The establishment of “dual carbon” targets has highlighted the significance of integrated industrial parks that combine energy generation, grid interaction, load management, and storage (“source-grid-load-storage”) for achieving efficient energy use and low-carbon transitions. A primary challenge, however, lies in the variable and uncertain nature of renewable power, which complicates both park operation and grid stability. This study investigates how hybrid energy storage systems can enhance park dispatch and support grid ancillary services. A dynamic optimization model and a two-stage cooperative algorithm are developed for the integrated park system, aiming to balance economic efficiency with grid support functions like peak shaving and frequency regulation. Findings demonstrate that the proposed two-stage framework successfully reconciles the inherent trade-offs between park economics and grid service provision, leading to improved overall system performance. This work offers both theoretical insights and practical strategies for optimizing industrial parks and enhancing grid ancillary services.