Mutation-Adaptive Mean Variance Mapping Optimization for Low Voltage-Ride Through Enhancement in DFIG Wind Farms

The widespread integration of wind energy conversion systems has fundamentally reshaped modern power grid architecture. However, the limited dynamic response of wind turbine (WT) converters during grid faults—particularly their inability to provide sufficient reactive current and maintain voltage stability under severe dips—necessitates a redefinition of the conventional low-voltage ride-through (LVRT) curve. This study addresses this challenge by proposing a Mutation-Adaptive Mean Variance Mapping Optimization (A-MVMO) algorithm for the control of grid-side converters (GSCs) in wind farms (WFs). To systematically assess post-fault voltage recovery, a Time-Segmented Analysis for Voltage Recovery (T-SAVR) approach is developed with a multi-objective function. The performance of the proposed A-MVMO is benchmarked against standard MVMO and conventional particle swarm optimization (PSO) under both moderate (0.7 pu) and severe (0.15 pu) voltage dips using the IEEE 39-bus system implemented in DIgSILENT/PowerFactory. The results demonstrate that A-MVMO achieves fast, oscillation-free voltage recovery with negligible overshoot (<1%) and lower current injection than PSO and MVMO, while satisfying all engineering constraints. Moreover, the co-optimization of Park-level and turbine-level controllers ensures seamless coordination, as evidenced by the close tracking between the farm-wide reactive power reference and the aggregated turbine response. The T-SAVR method proves essential for focusing optimization on controllable recovery dynamics, yielding a superior LVRT curve.