Hierarchical control of wind systems: neural maximum power point tracking and Lyapunov-based dual active bridge converter

Introduction. Permanent magnet synchronous generator (PMSG) wind power systems with battery storage are an attractive solution for autonomous microgrids. The problem of the proposed work consists of two major challenges: maximizing energy extraction in the face of wind variations, requiring fast and accurate maximum power point tracking (MPPT), and ensuring robust control of the power interface despite parametric uncertainties and external disturbances. Conventional methods (Perturb and Observe, PI controllers) have limitations in terms of convergence rate, oscillations, and parametric sensitivity. Goal. To develop an integrated hierarchical control for a PMSG-passive rectifier-dual active bridge (DAB) wind turbine system enabling fast and robust tracking of the optimal MPPT reference. Methodology. The proposed approach combines two levels: a feedforward neural network (NN) that instantly generates the optimal current reference from the wind speed, and an adaptive Lyapunov control (ALC) with online estimation of the dominant uncertain parameter, mathematically guaranteeing global asymptotic stability. Results. Simulation results validate the effectiveness of the proposed approach. The NN achieves excellent MPPT accuracy with an error of less than 0.5 %. The ALC demonstrates a significant improvement in response time compared to the conventional PI controller, a substantial reduction in current ripple, and near-zero tracking error in steady state. The absence of chattering confirms the applicability of the approach to DAB power converters. The neighborhood test validates that the generated reference corresponds to the maximum power point. Scientific novelty. The scientific contributions of this article lie in 3 main aspects. First, it is the first documented application of a hierarchical architecture combining MPPT NNs and ALC to the PMSG-passive rectifier-DAB topology, rarely explored for wind power systems. Second, it rigorously identifies the dominant parametric uncertainty and compensates for it through online adaptation with a mathematical guarantee of stability. Third, it provides comparative validation demonstrating quantified gains in speed, accuracy, and stability for the energy optimization of wind power systems with battery storage. Practical value. The proposed control strategy offers a practically implementable solution for wind energy conversion systems equipped with battery storage. It improves energy harvesting efficiency under variable wind conditions while ensuring robust and stable operation of the DAB converter without requiring precise knowledge of system parameters. The approach reduces sensitivity to uncertainties and enhances dynamic performance, making it suitable for real-time digital implementation in standalone or remote microgrid applications. References 35, tables 3, figures 11.