Abstract:The complex signal modulation characteristics brought by power quality disturbance make it difficult to extract the disturbance signal characteristics from the power quality monitoring data. In order to improve the detection accuracy of composite power quality disturbance under noise, the Adaptive Parameterless Empirical Wavelet Transform (APEWT) is used to decompose the disturbance signal into IMFs, which are demodulated based on the Frequency-Weighted Energy Operator(FWEO) respectively, and then the instantaneous frequency and amplitude feature quantities for disturbance identification are accurately obtained. On the one hand, in APEWT the output of the orthogonal filter bank based on adaptive band segmentation only contains meaningful mode mono-component, which effectively avoids the phenomenon of mode mixing. On the other hand, FWEO noise robustness effectively improves the accuracy of disturbance feature extraction in strong noise background. Furthermore, applying the proposed algorithms to simulated and measured signals to verify performance, the results show that the method can effectively track the instantaneous change of the disturbance signal, and the instantaneous frequency and amplitude obtained by demodulation further prove the feasibility and effectiveness of the method.