Research on Dynamic Planning of a Modular Cabin Assembly Sequence for Large Cruise Ships Based on an Improved Genetic Algorithm

Assembly sequence planning for prefabricated modular cabin units must satisfy geometric, precedence, stability, direction, and tool constraints while remaining responsive to workshop disturbances. This study formulates the task as a constrained, scalarized multi-criteria optimization problem, and combines constraint-aware greedy screening, blockwise split-and-recombination operators, and a feasibility-guided population injection in an improved genetic algorithm (IGA). The fair computational study used identical population sizes, evaluation budgets, stopping rules, crossover probabilities, and mutation probabilities for GA and IGA, with 30 independent runs on a reconstructed, anonymized 15-component benchmark derived from the component attributes and sequences reported in the submitted manuscript. IGA reached the best-known fitness of 0.115741 in 30/30 runs and required a mean of 10.6 generations to reach that value, whereas GA reached it in 26/30 runs and required 124.8 generations among successful runs. Simulated annealing and ant colony optimization reached the same best-known value, so global optimality is not claimed for the 15-component benchmark; exhaustive enumeration only verifies the optimum of a reduced nine-component instance. Ablation, diversity, parameter sensitivity, and synthetic 30–60-component tests clarify the contributions and limits of each mechanism. The event-triggered dynamic planning layer was additionally evaluated for six disturbance types and completed replanning in 0.52–0.62 s, reducing normalized waiting times by 33.3–100% in the release-delay scenarios. The results support feasibility and rapid replanning for this computational benchmark, while validation on complete shipyard matrices and measured task-time data remains necessary.