A Hybrid Computing Adaptive Filtering Methods for Parameter Estimation of Nonstationary Power Signals

This paper presents a new approach in the detection, localization, and classification of frequency and amplitude changes in nonstationary signal waveforms using a variable window shorttime Fourier Transform (STFT) known as ST in short and an Extended Complex Kalman Filter (CEKF). Unlike the fixed window STFT, the variable window Short-time Fourier Transform has excellent time-frequency resolution characteristics and provides detection, localization, and visual patterns suitable for automatic recognition of time-varying signal patterns. The CEKF, on the other hand, provides automatic classification and measurements of the frequent amplitude, and phase of sinusoids embedded in noise. The technique is applied to both simulated and experimentally obtained waveform disturbances in the presence of additive noise and the results reveal significant accuracy in completely localizing the changes in amplitude, frequency, and phase of nonstationary sinusoids in noise.