Adaptive multi-objective Bayesian optimization approach for capacity planning of the interconnected offshore wind farms

The concept of interconnected offshore wind farms (OWFs) presents a promising future that motivates this paper to develop a collaborative capacity planning model covering the system’s economy, sustainability, and security needs. The model is optimized by an improved adaptive multi-objective Bayesian optimization (AMBO) algorithm, which seeks the Pareto optimal capacity allocation schemes of OWFs, offshore cables, battery energy storage systems, and hydrogen electrolysis units to form a Pareto front. Compared with other commonly used multi-objective optimization algorithms, the AMBO demonstrates its superiority in generating a more informative Pareto front sample-efficiently. In addition, it gets rid of the need for parameter tuning, thereby avoiding the introduction of planner subjectivity into the planning. The proposed planning model accounts for the inevitable simulation deviation resulting from the inherent variability of RES by modeling it into a noise function, whose side effects on the optimization can be mitigated through a probabilistic surrogate model. This approach ensures accurate performance evaluation of capacity allocation schemes without extensive simulations. With the aforementioned methodologies, the potential of OWF interconnection and integrated green hydrogen production to increase renewable power consumption and fortify the stability of power systems is demonstrated in the study.