수산물 보관 창고용 냉동 압축기의 전력부하 예측 AI알고리즘 개발

Abstract

In this study, we use the monitoring information of the 214.5RT ammonia screw refrigerationsystem to show that it is possible to analyze operation patterns based power load usage, predictpower usage, and save energy. We further propose an algorithm to this end. Accidents occurfrequently, such as the leaking of ammonia in a recently dilapidated freezer warehouse, andmanagement is an important point of focus because most of the power load in freezerwarehouses is drawn by refrigerators. It is important to save energy and predict accidents byusing the temperature information of the Meteorological Agency and taking into considerationthe pressure and temperature of the refrigerator. In particular, the pressure of the refrigeratorchanges rapidly according to seasonal changes, and this information could be used to reduceaccidents. In addition, as interest in energy saving is increasing worldwide and the reduction ofcarbon dioxide emissions comes into focus, research on energy saving is being conducted invarious fields. In this paper, a prediction algorithm is implemented by analyzing data from thepast two years among the monitoring information of a large-scale animal center refrigerationsystem in Incheon area using deep learning, a type of AI algorithm that has recently attractedattention, as a prediction technique. Among the deep-learning models, an algorithm is designedusing RNN (Recurrent Neural Network) and LSTM (Long Short-Term Memory). The RNN andLSTM models have strengths in continuous data pattern analysis. Through this model, data arepredicted and performance evaluations are conducted by comparing the power load ofrefrigerators that have been actually consumed recently, considering the current limitingsituation and the direction to move forward