Leakage Error Compensation in Motor Current Signature Analysis for Shaft Misalignment Detection in Submersible Pumps

Excessive vibrations in motor operation can be a cause of premature failure in many industrial applications; therefore, effective solutions to accurately measure vibrations are required. A common approach to measure vibrations is through the motor current signature analysis, as torque oscillations produce characteristic frequency components in the current spectrum; this approach is suitable for those industrial applications where the rotating parts are not directly accessible for sensor installation, e.g., electrical submersible pumps which are usually deep underground or underwater. The detection of vibrations from the current spectrum is particularly challenging in high-power motors, where the ratio between the amplitude of the sideband spectral components caused by vibrations and the amplitude of the fundamental component is so small that the former is likely to be masked by the spectral leakage of the latter when a traditional Fourier analysis is employed. This article proposes a method to improve the sideband amplitude measurement by compensating for the leakage error in the current spectrum. The method is applied, in particular, to the detection of vibrations caused by a shaft angular misalignment, but it can be applied to other types of vibrations as well. The method is first illustrated through simulation results and then experimentally validated on a smaller-scale motor. Results confirm the ability of the proposed method to accurately measure the sideband amplitude even when it was originally completely hidden by the spectral leakage of the fundamental component.