Coordinated optimization of DFIG control parameters integrating grid similarity index screening and neural networks

This paper presents a coordinated optimization framework for control parameters in power systems integrated with doubly fed induction generator using machine learning algorithms. A sample screening method based on the similarity index of power system samples is developed with pattern recognition techniques and back propagation neural networks, ensuring robustness under varying noise levels. Incorporating sensitivity analysis, an optimization model is formulated to minimize parameter variations while maintaining feasibility across all operating scenarios. Case studies show that the proposed method reduces unstable operating modes under small-signal disturbances by approximately 98 \(\%\) . For the remaining unstable cases, active power regulation is introduced to eliminate weak damping and further enhance small-signal stability.