http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
GHP 운전시 COV에 의한 정상상태 판별 및 이상검출 방법 연구
신영기(Younggy Shin),오세재(Se-Jae Oh),정진희(Jin_Hee Jeong) 대한설비공학회 2011 설비공학 논문집 Vol.23 No.11
Fault detection has to be proceeded by steady state filtering to get rid of transient effect associated with thermal capacity. Coefficient of variance (COV), ratio of standard deviation devided by moving average, was employed as steady-state filter. Engine speed and refrigerant pressures were selected as parameters representing system dynamics. The filtered values were registered as members of steady-state DB. They were found to show good functional relationship with ambient temperature. The relationship was fitted with a second order polynomial and the distribution bounds of the data around the fitted curve were expressed by visual inspection because of varying average and random data interval. Fault data were compared with the steady-state data obtained during normal operation. The fault data were easily isolated from the fault-free one. To make such isolation reliable, tests to construct good DB should be designed in a systematic way.
서정아(Jeong-Ah Seo),신영기(Younggy Shin),오세재(Se-Jae Oh),차우호(Woo-Ho Cha),정진희(Jin_Hee Jeong) 대한설비공학회 2011 대한설비공학회 학술발표대회논문집 Vol.2011 No.7
Fault detection has to be proceeded by steady state filtering to get rid of transient effect associated with thermal capacity. Coefficient of variance (COV), ratio of standard deviation devided by moving average, was employed as steady-state filter. Engine speed and refrigerant pressures were selected as parameters representing system dynamics. The filtered values were registered as members of steady-state DB. They were found to show good functional relationship with ambient temperature. The relationship was fitted with a second order polynomial and the distribution bounds of the data around the fitted curve were expressed in terms of confidence intervals which were calculated from the assumption of Student's t-distribution. Fault data were compared with the steady-state data obtained during normal operation. The fault data were easily isolated from the normal one. To make the fault-detection algorithm sense, more fault cases have to be tested.