Digital technology for spectral analysis of oscillation parameters

Abstract

Performing the Fourier analysis of the harmonic composition of oscillations is an effective approach to the study of vibration processes in machines and mechanisms. This method of express control remains relevant and highly demanded in the acoustic diagnostics of machines due to its relatively high informativeness. The expansion of its analytical capabilities is implemented by introducing additional means for digital processing of vibration signals, including smart systems. The objective of this work consists in creating a digital method for spectral analysis of vibration signals, the implementation of which would be possible with the use of smart devices. This work presents a method of polyharmonic approximation of the parameters of the signals reflecting oscillatory processes in machines based on the additive combinations of sinusoidal functions. The technology of truncation of the time discretised decomposition of the above signals into a Fourier series is implemented using the Rayleigh theorem and the Lyapunov – Parseval equality. The distortion of the frequency spectrum of oscillations is eliminated by suppressing its side lobes using window filters. To compare the efficiency of different windows, the following special parameters are used: width of the main lobe of the frequency spectrum; coefficient of relative expansion of the main lobes; maximum relative level of side lobes; coefficient of relative attenuation of signal energy. In so doing, the windows of Hanning, Nuttall, Hamming, Barlett were considered, and the parameters of a rectangular window were taken as basic values. The limitations of the digit capacity of analogto-digital signal converters depending on the required accuracy of maintaining the energy balance when truncating the discrete Fourier expansions have been determined. An example of implementation of the digital technology of spectral analysis of the parameters of oscillatory processes is given.