Fiber-optic tube classification based on weighted least squares method

ZHANG De-chao， SHI Wei-min， YANG Liang-liang， LIU Yi-sheng
（School of Machinery and Automatic Control， Zhejiang Sci-Tech University， Hangzhou 310018， China）

Abstract： Aiming at the precise measurement of the optical tube diameter and rapid classification，the principle that the friction between the fiber-optic tube aperture and standard stick varies linearly was investigated. After the analysis of the actual friction data collected between the different kinds of standard fiber-optic tube and standard detection stick，the standard friction model of each kind of the fiber tube aperture with the weighted least squares method was gotten，and the friction model database was achieved by the calculation of the characteristic of the standard friction model，the relation between all kinds of standard fiber-optic tube and the friction model database was established. A classification method was presented to compare the friction characteristics of actual data collected with the friction model database. The method was evaluated on the DSP，the classification was tested. The experimental results indicate that the classification method is feasible，it can detect 18 fiber tubes per minute with accuracy of up to 99%，and improves the accuracy and speed of detection.
Key words：fiber tube classification； classification processing； weighted least squares method