Variable Frequency Room Air Conditioner Operation Pattern Recognition and Data Mining

Abstract

： The long running data of the variable frequency room air conditioners is the most direct reflection of their operation state in actual surroundings. Since the inherent law of air conditioners is hidden behind the long term operation data with a great quantity of redundancy and outlier data, data mining method is needed to solve this problem. The clustering algorithm is used for the analysis of air conditioners’ operation modes. Three modes are concluded ： morning high load mode, afternoon high load mode and stable low frequency mode. The temperature of air supply in the three modes is relatively similar, but the outdoor temperature is different. In the morning high load or afternoon high load mode, the outdoor temperature peaks in the morning and afternoon, meanwhile the energy consumption reaches the maximum and the energy efficiency ratio (EER) reaches the minimum respectively. In the stable low frequency mode, the outdoor temperature is not high, the operation is relatively stable after starting, and the operating parameters are relatively stable. Since then, the controlling variables method and polynomial fitting method are applied to analyze the relationships between the operation parameters and the performance of the air conditioners. It shows that air conditioner energy consumption