LGEP 2011 ID = 812International audienceThis paper presents a new technique to reconstruct faulty wiring networks and/or to localize the defects affecting the branches of the wiring network from the time domain reflectometry response. The method is also for characterization of defects in branches on the network. The direct model for wave propagation along the transmission lines is modeled by RLCG circuit parameters computed by finite elements method (FEM) or analytical solution and the finite difference time-domain (FDTD) method. Neural networks (NNs) are used to solve the inverse problem. A set of experimental results is carried out in order to validate the calculations

International audienceBecause of the increasing size of electrical networks in automotives, the degradation of wires and connectors has become a major concern. As a matter of fact, severe system failures are often merely due to broken wires, bad crimping or degraded connectors. Furthermore, the difficulty to localize those kinds of defects using nowadays techniques generally leads to costly repair. Reflectometry is a well-known method used to monitor the health of lines and wired networks. Wiring networks can be affected with two types of faults: "soft ones" are created by the change of the impedance along the line due to different kinds of defects (insulation, radial crack, and degradation of connector...) in the wire and "hard ones" which correspond to open and short circuits. For the first type of faults, the frequency-domain response of the faulty wiring presents a modification of the impedance, in the defect location. For that reason, this paper focuses on the application of the reflectometry technique to the broadband characterization of a vibration degraded Cu-Sn connector. Once the Sn plating is removed, the electrical properties of the contact interface are entirely modified. The results presented in this paper show that observing such kind of transition can be used as criteria for estimating the degradation level of the connector

No abstract

No abstract