Non-iterative Spectrum Correction to Close Components in Frequency
Zhang Qiang1, Zhang Pin2, Chen Kui-fu3
1.College of Civil Engineering, Shanghai Normal University, Shanghai 201418, China;2. College of Land Resources and Environment, JAU, Nanchang 330045, China 3. College of Science, China Agricultural University, Beijing 100083,China;
For a complex sinusoid model (CSM), there exists a set of explicit expressions for spectrum correction. However, it is not clear whether there exist corresponding formulas for double-frequency model (DFM). To determine feasibility of the explicit expressions to the DFM, the innate of ratio correction for CSM was analyzed. The explicit correction formulas were present for the DFM without windowing, and were examined by a DFM signal with one strong component amplitude 10 times the weaker one. The study shows, firstly, the essence of existence of simple correction expression for the CSM is that the spectrum functions of common windows can be factorized as transcendental part multiplying a rational fraction, and absolutes of the former are equal on the neighbor lines of discrete spectrum. Secondly, with the above properties, the frequency equations for DFM, only containing the rational fraction, can be deduced. Only for the case of the DFM without windowing, can the frequency equations be simplified to a quadratic polynomial, and will be implicated with at least cubic polynomial for other cases, such as Hanning window. In conclusion, it is not worthy or impossible to find the explicit correction expressions except for the DFM without windowing. The simulation results show that, the precision of the given correction expression can be achieved for the strong component better than the weaker one. In the frequency scanning 0.5~2.0 resolutions of the fast Fourier transform. But the CSM based correction is preferred if the frequency difference of DFM is greater than 2 canonical resolutions of FFT.