승강기 결함 진단을 위한 진동 신호 기반 특징 분석

Abstract

An elevator is a machine composed of various components. Extensive research has been conducted to determine the optimal life cycle of the components; however, there is a lack of methodological research on the diagnosis of the elevator condition. In this study, an efficient method for diagnosing faults through feature-based analysis on elevator vibration measurement three-axis sensor systems is proposed. The obtained data consists of normal and fault signals, and a sample is secured through a sampling process in a constant speed section of the signal. Subsequently, features with statistical and shape information are extracted from sampled signals and finally, machine learning consisting of Genetic Algorithm (GA)-based feature selection and Support Vector Machine (SVM) is applied to classify faults and evaluate diagnostic possibilities.