A comparison of power spectral estimates and applications of the maximum entropy method

摘要: A new spectral estimate, called the maximum entropy method, is described. This estimate was originated by John Parker Burg for use in seismic wave analysis. In method entropy, or information, of a signal maximized under constraint that estimated autocorrelation function Fourier transform power density. The estimates are calculated two ways: (1) minimization error to obtain coefficients prediction filter, as suggested Burg, and (2) direct solution matrix equation using an algorithm due Norman Levinson. For comparison Blackman-Tukey technique, with Hamming window, used also. We illustrate these three methods applying them composite consisting four sinusoids unit amplitude: one each at high low frequencies moderate respect Nyquist frequency, which added white noise 0.5 amplitude. Results shown indicate best correspondence input spectrum provided technique. Applications geomagnetic micropulsations reveal complex multiplet structure Pc 4, 5 range. Such structure, not previously resolved conventional techniques, has been predicted recent theory magnetospheric resonances. period range 7 orders magnitude longer than micropulsation periods, analysis annual sunspot means shows 11-yr band composed least distinct lines. With lines associated harmonic sequence. Long periods order 100 yr also revealed.