Fault Diagnosis Based on Novel Hybrid Intelligent Model

Abstract

Listen

Translate article

In order to solve the difficult problem of identifying incipient fault and compound fault for mechanical equipment,and improve diagnosis accuracy,a novel hybrid intelligent diagnosis model based on multiple feature sets from different symptom domains and multiple classifier combination,is proposed.This model combines empirical mode decomposition(EMD),the improved distance evaluation technique,adaptive neuro-fuzzy inference system(ANTIS)and genetic algorithm(GA)techniques etc.Prior to feature extraction,several signal preprocessing techniques,i.e.filtration,EMD and demodulation etc.,are employed to excavate the underlying fault information from dynamic signals.Time-domain and frequency-domain statistical features that reflect the equipment operation conditions from various aspects are extracted and six feature sets are obtained.In succession,the improved distance evaluation technique is proposed,and with it,six sensitive feature sets are selected from the six original feature sets,respectively. The six sensitive feature sets are input into the multiple ANFISs combined by GA to attain the final diagnosis result.The application to fault diagnosis of locomotive wheel pair bearings shows the model is able to reliably recognize not only different fault categories and severities but also the compound faults.Thus,a desired diagnosis effect is obtained via the hybrid model.Moreover,the application result also validates the power of the proposed feature selection method based on the improved distance evaluation technique.