Investigating Innovative Architectural Patterns Based on BIM and GA Using Design Typology in Railway Station Design

Spatial arrangement is a complex problem for architects, who must consider adjacencies, room sizes, and site constraints. This research optimizes spatial layout in two dimensions using genetic algorithms. Visual programming languages that interface with BIM software are used to allow for semantic and topological qualities of each station to be utilized for developing complete floor plans. A set of case studies are presented with the adoption of different machine learning techniques to demonstrate the relative performance of each approach, using the minimization of unusable space as a performance metric. Previous research has considered purely mathematical solutions to support the automation of these layouts, often resulting in irregular polygons or other room shapes. From an architectural perspective, this is undesirable, and therefore a rectangular packing algorithm has been developed and utilized. On the other hand, architectural heritage in historical areas, as a complex type of heritage, encompasses both the uniqueness of the building itself and the cultural and regional characteristics as a group. This paper develops a digital form generation method for a railway station using design typology analysis.