Skip to Main content Skip to Navigation

Comparaison et développement de méthodes avancées de détection de défauts sur des câbles coaxiaux

Abstract : In this thesis, we present new approaches of soft fault detection and location in simple andcomplex wire networks. The idea is to find a new approach to overcome the difficulties withstandard reflectometry techniques. We prove that before applying post-treatment methods,denoising techniques should be applied, such as empirical mode decomposition (EMD), localmean decomposition (LMD), or the discrete wavelet transform (DWT). These three methodsdecompose a signal into multiple levels to threshold them before signal reconstruction.Testing several applications shows that EMD is the most efficient method, although it hassome limitations as side effects. After the denoising step, the wiring faults can be detected.Time–frequency analysis is employed at this step. This approach, based on the FourierTransform, is able to detect wiring faults only if the noise level is low. To overcome thisdifficulty, the Bayesian approach is beneficial when system complexity increases. Its responseis based on estimation of prior parameters and prior distributions. In this work, the Bayesianapproach is applied via a formal mathematical study followed by simulation results validatingthe proposed approach, with analysis of the parameters that affect the method’s performance.In the domain of soft fault location, we derive a chaos time domain reflectometry approachbased on chaotic signal properties. Our simulation and experimental results prove that thismethod can synthesize signals and localize the soft fault position without the need forsupplemental methods.
Complete list of metadatas
Contributor : Abes Star :  Contact
Submitted on : Monday, March 23, 2020 - 9:05:18 AM
Last modification on : Tuesday, March 24, 2020 - 1:41:58 AM
Long-term archiving on: : Wednesday, June 24, 2020 - 1:10:17 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02514852, version 1



Ihssane Bzikha. Comparaison et développement de méthodes avancées de détection de défauts sur des câbles coaxiaux. Signal and Image processing. Université de Limoges, 2019. English. ⟨NNT : 2019LIMO0119⟩. ⟨tel-02514852⟩



Record views


Files downloads