Controlador de demanda de energia elétrica utilizando redes neurais artificiais

Abstract

In the current global situation of scarcity of resources, the saving of electric energy has become relevant. Therefore, the Federal University of Lavras, with the objective of avoiding the surpassing of contracted electric energy demand, acquired a demand controlling system. The present work had as objective to improve the performance of the demand controller, optimizing, intelligently and dynamically, the priority of load disconnection. For this, computational intelligence techniques were used to create an automated system that together with the demand controller can achieve better performance. A priority classification system for each load was developed, as well as a system for forecasting the demand for electricity for the next 15 minutes. The methodology used was based on Artificial Neural Networks for the development of the two computational systems working in parallel. The results obtained in the simulation showed satisfactory indexes for the prediction system and for the classification system. It is concluded through the results obtained that the demand controller, with the application of the forecasting and classification techniques, will have an optimization of the load disconnection process.