This paper presents a novel hybrid AI method combining an H-filter and an adaptive linear neural network for flicker component estimation in power distribution systems. The proposed method leverages the robustness of the H-filter to extract the voltage envelope under uncertain and noisy conditions, and then accurately identifies the flicker frequency contained within the envelope using ADALINE. This synergy enables efficient time-domain estimation with fast convergence and noise resilience, addressing the key limitations of existing frequency-domain methods. Unlike existing techniques, this hybrid AI model handles complex power disturbances without prior knowledge of noise characteristics or extensive training. To validate the method's performance, simulation studies, statistical analysis, Monte Carlo simulations, and real-world data were conducted based on IEC standard 61000-4-15. The results demonstrate superior accuracy, robustness, and reduced computational overhead compared to estimators based on the fast Fourier transform and discrete wavelet transform.
