Fault Diagnosis of Refrigeration Equipment Based on Data Mining and Information Fusion

Traditional monitoring system of refrigeration equipment is accurate in detection single fault, but difficult to judge concurrent faults. In light of this limitation, a data mining method is proposed based on the information fusion method combing the designated cell analysis and support vector machine with weighted evidence theory. First, a non-fully orthogonal designated cell analysis method is proposed to make up the limit of traditional designated cell analysis in the non-fully orthogonal mode. Then, experiments prove that both the non-fully orthogonal designated cell analysis and support vector machines models can identify concurrent faults, and each model has certain advantages in identification of different concurrent fault. Finally, the weighted evidence theory is used to synthesize the diagnostic results of the two models. The hit rate of the post-fusion diagnosis raised to 99.10%, and the false alarm rate is reduced to 0.21%. © 2021, Editorial Department of JVMD. All right reserved.