Application of Rank-order Morphological Filtering and Sample Entropy in Feature Extraction of Rotor Fault

摘要

After deeply analyzing the relation between reason and symptom of rotor fault, the sample entropy was introduced into the fault diagnosis field of rotating machinery. Combined with rank-order morphological filtering and sample entropy, a novel feature extraction method was proposed for rotor. Firstly, the line structure element was selected for rank-order morphological filter to de-noise the original signal. Secondly, the sample entropy of de-noised signal was calculated, the de-noised signal types were normal, unbalanced, misalignment, oil-film whirl and rubbing. Finally, the sample entropy was served as fault feature to evaluate the different fault condition. Practical results prove that the proposed method is effective on fault diagnosis of rotating machinery.