Research on Hybrid Mechanistic and Data-Driven Modeling of Wind Turbines Based on Particle Swarm Optimization

In response to the complex nonlinear operational characteristics of large wind turbines, a hybrid semi-mechanistic modeling method for multi-condition operations is proposed, based on a refined 5MW wind turbine model and co-simulation using FAST and Simulink. First, a complex nonlinear mechanistic model is constructed through mechanistic analysis, identifying the input and output variables, and completing data collection. Then, the aerodynamic system is linearized using a Taylor expansion, and an incremental model is built for key nonlinear components at steady-state operating points. Finally, guided by mechanistic insights, the particle swarm optimization (PSO) algorithm is applied to refine the aerodynamic partial derivatives, optimizing the results of the model linearization. The conclusions show that the proposed modeling method retains advantages of speed and precision under various operating conditions, while preserving the physical meaning of model parameters. This method effectively captures the complex nonlinear response characteristics of wind turbines across different operating conditions, laying the foundation for wind turbine condition monitoring and controller research.