Wind turbine control co-design using dynamic system derivative function surrogate model (DFSM) based on OpenFAST linearization

This research presents a comprehensive control co-design (CCD) framework for wind turbine systems, integrating nonlinear derivative function surrogate models (DFSMs) developed through OpenFAST linearization and data-driven approaches. The primary motivation for developing the DFSM is to accurately capture the nonlinear dynamics of wind turbine systems in a computationally efficient manner, thereby enabling effective and scalable optimization within the CCD framework. The developed DFSMs successfully represent state derivatives and system output responses across extensive ranges of plant, control, and state variables, validated against direct simulation outputs. By concurrently optimizing plant and control designs, the CCD approach leverages their synergistic interactions, resulting in significant reductions in the levelized cost of energy (LCOE) through an optimized balance of annual energy production (AEP) and costs associated with plant design parameters, while adhering to design and physical constraints. Comparative analyses demonstrate that CCD, particularly when utilizing open-loop optimal control (OLOC), outperforms traditional closed-loop control (CLC) strategies. Sensitivity and sparsity analyses reveal critical interdependencies among design variables, emphasizing key input–output parameter relationships that guide targeted design optimizations. These studies build on pioneering DFSM work that was limited to a handful of design and state variables; this work advances DFSM capabilities to the level of practical utility in engineering design for the first time. The work presented here serves as a foundational exploration; authors advocate for future research to incorporate broader constraints and other considerations to further advance CCD methodologies for wind turbine system optimization.