Urban parks can mitigate surface heat in high-density cities, but their cooling services may not align with population demand or social vulnerability. This study developed an equity-oriented planning framework for central Chengdu, China, linking park-level cooling assessment, cooling supply–demand analysis, social vulnerability, and land-constrained multi-objective optimization. Cooling intensity and effective cooling distance were estimated for existing parks from Landsat-derived land surface temperature (LST) and park-level temperature–distance profiles. Cooling demand was represented by LST, activity-informed population distribution, and building density; distributional inequality and vulnerability-sensitive deficits were assessed separately. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) optimized new-park locations and nominal areas, and a constrained Random Growth Algorithm (RGA) converted the selected allocation into contiguous polygons. A consistent linear distance-decay function was used for baseline and intervention assessments. Cooling demand was concentrated in the dense urban core, whereas relatively high modeled cooling supply occurred around larger peripheral green spaces. The selected solution comprised 71 proposed parks with a nominal target area of 1265.42 ha; 842.06 ha was realized after polygon growth. The realized configuration reduced population-weighted and Social Vulnerability Index (SVI)-weighted cooling-service deficits by 34.93% and 25.54%, respectively. Thirty independent runs and alternative cooling-response scenarios were used to assess optimization stability and model sensitivity. The framework distinguishes theoretical allocation from spatial realization when evaluating equity-oriented cooling provision.