Navegando por Data de Publicação, começando com "2022-10-22"

Filtrar resultados por ano ou mês

Agora exibindo 1 - 2 de 2

Predição de carga de energia elétrica no Brasil utilizando técnicas de Deep Learning
(Universidade Federal de São Carlos, 2022-10-22) Cunha, Leticia Amaral da; Suetake, Marcelo; https://lattes.cnpq.br/8630323559353535; Inoue, Roberto Santos; https://lattes.cnpq.br/6221209121565990
The open electricity market has stood out for being a competitive and dynamic way of trading an asset that was previously restricted to passive consumption. The freedom of negotiation provided to the participants of this market makes the prediction of its price, as well as the factors that affect it, extremely important. The purpose of this study is to develop three models based on different techniques of recurrent artificial neural networks and their comparative analysis. For this, the LSTM Feed Forward, LSTM bidirectional and GRU architectures must be applied to each of the time series explored. Average electric energy load data from the Brazilian free energy market between the years 2011 to 2021 are applied as an input for training, validation and testing processes. For the evaluation of the results, metrics of wide application were used, such as mean square error and mean absolute percentage error. The analysis of the models verifies the performance of the applied techniques, as well as possible causes for the performance of the established metrics. The results showed the bidirectional LSTM and GRU types as viable modeling results
Uma abordagem para predição da estabilidade transitória em sistemas elétricos de potência a partir de redes neurais artificiais
(Universidade Federal de São Carlos, 2022-10-22) Mancini, Gabriel; Fernandes, Ricardo Augusto Souza; https://lattes.cnpq.br/0880243208789454; Fernandes, Tatiane Cristina da Costa; https://lattes.cnpq.br/2590102586758022; https://lattes.cnpq.br/3923574181707452
In an electrical network, the assessment and prediction of dynamic safety are essential to avoid interruptions in the supply of electrical energy to consumers, in addition to ensuring that the system operates reliably. Regarding the transient security assessment, predicting the behavior of the electrical power system in the post-fault period within a short time interval is essential so that preventive and, if necessary, corrective actions are taken. Traditionally, in the planning and operation of an electrical network, this study can be developed through the numerical solution of nonlinear differential equations, however, the accuracy of this analysis is limited by the capacity of representation of the model and the need for its constant updating. Considering the operation of power systems nowadays, with a continuous increase in load demand as well as the participation of energy generation sources with intermittent characteristics, this task has become increasingly challenging, increasing the degree of complexity of the mathematical model. On the other hand, the improvement of phasor measurement systems boosts the use of machine learning techniques to predict the status of the transient stability of an electrical network, among which neural networks have stood out. In view of this, the present work aims to investigate the use of artificial neural networks to predict transient stability in an electrical power system through measurements that can be easily measured by phasor measurement units, such as magnitude and voltage angle of the buses. In the proposed structure, the smallest number of measurement cycles necessary to perform the prediction is investigated, so that the behavior of the system (stable or unstable) is identified in the shortest possible time, helping the operator to make a decision. The results obtained in the IEEE 68 bus system show the efficiency of the predictor, which obtained an accuracy of 97.5% in the classification from 6 consecutive cycles of measuring the system response. When only 1 measurement cycle was provided to the predictor, a high accuracy is also achieved 96.1 % by the proposed methodology.