Outliers Identification From Instantaneous Angular Speed Signal Generated by Zebra Tapes

Xiaoyang Bi,Jiansheng Lin,Youxi Wu,Teng Ma,Xiao Yang,Xiaobo Bi,Fengrong Bi

doi:10.1109/access.2019.2928598

Xiaoyang Bi, Jiansheng Lin + Show 5 more

Open Access

https://doi.org/10.1109/access.2019.2928598

Copy DOI

Journal: IEEE Access	Publication Date: Jan 1, 2019
Citations: 14	License type: CC BY 4.0

Affiliation: Tianjin University, Hebei University of Technology

Abstract

Instantaneous angular speed (IAS) measurement using zebra tape has been presented as an advantageous tool for rotating machinery surveillance. Unfortunately, both butt joint and mounting imperfection of the zebra tape result in stripes with different widths, which cause the outliers appear in the acquired signal. This paper proposed a solution to accurately identify the outliers and define the angular displacement values represented by outliers in the condition of periodic fluctuation of speed. The procedure is based on K-means clustering method and Fourier series model. Specially, a series of algorithms take K-means clustering partitioning method as core, and is used for outliers identification; the Fourier series fitting model is used to analyze the angular displacement values represented by outliers. Experimental study was performed on a test rig to validate the method under different measurement conditions. The result of error analysis shows the effectiveness of the algorithm.

Highlights

Rotating machinery is important equipment in the industry of petroleum [1], chemical, metallurgy, electric power [2] and machinery [3]
The number of sampling data that when each black and white group passing through the senor was exacted from raw sampling data, and the exacted data series are fitted with the Fourier Series model and the distribution of response errors are analyzed, according to the characteristics of error distribution, the K-means clustering partitioning method is used for outliers identification; the Fourier Series fitting Model was used again to analyze the angular displacement values corresponding to outliers, the difference is that the outliers have been removed from the fitting data sequence based on the previous outliers identified step
Practical tests were performed on a test rig to validate the method under different measurement conditions, the result of error analysis shows the effectiveness of the algorithm

Summary

INTRODUCTION

Rotating machinery is important equipment in the industry of petroleum [1], chemical, metallurgy, electric power [2] and machinery [3]. The signal generated by angular speed sensor represent the relationship between angular displacement and corresponding time duration. The measurement method involved in this paper is based on analog-to-digital conversion (ADC), using photoelectronic sensor and zebra tape with equal-width dark and light stripes for signal acquisition. The number of sampling data that when each black and white group passing through the senor was exacted from raw sampling data, and the exacted data series are fitted with the Fourier Series model and the distribution of response errors are analyzed, according to the characteristics of error distribution, the K-means clustering partitioning method is used for outliers identification; the Fourier Series fitting Model was used again to analyze the angular displacement values corresponding to outliers, the difference is that the outliers have been removed from the fitting data sequence based on the previous outliers identified step. The signal generated by speed sensor contains the information about angular displacement and corresponding time duration. If the stripe groups of zebra tape have same width, IAS ω can be expressed as follows: ω=

IAS MEASUREMENT ERROR ANALYSIS

OUTLIERS OF ZEBRA TAPE IDENTIFICATION

ANALYSIS OF ANGULAR DISPLACEMENT REPRESENTED BY OUTLIERS

EXPERIMENTAL STUDY

CONCLUSION