http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Prediction of component lifetime using simulation based on accelerated destructive degradation test
Munwon Lim(임문원),Gyu Ri Kim(김규리),Seunghak Chai(채승학),Suk Joo Bae(배석주) 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.4
In real industry, reliability evaluation of newly developed products is conducted through performance evaluation process. Due to the limitation of production period, accelerated degradation test (ADT) was developed to test the products as quickly as possible before the release of new products. By stressing the products beyond their use condition, the components of product are degraded faster than the normal environment. Using time-series performance measurements in accelerated condition, the degradation path or failure time distribution of the components can be predicted. In case the degradation of components should be measured only through destructive inspection, accelerated destructive degradation test (ADDT) is conducted. In this study, we propose an ADDT-based reliability analysis procedure. To describe the decaying pattern of ADDT data, various nonlinear degradation models are applied and simulated. Based on the generated degradation paths from Monte Carlo simulation, the failure time distribution and remaining useful life of the observations can be estimated. Through the application of bi-functional DC motor in a headlight system, the suggested approach estimates the reliability and remaining useful life of the components considering a variability of the degradation models.
웨이블릿 스펙트럼을 이용한 스마트 팩토리 설비의 이상감지 및 진단
문병민(Byeong Min Mun),임문원(Munwon Lim),김성준(Seong-Joon Kim),배석주(Suk Joo Bae) 한국신뢰성학회 2019 신뢰성응용연구 Vol.19 No.1
Purpose: Condition-based maintenance (CBM) is widely used to decrease the risk of equipment failures. A signal data indicating the health status of equipments is continuously measured in CBM. This article proposes a fault detection and diagnosis approach for smart factory equipments based on the signal processing and feature extraction techniques using a support vector machine (SVM). Methods: We propose a discrete wavelet transform (DWT) as one of signal processing methods. After processing the signal data, we derive the representative energy spectrum through various measures such as mean, median, variance, and interquartile range (IQR). Finally, the SVM is used to classify two classes based on Gaussian radial basis function (RBF) kernel. Results: we applied the proposed method to signal data collected from the equipment. We compared the classification accuracy of the SVM. At window length of 2⁹(J=9), the wavelet spectrum through the variance measure provides the best classification accuracy for the signal data of the equipment. Conclusion: In this article, fault detection and diagnosis methods for smart factory equipments are proposed.