Leaks detection in a pipeline using artificial neural networks Chapter in Scopus uri icon

Abstract

  • A system based on Artificial Neural Networks (ANN) is proposed to detect and diagnose multiple leaks in a pipeline leaks by recognizing the pattern of the flow using only two measurements. A nonlinear mathematical model of the pipeline is exploited for training, testing and validating the ANN-based system. This system was trained with tapped delays in order to include the system dynamics. Early results demonstrate the effectiveness of the approach in the detection and diagnosis of simultaneous multiple faults. © 2009 Springer-Verlag Berlin Heidelberg.

Publication date

  • December 1, 2009