-meters and data science techniques, using the facilities of the Legislative Assembly of Rio Grande do Norte (ALRN) as a case study. For method validation, high-precision meters were positioned at the Main Low Voltage Switchboard (MLVS/QGBT) and at the essential loads panel connected to the generator bus, enabling continuous acquisition of parameters such as voltage, current, power, power factor, and total harmonic distortion (THD). Communication is performed via the TCP/IP protocol, allowing for systematic collection and data export in digital format. The methodology includes information pre-processing, exploratory anal ysis of power quality, disturbance identification, and the subsequent application of exploratory predictive models, covering both statistical methods and machine learning. As a tangible result, a computational prototype was developed to centralize data monitoring, analysis, and prediction, generating technical reports and alerts for decision-making support. It is concluded that the proposed solution is scalable to various public and private administration scenarios, providing support for energy management strategies, predictive maintenance, and operational sustainability in electrical energy use.