SOME PROBLEMS IN EVALUATION OF ACCURACY OF CORRELATION FUNCTIONS (Nekotorye Voprosy - Otsenki Tochnosti Korrelyatsionnykh Funktsii),

Abstract

Further improvement in the methods of calculating correlation functions, with the source information in discrete form, should go along these two directions: (1) Development of methods for evaluating the error depending on the number of points of source information; with knowledge of the process frequency spectrum available, the highest-frequency region should contain 10-40 points; (2) With a sufficiently large number of source-information points, the accuracy of calculation of correlation function should be determined. The present article places particular emphasis on the case of a relatively small number of the source-information points. A formula for the centered autocorrelation function indicates two sources of errors: (a) level quantization and (b) determination of mathematical expectation. If the quantization interval is q = 1/31 maximum level, the quantization error is negligible. For practical purposes, it is recommended that estimation of the mathematical expectation be restricted to the second decimal place. With the above provisions, the correlation functions of interdependent parameters (30-70 points) of the rolling process at the Pervoural'sk Pipe Plant were estimated on a DK-1 digital correlograph developed by the Institute of Electronics, Automatics and Telemechanics, AN GruzSSR. (Author)

Document Details

Document Type
Technical Report
Publication Date
Oct 13, 1967
Accession Number
AD0689259

Entities

People

  • E. A. Saakyan
  • E. V. Melikhova
  • V. V. Imedadze

Organizations

  • National Air and Space Intelligence Center

Tags

DTIC Thesaurus Topics

  • Accuracy
  • Autocorrelation
  • Automatic
  • Control Systems
  • Correlation Techniques
  • Data Science
  • Electronics
  • Errors
  • Frequency
  • Information Science
  • Intervals
  • Spectra
  • Test And Evaluation

Readers

  • Computer Programming and Software Development.
  • Information Retrieval
  • Regression Analysis.

Technology Areas

  • Microelectronics