Fault diagnosis method for chiller

Abstract

The invention discloses a fault diagnosis method for chiller. The method comprises the steps of: S1, performing fault simulation experiments on a screw chiller and a centrifugal chiller, and collecting fault simulation experiment data of the screw chiller and the centrifugal chiller; S2, randomly selecting the fault simulation experiment data of the screw chiller and the centrifugal chiller on which a steady state processing and a noise removing processing are performed to form a second training set data and test set data; S3, establishing original training set data; S4, performing oversampling processing on the original training set data to obtain oversampled training sample set data with different oversampling rates; S5, training the oversampled training sample set data using an SVM model to obtain a diagnostic model; and S6, putting the test set into the diagnostic model for testing to obtain diagnosis results of the different chiller fault models. The diagnosis method can perform fault diagnosis on different types of chillers, and has wide application prospect.

Description

technical field [0001] The invention belongs to the refrigeration field, and in particular relates to a method for diagnosing a chiller fault based on a PCA-SMOTE-SVM algorithm. Background technique [0002] In the existing chiller fault diagnosis research, the chiller is in one-to-one correspondence with its fault diagnosis model, and one fault model can only diagnose one chiller. Simulation experiments and training new diagnostic models consume a lot of resources and time, which is not conducive to the promotion of chiller fault diagnosis technology in practical applications. [0003] When there is an efficient fault diagnosis model for a certain type of chiller, and fewer experimental samples of other types of chiller faults are obtained, when the new samples are mixed with the original samples to form a new training sample set, the training set of the original model The number of fault sample sets differs greatly from that of the new chiller, forming an unbalanced sampl...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of unbalanced sample set, the present invention adopts synthetic minority class oversampling techn

Images