Underground Metro Interstation Horizontal-Alignment Optimization with an Augmented Rapidly Exploring Random-Tree Connect Algorithm

Abstract

The design of an underground metro alignment in an urban area is complex and challenging. The metro alignment constantly invades districts with high-density buildings and specially planned districts, which results in large additional expenses due to reconstruction, land requisition, demolition, vibration absorption, and environmental protection. This paper presents a method for the optimization of the horizontal alignment between metro stations to minimize the overall cost. The proposed method includes three steps: (1) low-cost path search, (2) initial horizontal alignment generation, and (3) horizontal alignment optimization. First, an augmented rapidly exploring random-tree connect (RRT-connect) algorithm is developed to search for a low-cost path, in which the buffer zone of the low-cost path defines the corridor constraint of the horizontal alignment optimization. Subsequently, based on the low-cost path, an initial horizontal alignment is generated that determines the number of design variables, i.e., intersection points. Next, within the corridor constraint, the method optimizes the intersection point coordinates and curve radius on the initial horizontal alignment using a differential evolution (DE) algorithm. A real-world metro line was studied to verify the effectiveness of the proposed optimization approach.