Effective decision-making in urban water infrastructure optimization, particularly in sustainable urban drainage systems (SuDS), hinges on navigating complex multi-objective problems. This study addresses the challenge of sparse Pareto-fronts in many-objective SuDS design, often caused by algorithmic limitations and the intricate objective function interactions, impacting the availability of diverse design alternatives for decision-makers. To tackle these challenges, this research proposes a novel framework that integrates advanced data imputation and surrogate modeling techniques. The framework uses artificial intelligence methods to populate the sparse regions by replicating the Pareto-front structure, predicting decision variables to guide further simulations and find efficient solutions without repeated optimization runs. The methodology is validated through a SuDS design case study located in Ann Arbor, Michigan. Following the initial optimization, sparse regions were identified in four of the eight objective functions. Using the proposed framework, 32 new and efficient SuDS designs were introduced into the sparse regions without additional optimization, enhancing the uniformity of the Pareto front. This study enhances decision support tools in urban flood management by increasing the informativeness of design alternatives available to planners and engineers.
