Fault Diagnosis of Water Chiller Based on MSSMOTE-CNN Model
To deal with the unbalanced data types during the operation of water chillers,this paper proposes the MSSMOTE method based on Mahalanobis distance and"triangle"area interpolation to expand the fault data,and input the obtained data into the CNN model for training,so as to realize the diagnosis of seven kinds of faults in water chillers.Simulation tests were conducted under different expansion ratios and the same data type.The results showed that when the expansion ratio was 4,the MSSMOTE-CNN model achieved an accuracy of 0.961 and a F1-score of 0.971 respectively for normal sample testing,which was capable of accurately identifying the fault type of the chiller.
MSSMOTE-CNN modeldata imbalancefault diagnosiswater chilling unit
万方数据
维普
