One-Cycle Windowed-DFT Harmonic Estimation with Spectral-Interference Compensation.

Accurate harmonic estimation at the per-cycle timescale is increasingly required in modern power-quality (PQ) monitoring, where fast-varying distortion sources demand high temporal resolution. However, when harmonic phasors are estimated from a single cycle using windowed discrete Fourier transform techniques, off-nominal fundamental frequency introduces spectral interference between harmonics, leading to systematic amplitude and phase errors that conventional correction methods cannot remove. This paper presents a lightweight, non-iterative harmonic estimation module designed to operate on fixed-rate, one-cycle data streams. The method leverages a frequency estimate provided by an external tracker to explicitly model the spectral interference induced by windowing under off-nominal conditions. By formulating this effect as a linear mixing process, the proposed approach applies an algebraic inversion to recover unbiased harmonic phasors without requiring adaptive resampling, variable window lengths, or modifications to the acquisition system. The module is designed as a plugin component compatible with existing PQ processing chains and shared sampled-value architectures. Experimental validation across frequency sweeps, Monte Carlo noise trials, and dynamic streaming scenarios demonstrates machine-precision accuracy in ideal conditions and noise-limited performance in realistic settings. Compared to iterative alternatives, the proposed solution achieves equivalent accuracy with a 484× reduction in computation time. A sensitivity analysis further quantifies the relationship between frequency-tracking accuracy and harmonic estimation error, providing practical guidelines for system integration. These results show that accurate, real-time harmonic estimation can be achieved from single-cycle data using fixed-rate acquisition, enabling improved monitoring and protection capabilities in modern power systems.