IET Science Measurement & Technology · 2019 · 13 citations · 16 references
EngineeringElectrical EngineeringCircuit TheoryMicroelectronicsTime Series AnalysisDiagnosisDiagnosticsElectrical System DiagnosticsFault DetectionFault DiagnosisSupport Vector MachinesExperimental Results InterpretationFault Localization AlgorithmsMachine Learning For Fault DetectionReflectometrySupport Vector ClassificationSupport Vector Machine ClassificationSupport Vector RegressionTime-domain Reflectometry
An efficient diagnosis method dedicated to embedded wiring network based on reflectometry technique is developed in this study. The proposed methodology is based on the two complementary steps. In the first step, the time‐domain reflectometry (TDR) method is simulated, by RLCG ( R : resistance, L : inductance, C : capacitance and G : conductance) circuit model and the numerical finite‐difference time‐domain method, and at the same time the datasets are created. In the second step, the support vector machine (SVM) algorithm is combined with a principal component analysis to identify the faults on wiring network from the TDR response. Two types of SVM models have been used in the diagnosis procedure: SVM classifiers and SVM regression models. In order to illustrate the performances and the feasibility of the proposed approach, numerical and experimental results are presented.
16
Support Vector Regression Machines
Harris Drucker, Christopher J. C. Burges, Linda Kaufman et al. · 1996 · 4.2K citations