Feature extraction of time-varying power signals

This paper presents a wavelet based technique for monitoring and measuring nonstationary power system disturbances. A significant improvement in monitoring efficiency is achieved by processing signals through Kaiser's window. The maximum expansion coefficient extracted at each resolution level, the indices and sign of these coefficients at a super-resolution are used to monitor and measure the nonstationary behavior of signals. The proposed tool depends on the expansion coefficients and no reconstruction of these coefficients is required. The proposed monitoring technique is evaluated using large data sets of randomly variable magnitudes and frequencies.