Accession Number : AD0716600

Title :   The Classification of Engine Vibration Records Through Quasi-Singular Detection.

Descriptive Note : Technical rept.,

Corporate Author : PRINCETON UNIV N J DEPT OF ELECTRICAL ENGINEERING

Personal Author(s) : Fang,Geng-Seng ; Pavlidis,Theodosios

Report Date : DEC 1970

Pagination or Media Count : 36

Abstract : The report deals with the classification of signals in term of their autocorrelation functions. Such an approach is necessary for the classification of vibration records from reciprocating internal combustion engines because the mean of such records contains no significant information. For each signal the autocorrelation is calculated for a narrow 'window'. The eigenfunctions of ach class are determined and an unknown signal is classified by comparing the norms of its projections to subsets of such eigenfunctions. By working with eigenfunctions corresponding to the smallest eigenvalues the method approximates singular detection. (Author)

Descriptors :   (*INTERNAL COMBUSTION ENGINE NOISE, CLASSIFICATION), (*INTERNAL COMBUSTION ENGINES, VIBRATION), (*PATTERN RECOGNITION, *INFORMATION THEORY), STOCHASTIC PROCESSES, OPERATORS(MATHEMATICS), SET THEORY, HILBERT SPACE, INEQUALITIES, DECISION THEORY, DISTRIBUTION FUNCTIONS, ERRORS

Subject Categories : Cybernetics
      Acoustics
      Reciprocating and Rotating Engines

Distribution Statement : APPROVED FOR PUBLIC RELEASE