Objective  To address the significant challenges posed by motorized transportation, including traffic congestion, air pollution and excessive carbon emission, the global promotion of bicycles as a green mode of travel has gained momentum. As one of the most important green infrastructure in cities, the urban cycling route network serves as a vital support for residents to travel in an environmentally-friendly manner. In recent years, the planning and construction of urban cycling route networks have gained more attention due to national strategic initiatives such as the carbon peaking and carbon neutrality goals and Healthy China. However, the current status of cycling route network planning in China remains unsatisfactory, characterized by issues such as the inadequacy of quantity, lack of coherence and consistency, and difficulty in safeguarding cyclists’ right of way. Additionally, existing planning methods for cycling route network may have such shortcomings as excessive reliance on subjective experience, disregard for potential cycling demand, and overlooking constraints during construction practice. To remedy these shortcomings and achieve effective planning of cycling route networks in China, this research aims to propose an innovative planning method for cycling route network based on multi-source urban big data.

Method  Firstly, the proposed method combines the origin-destination (OD) data of bike sharing with the shortest path algorithm to accurately depict cycling trajectories and calculate the volume of cycling traffic at the road level. Additionally, relevant POI (point of interest) data are introduced to quantitatively represent the spatial characteristics of urban roads. Subsequently, a multi-objective combination optimization model is proposed to address the problem of cycling route network planning. This model incorporates various targets including the current usage of the cycling route network, and potential cycling demand measured by the number of POIs and the mixedness of road functions, as well as some construction constraints such as planning budget and road connectivity. To account for different planning preferences, linear weighting with distinct parameters is employed to design differentiated scenarios for each planning objective. Furthermore, a modified ant colony algorithm is proposed, enhancing the traditional ant colony algorithm in three key aspects: neighbourhood selection rules, starting point selection, and pheromone updating rules. This modification aims to adapt the classic ant colony algorithm to the specific problem of cycling route network planning. Finally, the proposed method is applied and validated in the central area of Ningbo that encompasses the five administrative districts of Zhenhai, Beilun, Haishu, Jiangbei, and Yinzhou, aiming to deliver cycling network planning schemes that can maximize the benefits for Ningbo.

Results  The results demonstrate that the proposed method can effectively identify urban roads and functional nodes that should be prioritized in future cycling route network planning. The research also provides differentiated planning schemes under different scenarios and objectives. It is noteworthy that although the number of POIs and the mixedness of road functions are considered in the planning objectives, the optimization results align closely with the distribution of road-scale cycling traffic volume, indicating that planners should prioritize realistic cycling demand when planning the cycling route network. Besides, in every scenario, the optimization results consistently concentrate on the planning routes in the three-river catchment area, suggesting the importance of this area for green transport travel in Ningbo. Several significant streets, such as Yaxing Street, Kaiming Street, Jiefang North Road, Dazha South Road and Renmin Road and some key functional areas including Tianyi Square, Chenghuang Temple and Ningbo Old Bund, frequently appear in optimization results and need to be prioritized in future planning practice. Finally, the circular cycling route network obtained from the multi-objective optimization scenario holds significant implications. Planners should consider the urban form of Ningbo and the unique morphology of the three-river catchment area when planning the circular cycling route network to foster regional synergistic development and increase the overall ratio of green travel in Ningbo.

Conclusion  This research uses a swarm intelligent optimization algorithm to solve the problem of cycling route network planning by introducing multi-source urban data and considering realistic constraints, with a view to promoting green mobility represented by bicycles. The results demonstrate the effectiveness of the proposed method in tackling the aforesaid problem, offering diverse planning schemes tailored to different scenarios and objectives. Particularly in densely populated urban areas, the method successfully identifies priority roads for cycling route construction and upgrade, thereby enhancing the utilization efficiency of stock road space. This method can also be transferred to other urban planning problems and serves as a theoretical foundation for urban renewal in high-density urban contexts. More importantly, the introduction of mathematical modelling and data-driven problem solving is an enhancement to traditional empirical-based planning methods and has important theoretical implications for the development of innovative urban planning methods. To further improve the method, three directions can be considered. Firstly, bring problem modelling closer to reality by introducing comprehensive objectives and constraints. Secondly, benefit the parameter setting of the algorithm by incorporating theoretical justifications and suggestions from stakeholders, thus achieving more reasonable parameter selection. Lastly, yield more robust findings by cycling data spanning a longer period from multiple operators.