Integrated experimental platform for identifying control parameters of doubly fed induction generators based on hardware-in-the-loop impedance testing

This study aims to enhance the teaching quality of electrical engineering courses and address students’ challenges in traditional laboratory instruction, specifically, difficulties in bridging theory with practice and insufficient hands-on experience with cutting-edge engineering problems. It considers two critical industry trends: increasing penetration of doubly fed induction generators (DFIGs) in power systems and the growing prominence of DFIG-related grid stability issues. Commercial wind turbines typically utilize encapsulated controllers with opaque parameters, which pose considerable challenges for power system stability analyses and controller optimization. To tackle these interconnected issues, this paper designs and develops an experimental platform dedicated to the parameter identification of DFIG controllers.

The platform leverages an improved particle swarm optimization (PSO) algorithm, and its operation is driven by impedance scanning data obtained from a hardware-in-the-loop (HIL) system. First, the study derives an analytical equivalent impedance model for DFIGs that explicitly considers rotor-side converter (RSC) and grid-side converter (GSC) control loops and phase-locked loop coupling effects. Building upon this model, the study establishes a clear mapping relationship between key controller parameters and the system’s frequency-domain impedance response. The model focuses on four critical controller parameters: for the rotor side, the d-axis current loop proportional gain (K_{P, d, RSC}) and integral time constant (T_{I, d, RSC}), and for the grid side, the d-axis current loop proportional gain (K_{P, d, GSC}) and integral time constant (T_{I, d, GSC}). The core of parameter identification relies on the enhanced PSO algorithm, which employs Latin hypercube sampling combined with Gaussian perturbations for population initialization to effectively enhance population diversity. During the iterative process, the algorithm adopts a dynamic frequency weighting approach, assigning distinct weights to impedance errors across different frequency bands. This weighting prioritizes frequency ranges that are critical for system stability analysis, thereby ensuring more targeted optimization. Concurrently, the algorithm integrates a dynamic parameter management module, which successfully prevents the algorithm from being trapped in local optima by implementing particle perturbations based on boundary expansion and clustering detection. To ensure that the identification results are comprehensive and accurate, the fitness function integrates three key components: impedance magnitude-frequency error, phase-frequency error, and parameter grouping error. The experimental platform, constructed using the OP4510 real-time simulation system and National Instruments data acquisition boards, can perform standard impedance scans and collect high-precision frequency response data. Experimental tests were conducted on three double-fed wind turbines under varying active power output conditions (1.0, 0.5, and 0.1 per unit). For each test condition, the proposed improved PSO parameter identification method was applied to identify the four key controller parameters.

The results indicate that the improved PSO algorithm effectively fits the measured impedance curves, demonstrating strong approximation capabilities. To enhance the reliability of parameter estimates, identification results across multiple operating conditions were weighted and averaged, yielding robust parameter values. These weighted parameters were then substituted back into the DFIG impedance model for validation. This step revealed significant reductions in impedance fitting errors, confirming the effectiveness of the proposed method and its engineering feasibility. Beyond its research applications, the developed experimental platform serves as a cutting-edge engineering practice tool, addressing a notable gap in current experimental teaching protocols for new energy power systems. By employing a visual, hands-on approach, the platform enables students to develop a deeper understanding of the intrinsic relationships between system impedance, controller parameter identification techniques, and system stability. This enriches electrical engineering students’ practical knowledge and holds significant value for cultivating innovative thinking and the ability to solve complex engineering problems.

The methodological framework and experimental validation presented herein provide a concrete contribution to the field of wind turbine controller analysis and pedagogical development in practical engineering education. By synthesizing advanced algorithmic optimization with real-time HIL experimentation, this study establishes a reproducible and effective paradigm to tackle similar black-box identification challenges in modern power electronic systems while serving as an invaluable resource to bridge the gap between theory and industrial practice.