Open AccessNONCORRALATED DATA ORDERED SAMPLES AS A SINGLE-ELEMENT MULTICHANNEL CONVERTER
Author(s) -
R.O. Mazmanian
Publication year - 2021
Publication title -
tehnìčna elektrodinamìka
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.22
H-Index - 13
eISSN - 2218-1903
pISSN - 1607-7970
DOI - 10.15407/techned2021.03.074
Subject(s) - closeness , statistical hypothesis testing , statistics , mathematics , sample size determination , statistical power , uncorrelated , normal distribution , probability density function , sample (material) , mathematical analysis , physics , thermodynamics
The sufficient closeness of the medians of the ordered samples of random data to the normal distribution is used in computer systems for control, monitoring and diagnosing electric power equipment. However, it remains what other probability density function (pdf) of elements (sample statistics) have such similarities. This paper presents various methods for statistical testing hypotheses for pdf-converter channels as statistics of given sizes odd numbered and ordered samples of bounded and uniformly distributed random numbers. The use of various different criteria and the results of estimates studied under the same conditions showed a sufficient conformity of the results of tests for three statistical criteria. It made possible to draw a reasonable conclusion about the preferable use of the adapted chi-square test for assessing the congruence of analytical pdf channels of the converter with normal distribution. We also suggested using the "statistical closeness window" to define those channels of the converter that do not significantly differ from the normal distribution. In addition, we presented an empirical formula determining the dependence of the size of the window of the statistical closeness window on the sample size. The results of the research are summarized in a statistical model of a multichannel uncorrelated data converter. References 27, figures 7.