Introducing Parallelism In Power System Diagnosis
C. Rodriguez, Javier Muguerza, Jose Ignacio Martin, Alberto Lluch Lafuente, Santiago Rementeria · 1994
In this paper a fault diagnosis system for transport electrical grids based on neural networks is presented. l3.e functional modularity adopted in the solution allows to partition the electrical network in a way such that parallel architectures can be used to obtain acceptable response times in realistic operation. Using this type of architecture permits to improve the cost @netion used by experts to determine the faulty element. In order to achieve that, it is proposed to obtain the cost function in a segmented way. The particular characteristics of the power system to monitor will condition the parallel architecture selection.