As a researcher specializing in intelligent bridge systems and structural control, this study presents a novel, expert-informed framework that leverages machine learning to optimize Proportional-Integral (PI) control parameters for bridge systems equipped with the Neutral Equilibrium Mechanism (NEM). A precisely scaled bridge model integrated with dual NEM units was employed to collect over 21 million high-resolution displacement data points under four carefully designed control scenarios. Advanced computational methods—including neural networks and random forest regression—were used to accurately model the system’s dynamic displacement behavior. Subsequent analysis using K-means clustering and sensitivity evaluation identified a medium-gain setting (GP = 1.0, GI = 0.010) as the optimal configuration, achieving a balanced trade-off between displacement suppression and system synchronization. These findings highlight the significant potential of AI-assisted control strategies in enhancing the performance, adaptability, and resilience of next-generation smart bridge systems. This work contributes meaningful insights at the intersection of machine learning and structural engineering, with far-reaching implications for future infrastructure design and control methodologies in civil engineering.