Application Of Higher-Order Spectra To The Analysis Of Pulse Signals

Traditional Chinese medicine has received world's wander for its unique diagnostic methods and particularly curative effects. Along with the development of science and technology, people hope to apply biomedical signal processing technology to human pulse diagnosis,in order to reveal the essence and features of pulse's phenomena scientifically and objectively,which is the main research aspect in this paper.Higher-order statistics is an advancing research aspect in signal processing field . HOS are the higher-order statistical description of a signal. HOS preserve the magnitude information as well as the phase information. The two main motivations behind the use of HOS are extracting information due to deviations from Gaussianity and detecting nonlinear properties in signals. HOS can be used to suppress additive colored noise in theory. Bispectrum is the most common method of Higher-Order Spectra mathematical analysis,and it includes two chief approaches that can be used to estimate higher-order spectra,namely the conventional and the parametric approach.In this paper,the direct algorithm of conventional bispectrum estimator and parametric approach bispectrum estimator the third-order recursion(TOR) to identify heroin addicts from the pulse signals of 15 heroin addicts and 22 healthy persons. Direct class of conventional bispectrum estimators is proposed for calculating the bispectrum of pulse signals. Characteristic parameters of magnitude bispectrum of the pulse signals are obtained by using diagonal slices of magnitude bispectrum.Extract feature via the principal component analysis, then accomplish classification using BP neural network.Thus Only one heroin addict b14 and healthy persons z14,z22 is misjudged. Third-order recursion method of non-Gaussian AR model is proposed for calculating the bispectrum of pulse signals for 15 heroin addicts and 22 healthy persons. Characteristic parameters of magnitude bispectrum of the pulse signals are obtained by using diagonal slices of magnitude bispectrum.Extract feature via the principal component analysis, then accomplish classification using BP neural network. Thus, all of the 22 healthy persons are identified. Only one heroin addicts b14 is misjudged.Exactness ratios reach 97.3%. The research result shows that bispectrum estimation for analyzing pulse signals of heroin addicts and healthy persons is really an effective method.Besides the two methods of bispectrum estimators in detail mentioned above, HOS are deduced and proved in this paper. Meanwhile, the theory of principal component analysis and bp neural networ is also concluded .