Enhancing synchronization stability: Adaptive PLL implementation using double internal loop RNN in type-3 wind power systems

Abstract

Doubly fed induction generators (DFIG)-based wind turbines (WTs) are prevalent in modern power grids. However, they often encounter synchronization issues when connected to weak grids. Synchronization instability primarily arises from controller dynamics, particularly during a phase-locked loop (PLL) interface with a voltage source converter. This study investigates the influence of PLL control gains on grid synchronization and the characteristics of power system oscillations induced under weak grid conditions. Based on this analysis, a novel adaptive double internal loop recurrent neural network (DILRNN) is employed to adjust the PLL gains dynamically under various system conditions. A dynamic back-propagation learning algorithm is utilized to fine-tune the weights of the DILRNN. Moreover, dynamic (adaptive) learning rates are selected using the Lyapunov stability method to ensure faster convergence and enhance the stability of the control scheme. The effectiveness of the proposed controller is validated under varying wind speeds, grid disturbances, faulty conditions, and torsional interactions resulting from the interplay between DFIG-based wind power systems and synchronous generators. A controller hardware in Loop (CHIL) simulation using OPALRT is conducted to demonstrate the efficacy of the proposed DILRNN-tuned PLL in enhancing synchronization stability. The findings showed that the adaptive PLL achieved an 80% reduction in power fluctuations and a 78% decrease in voltage variations under severe fault conditions as compared to those using conventional PLL designs.