딥러닝을 이용한 히트펌프 시스템의 냉매량 고장 감지에 관한 연구
엄용환, 류진우, 홍성빈 et al. · 대한설비공학회 2018년도 하계학술발표대회 · 2018
Refrigerant leakage is one of the frequent faults in heat pump systems. Since optimal refrigerant charge influences the performance of heat pump system, it is very important to detect refrigerant charge fault. In recent years, the data-driven methods such as artificial neural network(ANN) have attracted more attention because of big data and an improved computing speed. However, the conventional ANN based refrigerant charge fault detection(RCFD) methods need the features extracted by experts' knowledge, and the feature extraction is not easy. Hence, a new RCFD approach based on convolutional neural network(CNN) which has an effective way to extract features automatically was proposed in this study. Also, this research suggested a quantitative prediction model using classification method. The CNN based RCFD model has an accuracy of 99.3% in classifying the trained refrigerant charge amount.
Cited by 0 works.