MSCNASTRAN振动疲劳分析振动疲劳.ppt

What is a PSD? PSD’s are effectively obtained by taking the modulus squared of the FFT. Some normalization has to take place (in this case 1/2T) but this can differ depending on the FFT algorithm used. Frequency domain techniques were pioneered by the electronics industry in the early 1940’s. The method was employed to investigate the cause of electrical noise in circuits. As power was generally the parameter being measured, it found its way into the title of the spectrum. The mean square amplitude of the sinusoidal waves is used because this was easily obtained by the early analog circuits of the day, it would be more difficult to derive the amplitude spectra and so this step was never carried out. Today, logic would suggest that we use the ‘Amplitude Spectral Density (ASD)’, after all this is more intuitive than the ‘Power Spectral Density (PSD)’ and we no longer rely on analog computers, however, many design standards refer to the PSD and for this reason it is still the preferred option. Moments from a PSD Since we are concerned with structural systems analyzed in the frequency domain we require a method for extracting the pdf of rainflow ranges directly from the PSD of stress. The characteristics of the PSD which are used to obtain this information are the nth moments of the PSD function. The relevant spectral moments are easily computed from a one sided PSD G(f) in units of Hertz using the following expression. The nth moment of area of the PSD (mn) is calculated by dividing the curve into small stripes as shown in the slide. The nth moment of area of the strip is given by the area of the strip multiplied by the frequency raised to the power n. The nth moment of area of the PSD is then found by summing the moments of all the strips. In theory all possible moments are required to fully characterize the original process. However, in practice we find that m0, m1, m2 and m4 are sufficient to compute all of the information required for the subsequent fatig