Detection and Assessment of I&C Cable Faults Using Time–Frequency R-CNN-Based Reflectometry
Chun-Kwon Lee, Yong–June Shin
Abstract
In this article, we propose a fault detection and assessment technique for instrumentation and control cables based on time-frequency image classification using the faster region-based convolutional neural network (R-CNN). To train the faster R-CNN while compensating for multiple reflections, the reflected signal estimation is utilized, which divides the reflected signal into the signal propagation along the cable and the reflection from the impedance discontinuity point. Experimental results on two fault scenarios under the circumstance of multiple faults detection and branched networks demonstrate the effectiveness of the proposed method.
Topics & Concepts
ReflectometryConvolutional neural networkSIGNAL (programming language)Fault detection and isolationComputer scienceFault (geology)Reflection (computer programming)Electrical impedanceDiscontinuity (linguistics)Time–frequency analysisElectronic engineeringEngineeringArtificial intelligenceElectrical engineeringTelecommunicationsTime domainComputer visionGeologyMathematicsMathematical analysisProgramming languageRadarSeismologyActuatorElectrical Fault Detection and ProtectionIntegrated Circuits and Semiconductor Failure AnalysisUltrasonics and Acoustic Wave Propagation