Possibilities of a Hybrid Method for a Time-Scale-Frequency Analysis in the Aspect of Identifying Surface Topography Irregularities

Abstract

The article presents research results related to assessing the possibilities of applying modern filtration methods to diagnosing measurement signals. The Fourier transformation does not always provide full information about the signal. It is, therefore, appropriate to complement the methodology with a modern multiscale method: the wavelet transformation. A hybrid combination of two algorithms results in revealing additional signal components, which are invisible in the spectrum in the case of using only the harmonic analysis. The tests performed using both simulated signals and the measured roundness profiles of rollers in rolling bearings proved the advantages of using a complex approach. A combination of the Fourier and wavelet transformations resulted in the possibility to identify the components of the signal, which directly translates into better diagnostics. The tests fill a research gap in terms of complex diagnostics and assessment of profiles, which is very important from the standpoint of the precision industry.

Multiscale Data Treatment in Additive Manufacturing

“…In this case, the application of a classic approach [54] often appears insufficient [55]. In recent years, the newly developed solutions, algorithms used in surface metrology in data processing, multiscale [5] or hybrid methods [31], make for an effective alternative that is increasingly more often used in 2D and 3D surface analysis. The new aspect of the research is the comprehensive assessment of the spherical surfaces manufactured additively with a simultaneous analysis of the influence of additional operations and measurement data processing.…”

“…The research has shown that the size of the edge effect created is not equal to the support width of a particular mother wavelet. An analysis of the effect of support width on filtering results was presented in the paper [31]. The research showed that the use of wavelets with a smaller support width results in better, more complex filtering in the initial stages of decomposition which consist of additional high-frequency information.…”

An In-Depth Study of Vibration Sensors for Condition Monitoring

“…Machine learning algorithms are trained to identify patterns in vibration data, enabling predictive maintenance by identifying prior indications of future issues [50]. In order to determine the frequencies of resonance or other anomalies, spectral analysis is also used for analyzing frequency content [51]. By creating virtual models of actual machinery, digital twins provide predictive maintenance, real-time monitoring, and simulation-based continuous improvement [52].…”

Research on the Roundness Approximation Search Algorithm of Si3N4 Ceramic Balls Based on Least Square and EMD Methods

“…Although the calculation method of spherical error has certain advantages, the results easily fall into the local extremum in the iterative process, which affects the calculation accuracy. Damian Gogolewski [ 22 , 23 ] and his team creatively proposed to evaluate the contour by combining the Fourier transform and wavelet transform. This method has high accuracy and reliability, but there is a lack of application examples for ceramic balls.…”