State estimation for DFIG-based wind turbines under voltage dips using multiresolution sinusoidal neural network-Tasmanian Devil optimization in Internet of Things enabled systems

State estimation for Doubly Fed Induction Generator (DFIG)-based Wind Turbines (WTs) in IoT-enabled systems is crucial for maintaining stability and performance, especially during voltage dips that cause grid fluctuations. To address these challenges, this research proposes a hybrid approach for state estimation using a Multiresolution Sinusoidal Neural Network (MSNN) combined with the Tasmanian Devil optimization (TDO) algorithm, referred to as the MSNN-TDO method. The objective is to enhance system stability, improve forecasting accuracy, and reduce estimation errors. MSNN is utilized to predict the dynamic behavior of DFIG-based WTs during voltage dips, while TDO optimizes the MSNN parameters for better accuracy and stability. The model's performance is benchmarked against existng techniques, including Radial Basis Function Neural Network (RBFNN), Long Short-Term Memory-Swarm Intelligence Optimization (LSTM-SIO), Lightweight Multiscale Neural Network (LMNN), Transformer-Based Deep Neural Network (TDNN), and Artificial Neural Network (ANN) using MATLAB. The proposed MSNN-TDO method achieves a prediction accuracy of 98.8 % with an error rate of 1.5 %, significantly outperforming existing methods in state estimation accuracy and error reduction. These findings shows that the MSNN-TDO performs in improving system stability and dependability, which makes it a useful tool for controlling state estimation in WT systems with IoT support.