This study proposes an optimization framework for AR-based digital museum exhibition design to enhance visitor satisfaction and multisensory engagement. Despite the growing prevalence of interactive exhibitions powered by digital technologies, few existing works employ quantifiable metrics to evaluate user experience. To address this gap, the study integrates the Analytic Hierarchy Process (AHP), Gray Correlation Analysis (GCA), and Genetic Algorithms (GA) into a unified methodology. AHP and GCA are used to extract and weight user needs and behavioral preferences, forming a demand model that guides exhibit prioritization. Then GA is applied to optimize multiple design parameters and generate personalized exhibition sequences. This approach ensures the alignment between user preferences and curatorial intent, while improving design efficiency. Experimental results suggest that the integration of data-driven evaluation with intelligent optimization significantly enhances both the quality and effectiveness of exhibition design. The proposed framework is applicable not only to digital museum experiences, but also to broader digital and interactive systems, offering valuable insights for future research and practical deployment.