Harmonics filtering and detection of disturbances using wavelets

Traditional mathematical tools, like Fourier analysis, have proven to be efficient when analyzing steady-state distortions; however, the growing utilization of electronically controlled loads and the generation of a new dynamic in industrial environments signals have suggested the need of a powerful tool to perform the analysis of nonstationary distortions, overcoming limitations of frequency techniques. Wavelet theory provides a new approach to harmonic analysis, focusing the decomposition of a signal into nonsinusoidal components, which are translated and scaled in time, generating a time-frequency basis. The correct choice of the waveshape to be used in decomposition is very important and discussed in this work, a brief theoretical introduction on wavelet transform is presented and some cases (practical and simulated) are discussed. Distortions commonly found in industrial environments, such as the current waveform of a switched-mode power supply and the input phase voltage waveform of motor fed by an inverter are analyzed using wavelet theory. Applications such as extracting the fundamental frequency of a nonsinusoidal current signal, or using the ability of compact representation to detect nonrepetitive disturbances are presented