Efficient Integration of Machine Learning and Optimization Techniques into Passenger Ships Structural Design

The structural design of passenger ships is undergoing a transformative shift towards more sustainable and efficient solutions. This study focuses on practical applications of Machine Learning (ML) and optimization techniques to streamline the design process. By addressing challenges in structural optimization, such as high-dimensional parameter spaces and expensive simulations, we propose an advanced framework that integrates ML-driven parameterization and optimization workflows directly into the design pipeline. Our methodology emphasizes the development of an efficient optimization strategy to balance conflicting objectives, such as minimizing material usage while ensuring structural resilience under extreme conditions. The construction of efficient surrogate models reduces dependency on computationally expensive Finite Element Analysis, leveraging predictive models to identify promising configurations before final validation. In addition, multi-objective optimization enables exploration of trade-offs between steel usage, safety constraints, and manufacturing costs. The methodology includes a parameterization refinement step, in which the addition of a limited number of new decision variables ensures an increased ability in capturing the structural behavior and stress distribution, with the consequent achievement of a better value of the objective function. The framework has been successfully applied to multiple passenger ship projects, demonstrating significant improvements in the early design phase. This practical integration not only speeds up the design process but also facilitates innovative ship designs that meet stringent environmental and performance criteria. The approach highlights the potential of ML and optimization to support the maritime industry, bridging the gap between computational advancements and real-world applications.