高架铁路客运站通常为大跨度建筑，需设置天窗。传统的天窗设计方法难以解决复杂的采光需求和节能需求构成的多目标问题。为实现高铁站平天窗的多目标优化，文中基于高架高铁站平天窗的前期设计参数设置，采用Rhino和Grasshopper平台、Ladybug建筑性能模拟工具、Octopus多目标优化工具构建了一套基于遗传算法的平天窗多目标优化方法。该方法经过确定变量、确定优化目标、构建模型和程序编写等步骤，用Rhino和Grasshopper建立简化的参数化模型，导入Ladybug工具进行性能分析，依据分析结果用Octopus工具进行迭代的多目标优化；在优化过程中能够自动地对模型参数化的部分进行不断变更和模拟，记录每次变更、模拟的结果并进行比较，最终找出最满足设定的多个目标的参数；将参数返回到参数化模型即可得到最优结果的模型及对应的建筑性能模拟结果。对广州白云站的候车空间进行建模，依据国内外主要采光规范的要求，将采光强度达标、采光均匀度达标、采光有效性尽量大、眩光发生可能性尽量小、太阳辐射量尽量小作为目标体系，使用文中提出的方法进行多目标优化。结果表明：相比于原方案，经该流程多目标优化后最终生成的方案在满足规范要求的采光强度条件下，采光均匀度、采光有效性、眩光发生可能性、太阳辐射量方面均有很大程度的改善。文中提出的方法具有广泛的应用场景和较强的灵活性，可以为相关研究提供参考。

Elevated railway stations are usually large-span buildings and require skylights. Traditional skylight design methods have difficulties in solving the multi-objective problem of complex requirements in lighting and energy-saving. In order to realize the multi-objective optimization of the flat skylight of the high-speed railway station, based on the pre-design parameter settings of the flat skylight of the elevated high-speed railway station, this paper constructed a set of genetic algorithm-based multi-objective optimization methods using Rhino and Grasshopper platforms, building performance simulation tool called Ladybug, and multi-objective optimization tool called Octopus. Multi-objective optimization method for flat skylight goes through the steps of determining variables, determining optimization objectives, building models and programming, using Rhino and Grasshopper to build a simplified parametric model, importing the Ladybug tool for performance analysis, and using Octopus tool to carry out iterative multi-objective optimization according to the analysis results. The optimization process can automatically change and simulate the parameterized part of the model, and record and compare the results of each change and simulation. And finally, it finds out the parameters that best meet the set multiple objectives. Returning the parameters to the parametric model can yield the optimal model and the corresponding building performance simulation results. Furthermore, an empirical analysis was carried out by taking Guangzhou Baiyun Station as an example. According to the requirements of the main lighting standards at home and abroad, the study first set the daylighting factor and the daylighting uniformity up to the standard, the useful daylighting illuminance as significant as possible, the possibility of glare occurrence as small as possible, and the solar radiation as small as possible as the target system. Then it used the method for multi-objective optimization. The results show that compared with the original scheme, the final scheme meets the basic standard of daylighting factor and has better lighting uniformity, useful daylighting illuminance, glare occurrence possibility, and solar radiation under the lighting intensity conditions. The proposed method has a wide range of application scenarios and more flexibility and can provide references for related research.