Abstract: China's urban construction has entered the stage of stock updating, refined planning, and management. Comprehensive parks are a scarce and important public resource. The reasonable layout of comprehensive parks helps optimize the park system and improve resident wellbeing. Presently, the spatial layout of comprehensive parks is mainly implemented based on per capita indicators and service radius, but the refined layout decisions lack quantitative support. The layout optimization model is often used to solve layout problems of public service facilities. Reasonable allocation of comprehensive park resources based on the layout optimization model is important to optimize the supply-demand of parks and promote scientific layout decision-making for comprehensive parks.
　　The new gravity P-median model is a layout optimization method that integrates multi-objective orientation, facility supply-demand matching relations, and complicated constraints. It can meet practical needs of accurate spatial layout of comprehensive parks, but had the following limitations: (1) With respect to model construction, the new gravity P-median model ignores differences of supplying subjects and is difficult to adapt to differing scale requirements between existing facilities and new facilities during practical layout design. (2) With respect to the solving algorithm, as layout problems of comprehensive parks become increasingly complicated the model considers more comprehensive and refined factors of park layout, thus making it difficult to solve the new gravity P-median model. With comprehensive consideration of differences between supplying subjects and difficulties in model solving, an improved gravity P-median model for layout optimization of comprehensive parks was constructed. Major corrections were: (1) Refining the supply subjects and improving the scale distribution rules means the improved new gravity P-median model ignores limitations from the service supply differences between existing parks and new parks. (2) The genetic algorithm was introduced for optimization, which solves the layout problems of nonlinear and complicated multi-constraint facilities. It solved the model successfully and attained the scientific layout optimization scheme. This is convenient for promotion and application of the model in practical layout planning of comprehensive parks.
　　A case study based on the central urban area of Suzhou was carried out. Supported by high-precision multi-source data, the research scale was narrowed to the residential area scale. According to the empirical study results, (1) the improved gravity P-median model was proven valid. After optimization some indicators were improved, such as maximum travel cost of residents to nearby comprehensive parks, average travel cost, and accessibility Gini coefficient. The spatial layout of comprehensive parks was more reasonable. (2) The improved gravity P-median model is applicable to layout scheme design of comprehensive parks in new areas. It not only provides scientific and logical decision references and effectively lowers costs of trial and error, but is also applicable to spatial layout optimization of comprehensive parks in urban areas under the stock updating conditions. Moreover, it can guide the short-term, middle-term, and long-term planning of comprehensive parks in combination with current park resource allocation conditions in cities. This realizes the spatial efficiency layout optimization goal, with considerations for spatial equity. These results provide effective decision-making tools for layout planning of comprehensive parks and offer some references for refined layout of green space in parks.

Key words: comprehensive parks, spatial layout, new gravity P-median model, genetic algorithm, Suzhou

摘要： 综合公园作为重要且稀缺的城市公共资源，其合理布局对城市存量更新背景下优化公园体系、增进民生福祉具有重要意义。布局优化模型常用于解决公共服务设施布局问题，新重力P中值模型是一种综合多目标导向、设施供需匹配关系和复杂约束条件的布局优化方法。在辨析新重力P中值模型的基础上，综合考虑供给主体差异和模型求解难度，构建出面向城市综合公园布局优化的改进新重力P中值模型，主要修正内容如下：（1）通过细化供给主体，改良规模分配规则，改进新重力P中值模型未考虑现有公园和新建公园服务供给差异性的局限；（2）引入遗传算法寻优，解决非线性、多约束的复杂设施布局问题，实现模型求解以促进模型在公园实际规划中的推广应用。以苏州市中心城区为例，以多源数据为支撑，将研究尺度下沉至居住区尺度，实证结果表明模型优化后居民至邻近综合公园最大出行成本、平均出行成本和可达性基尼系数等指标均有改善，综合公园空间布局更为合理。研究结果可为综合公园布局规划提供有效决策工具，为公园绿地的精准化布局提供参考。

关键词: 综合公园, 空间布局, 新重力P中值模型, 遗传算法, 苏州