Unsupervised detection of rotary machine imbalance based on statistical signal properties. (15th March 2022)

• Record Type: Journal Article
• Authors: Jablonski, Adam; Bielecka, Marzena; Bielecki, Andrzej
• Is Part Of: Mechanical systems and signal processing. Volume 167:Part A(2022)
• Publication Date: 2022-03-15
• Subjects: Shaft imbalance -- Fault development -- Unsupervised fault detection -- Chi-square test for homogeneity; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2021.108497
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: Imbalance fault generates shaft harmonic components. Wideband statistical health indicators easily give contradictory result. Narrowband spectral analysis is effective, but it requires detailed configuration and extra PM signal. Chi-square test for homogeneity can detect multiple changes in vibration signal. Abstract: In the paper, a two-step method of unsupervised monitoring of rotary machine, based on statistical data processing, is proposed. The first step uses the chi-square non-parametric homogeneity test χ 2 in the form introduced by Snedecor for checking whether two signals, the monitored one and the referential one, have the same distribution. This first step is used to detect the first significant fault occurrence in the monitored machine. In the second step, the value of the statistics is utilized as the measure whose dynamics reflects changes in the nature of the incompatibilities of the given signal and the referential ones. The appearance of a new change manifests the occurrence of a new fault in the tested machine. In a such way, the second step allows to show further fault development. The method is evaluated using real signals recorded from a machine experiencing shaft imbalance development and it is compared with results of detailed analysis using classical supervised methods.