作者: T. Lobos , Z. Leonowicz , J. Rezmer , P. Schegner
关键词:
摘要: The spectrum-estimation methods based on the Fourier transform suffer from major problem of resolution. were developed and are mostly applied for periodic signals under assumption that only harmonics present periodicity intervals fixed, while in presence interharmonics variable very long. A novel approach to harmonic interharmonic analysis "subspace" is proposed. Min-norm music retrieval examples high-resolution eigenstructure-based methods. Their resolution theoretically independent signal-to-noise ratio (SNR). Prony method as parameter estimation signal components was also tested paper. Both do not show disadvantages traditional tools allow exact frequencies. To investigate methods, several experiments carried out using simulated signals, current waveforms at output an industrial frequency converter, during out-of-step operation a synchronous generator. For comparison, similar repeated fast (FFT). comparison proved superiority new