http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Y. Lee(이윤제),M. Yeo(여명환),H. Oh(오효근),C. Lee(이창우) Korean Society for Precision Engineering 2021 한국정밀공학회 학술발표대회 논문집 Vol.2021 No.11월
As the role of roll-to-roll systems within manufacturing industry expands, qualities for high maintenance and mass production relies on the accuracy of diagnosing the operating condition. Diagnosing machinery failure prevents unexpected expenditures by forcing maintenance to respond beforehand. As spindle bearings are a key element of roll-to-roll systems, this research proposes a method of optimization on statistical feature variables for diagnosis of spindle bearings. As fault diagnosis based on machine learning models require complex mathematical modeling and certain amount of time to structure, data-driven methodologies are an advanced approach commonly applied. The proposed method for optimizing statistical feature variables are based on a three-dimensional approach by observing the volume of the data. The diagnosis accuracy relies on the quantity and quality of the input data. Random features used for input data can cause increase in learning time as well as decreasing accuracy. The proposed method optimized appropriate combination of indicators based on skewness, kurtosis, mean, etc. The results of diagnosing fault conditions of spindle bearings on Roll-to-Roll system by optimizing statistical feature variables improved the classification performance to 99.3% of accuracy.