배관 균열형태와 밸브개폐에 따른 음향방출신호 및 진동신호의 특징분석

Abstract

Machinery typically consists of numerous pipes and valves. In particular, Pipes in power plant, are prone to defects owing to high-temperature, high-pressure steam. Such defects may result in reduced efficiency and increased risk of accidents due to fluid leakage. In this study, an experimental model used pipe used in heat exchangers to detect early defects in pipes. The classification performance of the vibration and acoustic emission signals was evaluated; specifically, the classification performance of acoustic emission signals before and after the pre-processing was evaluated. The characteristics used for classification were selected using three discriminative features by applying a GA(genetic algorithm), and the classification performance was evaluated using a support vector machine. The results of the study are expected to be the basis for early detection of pipe defects.