Algoritmo híbrido neural-imuno aplicado ao diagnóstico de distúrbios de tensão em sistemas de distribuição de energia elétrica

• Ano de defesa: 2015
• Autor(a) principal: Lima, Amanda Parra dos Anjos [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Sistemas de energia eletrica - Disturbios; Energia elétrica - Distribuição; Redes neurais (Computação); Algoritmos; Electric power distribution
• Link de acesso: http://hdl.handle.net/11449/124493; http://www.athena.biblioteca.unesp.br/exlibris/bd/cathedra/25-06-2015/000839340.pdf

In this work, is presented a new approach for the detection and classification of voltage disturbances in distribution electrical systems using a hybrid algorithm as a tool Neural- Immune. This is, the junction of an artificial neural network ARTMAP-Fuzzy with the negative selection immune algorithm, creating the hybrid method. Thus, starting from measurements made on a power distribution feeder for SCADA data acquisition system, a Neural module performs the detection process, identifying abnormalities (disturbances), and an Immune module performs classification of detected abnormalities. The main application of this new approach is to assist decision making in order to automate the system operation process during the occurrence of failures. To evaluate the efficiency of the proposed method, were performed simulations of voltage disturbances in distribution electric systems in the ATP/EMTP software, using the test systems of 5, 33 and 84 bars and the real system of 134 bars. The results demonstrate efficiency and accuracy when compared to literature