A hyper-heuristic approach to the strategic planning of bike-sharing infrastructure. (November 2022)

• Record Type: Journal Article
• Authors: Cheng, Rong; Zhong, Shaopeng; Wang, Zhong; Anker Nielsen, Otto; Jiang, Yu
• Is Part Of: Computers & industrial engineering. Volume 173(2022)
• Publication Date: 2022-11
• Subjects: Multimodal transportation -- Bicycle network design -- Bi-level programming -- Hyper-heuristic; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2022.108704
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: Develop a bi-level model for strategic planning of the bike-sharing system. Model the cycling demand as an induced travel demand. Develop a sequence-based selection hyper-heuristic to solve the bi-level model. Demonstrate that constructing bike infrastructure could improve cycling demand. Abstract: With the increasing concern on carbon emission, climate change, and human well-being, governments worldwide are exploring ways to encourage the usage of sustainable modes of transport. Particularly, cycling is gaining attention as a healthy and green travel mode, and bicycle-sharing systems are experiencing world-spread adoption. Moreover, in response to the COVID-19 pandemic, countries have begun to expand cycling infrastructures to promote cycling considering its advantages of keeping proper social distance. This study thus develops a bi-level model for the strategic planning of the infrastructure for a bike-sharing system. The upper-level problem is to simultaneously determine the locations of bike stations and lanes to minimize the sum of the construction cost and total travelers' travel time cost. The lower-level problem is the combined mode and route choice network equilibrium problem with elastic cycling demand. One of the novelties of this study to the existing bike network literature is that it captures the reality that some travelers only begin to cycle and use bike-sharing services when there are bike stations close to both their origins and destinations. To solve the proposed bi-level model, a sequence-based selection hyper-heuristic is developed, which employs a hidden Markov model as the online learning method to determine a set of problem-tailored heuristics to explore the solution space. Numerical examples are carried out to examine the model properties and algorithm performance. The results demonstrate the positive impact of bike infrastructures on promoting cycling measured by the mode share increment.