Non-linear least squares methodology for suppressing the fundamental frequency in the analysis of electric signals

Energy from the fundamental frequency component (FFC) in electrical signals is usually much higher than the energy from the rest of spectral components. This situation makes it difficult to obtain a proper analysis of the whole frequencies involved in a particular signal. To improve the spectral analysis of electrical signals, some works have proposed the use of filters and digital signal processing techniques for suppressing the influence of the FFC. However, these methodologies suppress a frequency band, introducing undesired effects in the frequencies close to the FFC. Thus, this paper proposes the use of non-linear least squares to identify the amplitude, frequency and phase that describe the FFC to suppress only one frequency component instead of a frequency band. The methodology is tested using synthetic signals and experimentation is performed with real electric signals from three different scenarios: current signals from two induction motors operating at two different frequencies (60 Hz and 31 Hz), and a voltage signal from a photovoltaic generation plant. Results show that the methodology can adequately recognize and subtract the FFC. This methodology aims to be a tool to enhance the results delivered by methodologies for condition monitoring of induction motors and power quality assessment.