研究内容

Research contents

滑雪者行为对空间布局的影响：研究分析了滑雪者穿着厚重滑雪服等特殊行为如何影响人群流动和密度。

心理距离与满意度：采用Proxemics理论，通过对数算法量化滑雪者之间的心理距离，探讨其对旅游满意度的影响。

空间布局优化：利用机器学习技术（Unity的ML-Agents）优化更衣室布局，以平衡人群感知密度和空间利用效率。

The impact of skier behavior on spatial layout: The study analyzed how specific behaviors such as the wearing of heavy ski clothing by skiers affect crowd flow and 

density.

Psychological distance and satisfaction: Proxemics theory is used to quantify the psychological distance between skiers through a logarithmic algorithm and explore its 

impact on tourism satisfaction.

Space layout optimization: Use machine learning technology (Unity’s ML-Agents) to optimize the locker room layout to balance crowd perception density and space 

utilization efficiency.

研究方法

Research methods

使用Unity3D进行基于代理的建模，模拟滑雪者的行为和更衣室内的拥挤情况。

引入适当的身体碰撞器，模拟不同着装阶段滑雪者的行为，增强模拟的真实性。

应用人际距离理论，将人群拥挤感量化为基于距离的评分。

采用对数函数表示拥挤感知与距离之间的关系，量化滑雪者的拥挤感受。

利用Unity的ML-Agents工具包，通过模拟生成的数据训练优化模型。

设置锁柜对、定义训练边界和网格、指定放置规则、确定奖励机制，以优化更衣室布局。

Use Unity3D for agent-based modeling to simulate skier behavior and crowding in locker rooms.

Introduce appropriate body colliders to simulate the behavior of skiers in different stages of clothing and enhance the authenticity of the simulation.

Applying interpersonal distance theory, crowd crowding perception is quantified into a distance-based score.

A logarithmic function is used to represent the relationship between crowding perception and distance to quantify the crowding experience of skiers.

Utilize Unity's ML-Agents toolkit to train and optimize the model through data generated by simulations.

Set up locker pairs, define training boundaries and grids, specify placement rules, and determine rewards to optimize locker room layout.

研究结果

Research conclusions

经过500小时的机器学习训练，得出了优化后的更衣室锁柜布局方案。

与传统“I”形布局相比，优化后的布局呈现出“<”形，有助于提升交通流畅性和滑雪者的舒适度。

优化后的布局在空间质量分布上更加均匀，减少了空间质量的差异。

每秒Proxemics评分和滑雪者个体的Proxemics评分均显著提升，表明滑雪者对社交距离的满意度更高。

通过定量分析比较了传统布局与优化布局的性能，证明了新方法的有效性。

研究结果为建筑师、规划师和管理者提供了数据驱动的决策支持，有助于在未来冰雪运动设施改造和发展项目中优化空间设计。

After 500 hours of machine learning training, an optimized locker layout plan for the locker room was derived.

Compared with the traditional "I"-shaped layout, the optimized layout presents an "<" shape, which helps improve traffic flow and skier comfort.

The optimized layout is more uniform in spatial quality distribution and reduces the difference in spatial quality.

Both the per-second Proxemics score and the individual skier's Proxemics score improved significantly, indicating that skiers are more satisfied with social distancing.

The performance of traditional layout and optimized layout is compared through quantitative analysis, which proves the effectiveness of the new method.

The findings provide architects, planners and managers with data-driven decision support to help optimize space design in future ice and snow sports facility renovation 

and development projects.