Investigation and Application of the BOUAD Characteristics in Multidimensional Vibration’s Parametric Space of Drilling Pump

Abstract

A large amount of drilling pump vibration signals are acquired and decomposed optimally into time-frequency atoms by the Pattern Filter Method. All of these time-frequency atoms are falling into the multidimensional statistical cube which is formed by the time-frequency atomic parameter (α,f,φ,A) space with certain interval length. The atoms in every cube are counted and the results are stored in a database. When keeping any two parameters in a certain range, the quantitative distribution contour maps in the plane of other two parameters are demonstrated by the Graf software tool, then all sub-space results were aggregated to construct the macro-views of the BOU quantity distribution in the four-dimensional space ,meanwhile the quantitative distribution of these atoms are investigated, the distribution characteristics for various vibration signals of drilling pump are obtained and all atoms are classified according to their clustering state in this parametric space. Furthermore, every special signal corresponding to the parametric feature clustering of atoms in this multidimensional space have been identified. The desired results have been achieved in signal identification and drilling pump fault diagnosis by use of these consequences.