A Measuring Method for Angular Displacement Based on Correlation Algorithm

Abstract

The existing angular displacement measurement methods rely on the manufacturing precision of the fixedplate and moving-plate with high manufacturing cost, and it is difficult to overcome a static error and drift for the static measurement method. A kind of angular displacement measuring method based on correlation algorithm is presented with the characteristics of a low manufacturing cost, high precision, anti-noise and anti-partially damaged properties and so on. The method is that firstly full circle broadband or white noise, random data is a pre-prepared coaxially in a rotating body, then the periodic random signal is continuously formed using pickup head reads this data in the basic uniform rotation process of the rotating body. The instantaneous angular displacement of the pickup head relative to the rotating body is obtained by means of the correlation operations between the periodic random signal and the signal sequence of pre-stored data. The angular displacement among the different pickup heads is gained by subtraction for the instantaneous angular displacement of different pickup heads relative to the rotating body at the same instant. The functional relationship between the relative angular displacement of the different pickup heads and the measured angular displacement is determined by linkage equations, which is used to calculate the measured angular displacement. Furthermore, the schematics of detective devices and the principle of the signal processing are developed to implement the method.

Case Retrieval Algorithm Based on Structure Matrix

“…To avoid enormous amount of calculation by the inversion of the matrix, we can use an iterative approach to solve characteristics similar matrix of S j in the former equation [11] . We proposed a recursive form of solving the characteristic similarity matrix S j :…”

“…Using structure matrix, with each case as a test case, to find whom the most similar to the k (k were taken10, 20, 30, etc. ) cases, and by judging the case and manual annotation of the most similar cases comparison to evaluate the accuracy of the similarity search [11] . Specifically, we use the following formula for the evaluation of the results:…”