Data-driven PID controller of wind turbine systems using safe experimentation dynamics algorithm

The stochastic nature of wind speed and turbulence between turbines commonly stress wind turbines, emphasizing the importance of regulating rotor speed based on desired reference speed. Employing a PID-based controller is crucial for wind turbine system performance. Recent interest in optimizing PID control parameters offers advantages in output response enhancement while preserving robustness and simplicity. However, existing optimization tools, especially those using multi-agent optimization, often entail a high computational burden due to a large number of function evaluations (NFE). This study presents a novel approach employing a safe experimentation dynamics algorithm (SEDA) to tune PID controllers in wind turbine systems. SEDA, a single-agent based optimization technique, requires only one function evaluation per iteration, alleviating computational burdens. Simulation analyses, encompassing convergence curves of the fitness function, time response specification analysis of step response, stability analysis using Bode plots, and computational effort analysis based on NFE, evaluate the effectiveness of the proposed SEDA-based PID controller for wind turbine systems Furthermore, the study reveals that the settling time (Ts) and percentage of overshoot (Mp) are notably low, measuring 1.27E-4s and 0%, respectively, compared to other algorithms. These results underscore the efficacy of the SEDA method in providing optimal PID control parameters while reducing computational burdens by 52% compared to other multi-agent optimization-based methods.