An approach, based on a global averaging procedure, is presented for estimating the power spectrum of a second order stationary zero-mean Ergodic stochastic process from a finite length record. This estimate is derived by smoothing, with a cubic smoothing spline, the naive estimate of the spectrum obtained by applying FFT techniques to the raw data. By means of digital computer simulated results, a comparison is made between the featurcs of the present approach and those of more classical techniques of spectral estimation.
展开▼