Mountain railway alignment optimization integrating layouts of large‐scale auxiliary construction projects

Abstract

Mountain railway alignment design is an important but complex civil engineering problem. To overcome the drastically undulating terrain, long tunnels and high bridges are major structures used along a mountain railway, which poses great challenges for railway design and construction. Unfortunately, despite being studied for many years, the crucial construction factors of complex structures have received slight attention in alignment optimization. In this paper, for the first time, the layout of large‐scale auxiliary construction projects (LACPs), including tunnel shafts and access roads, is incorporated into the alignment design process in order to consider construction practicability and economy. Primarily, an alignment–LACPs concurrent optimization model is built. After defining the comprehensive design variables, the alignment–LACPs total construction cost is formulated as the objective function. Besides, the separate constraints for designing the alignment and LACPs are considered. Also, a construction duration computation is proposed for constraining the alignment–LACPs integration. To solve the model, a four‐step hybrid solution method is developed. Specifically, the alignment is first generated with a particle swarm optimization (PSO). Afterward, a new divide and conquer approach is devised to search for shaft alternatives along the alignment. Then, a customized Dijkstra algorithm is developed to search for complex access roads. Finally, a novel polynomial mechanism for time‐varying acceleration coefficients (TVAC) is designed for PSO to evolve the alignment–LACPs solutions. The above model and methods have been applied to two complex actual mountain railway examples. Their effectiveness is demonstrated through detailed analysis of resulting railway solutions and control experiments with contemporary TVAC‐based methods.

A sequential exploration algorithm for the design optimization of horizontal road alignment

“…By using a data structure akin to a chain table to store the information about all of the investigated nodes and avoid computing duplicate nodes, this strategy can greatly lower the algorithm's temporal complexity. Dijkstra's algorithm and the A‐star algorithm, which have been implemented in road alignment designs (Pushak et al., 2016; T. Song, Pu, Schonfeld, Liang, et al., 2022), are typical methods based on this strategy which are used in path planning problems. These two algorithms are based on the grid cell's centroids or vertices and directly create alignments.…”

“…On the basis of a grid with i × j cells, Dijkstra's algorithm, which is a typical U‐SEA, can be improved to conduct sequential exploration (Song, Pu, Schonfeld, Liang, et al., 2022). This section presents the flow of implementing the basic Dijkstra's algorithm.…”

“…Then, the original DT was widely extended and improved by a research group at Central South University, using an improved DT to optimize railway alignments (Li et al., 2017; T. Song et al., 2020) and station locations (Pu, Zhang, et al., 2019; Pu et al., 2021). In addition to DT, researchers have also developed many other optimization methods combined with grid‐based techniques for designing railway or road alignments, such as the deep reinforcement learning method (T. Gao et al., 2022) and DA (Hirpa et al., 2016; Pushak et al., 2016; T. Song, Pu, Schonfeld, Liang, et al., 2022). In addition to the above two methods of dividing the search space, Roy and Maji (2022) suggested a variant of the ant colony optimization (ACO) approach to the problem of railway alignment design in which random sampling points were adopted to represent the search space.…”

Mountain railway alignment optimization based on landform recognition and presetting of dominating structures

“…Railway alignment development is a complex problem, which significantly influences the construction cost and operation safety of a railway project (Zhang et al, 2022).…”

“…Afterward, Pu et al (2019) further extended the search spaces and developed a three-dimensional DT (3D-DT). In addition to DT-based methods, Dijkstra's algorithm (Pradhan & Mahinthakumar, 2013;Song, Pu, Schonfeld, Liang, et al, 2022), a motion planning algorithm (Sushma & Maji, 2020;Yang et al, 2020) and a k-shortest path algorithm (Pushak et al, 2016), are also growth-based methods, which have been widely used in alignment optimization.…”