Data Mining from Haier Air-Conditioner Equipment Running Data for Fault Prediction

摘要

During commercial air-conditioner running, various equipment faults happen accidentally. Fault prediction of air-conditioners is crucial to reduce the incidence of faults, which lower the maintenance cost for users and manufacturers. However, detecting the cause of faults is a challenge because of air-conditioner running environment, complexity of parameters and interaction between equipment. In this paper, we propose data mining from Haier air-conditioner equipment running data for fault prediction. We calculate change rates of parameters to find unstable parameters which changing quickly before the fault occurrence, and also exploit dual association rules to mine association relationships among air-conditioner parameters. Relevant parameters are selected and their thresholds are calculated according to the underlying change rates for deducing the probability of a certain fault.