To address voltage violations, frequency fluctuations, and other challenges caused by the high-penetration integration of distributed photovoltaic (PV) generation into distribution networks under the “dual carbon” goals and energy transition, as well as the limitations of conventional grid-following energy storage systems due to their passive response characteristics, this paper proposes a grid-forming energy storage-based solution. A bi-level coordinated optimization model integrating site selection, capacity allocation, and control is developed. Scenario analysis is employed to handle PV output uncertainty, and a hybrid optimization method combining an improved particle swarm optimization algorithm with an interior-point method is adopted to solve the model, achieving a multi-objective balance between economic and technical performance. The proposed grid-forming energy storage effectively mitigates reverse power flow from PV systems and significantly improves PV curtailment reduction. Under fault conditions, it enhances the self-healing capability of the distribution network. By integrating virtual synchronous generator control with a multi-objective coordinated optimization strategy, the approach overcomes the technical bottleneck of passive response inherent in traditional energy storage, offering a systematic solution for the secure and stable operation of distribution networks with high renewable penetration.