Adaptative and Structured Optimization Stockwell Transform for Robust Power-Quality Disturbances Detection

The monitoring of electrical power quality requires advanced analysis techniques capable of accurately detecting and localizing non-stationary disturbances. This work proposes a structured and adaptive optimization of the Stockwell Transform (ST), referred to as Two-Power Optimization (TPO). A multiparametric Gaussian window governed by four parameters (m, p, k, r), is introduced, and a genetic algorithm is applied independently to each signal to optimize these parameters. The optimization process adapts the analysis scale to the local characteristics of the signal while preserving the physical consistency of the window function. Simulation results on synthetic power quality signals demonstrate that the proposed method produces sharper and more stable spectral peaks, higher energy concentration, and improved time–frequency compactness compared to the standard ST and single-parameter variants. Furthermore, the TPO better preserves low-frequency modulation patterns, enabling more accurate detection and localization of electrical disturbances and enhancing the robustness of monitoring in modern power systems.