DocumentCode
3404466
Title
Fifth-order cumulants based higher order coupling feature extraction algorithm
Author
Guo, Yeczli ; Zhao, Junwei
Author_Institution
Anhui Univ. of Sci. & Technol., Huainan, China
Volume
1
fYear
2004
fDate
31 Aug.-4 Sept. 2004
Firstpage
276
Abstract
Quadratic frequency coupling feature radiated by actual nonlinear vibration system can be extracted using third-order cumulants (TOC). Cubic frequency coupling feature and pairing frequency coupling one may be extracted using fourth order cumulants (FOC). Much higher order frequency coupling feature extraction isn´t studied now. In the paper, quadrtic frequency coupling and quadratic-to-cubic frequency coupling are firstly defined. The direct estimation formula of fifth-order cumulants (FFOC) is derived and two type different definitions of stationary complex random process FFOC are obtained. FFOC feature and relation between recursive estimation of FFOC and its computational load are analyzed. FFOC diagonal slice based adaptive line enhancer (FFOC-DSBALE) are developed. The FFOCDSBALE can recover or extract higher frequency coupling feature in Gaussian noise. Theoretic analysis and simulation result prove the validity of the proposed algorithm.
Keywords
Gaussian noise; acoustic signal processing; computational complexity; feature extraction; higher order statistics; optimisation; recursive estimation; Gaussian noise; cubic frequency coupling feature; diagonal slice based adaptive line enhancer; fifth-order cumulant; higher order coupling feature extraction algorithm; nonlinear vibration system; quadratic frequency coupling; quadratic-to-cubic frequency coupling; recursive estimation; stationary complex random process; Algorithm design and analysis; Analytical models; Computational modeling; Couplings; Feature extraction; Frequency; Gaussian noise; Line enhancers; Random processes; Recursive estimation;
fLanguage
English
Publisher
ieee
Conference_Titel
Signal Processing, 2004. Proceedings. ICSP '04. 2004 7th International Conference on
Print_ISBN
0-7803-8406-7
Type
conf
DOI
10.1109/ICOSP.2004.1452636
Filename
1452636
Link To Document