Pursuing optimal electric machines transient diagnosis: The adaptive slope transform

• The adaptive slope transform is defined, to analyze electric machines transient currents.
• It is based on the criterion proposed to obtain an optimal capture of the energy density.
• Improved t–f resolutions are obtained when capturing the current components.
• The results are compared with classic linear transforms.
• The entire evolution of the Lower Sideband Harmonic is captured during a startup and the subsequent load oscillations.

The aim of this paper is to introduce a new linear time-frequency transform to improve the detection of fault components in electric machines transient currents. Linear transforms are analysed from the perspective of the atoms used. A criterion to select the atoms at every point of the time-frequency plane is proposed, taking into account the characteristics of the searched component at each point. This criterion leads to the definition of the Adaptive Slope Transform, which enables a complete and optimal capture of the different components evolutions in a transient current. A comparison with conventional linear transforms (Short-Time Fourier Transform and Wavelet Transform) is carried out, showing their inherent limitations. The approach is tested with laboratory and field motors, and the Lower Sideband Harmonic is captured for the first time during an induction motor startup and subsequent load oscillations, accurately tracking its evolution.