Objective In the context of multidisciplinary and multi−technology integration, this research intends to collect residents’ activity data, comprehensively analyze residents’ real travel tendencies through spatio−temporal behavior analysis and machine learning, explore the change pattern of relevant indicators of residents’ real trajectories, and then simulate residents’ travel behavior and accordingly figure out the route selection law for urban greenway, so as to propose an intelligent route selection method for greenway in central urban area and promote the update of the planning method for greenway route selection.
Methods This research adopts classical design theory as the theoretical support, collects multi-source urban big data, and proposes an analysis framework for intelligent route selection for urban greenway with the help of quantitative analysis methods such as artificial intelligence (AI), machine learning, and spatio-temporal behavior analysis. The research integrates multi-source data such as geographic data, POI data, cell phone signaling data, resident activity data, streetscape data and land use data. By analyzing the spatio-temporal characteristics of residents’ travel in reality, measuring the influence elements of greenway route selection, and then simulating residents’ real travel behavior, the research extracts potential urban greenway networks, and finally forms a planning scheme for greenway route selection. 1) To be specific, the data about cell phone signaling, motion trajectory, streetscape and road network are collected to build the current situation database; 2) GIS, spatio-temporal behavior analysis and machine learning are combined to extract 10 indicators in 5 dimensions based on the analysis of all roads within the research area; 3) the information about roads passed through by the motion trajectory and about the passage order of such roads is extracted, and corresponding interoperability relationship between these roads is converted into the Networkx library format that can be recognized by AI framework; additionally, the LSTM neural network is adopted for training to get different attribute change laws of the motion trajectory, the start and end points of the real trajectory are input into the AI framework, and the A*-like algorithm is adopted to simulate the motion trajectory and verify and compare the simulation results with the real motion trajectory; 4) the start and end points of greenway route selection are extracted and input into the trained AI framework to obtain the simulated resident travel routes and identify the high-frequency travel routes as the basis for greenway selection planning, based on which a greenway planning scheme is finally formed.
Results It is found that the roads with high-heat slow motion trajectories in the central urban area are mainly concentrated in the central areas of Suyu and Sucheng Districts, the heat of road trajectories in the peripheral suburbs and towns is relatively low, and the heat of slow motion trajectories is mainly concentrated in the central urban area near large park. It can be seen from the nucleus density at the starting and ending points of slow motion trajectory that, the nucleus density at the starting point is mainly concentrated in the historic city area of Suyu District, while that at the ending point is more concentrated and mainly distributed along the south side of the ancient Yellow River Wetland Park. Combining LSTM and A*-like algorithms to build a trajectory simulation model can effectively enable the evaluation of greenway route selection, simulate the real travel behavior of residents, and extract the potential greenway network. By simulating residents’ travel in a slow-moving manner, the research finds that the residents living along the Lakeside Avenue, Xiangwang Road, Jinshajiang Road and Zhenxing Avenue typically travel at a high frequency and are suitable for creating a greenway system in series. The finalized greenway route connects the two important water systems of the Middle Canal and the Ancient Yellow River from east to west, and connects resource points such as Santai Mountain Forest Park, Lakeside Park, Xiangwang’s Hometown, Yuji Park, Xuefeng Park, and Dongguankou Historical and Cultural Park, while combining the surrounding public space and cultural tourism space to achieve the integration between greenway and other spaces.
Conclusion This research builds an intelligent route selection method for greenway in the central urban area, which is different from the traditional indicator evaluation method. The method integrates the real activity trajectories of residents into the analysis of spatial elements, blends the characteristics of residents’ travel behavior with the elements of street environment, and provides guidance for greenway route selection planning and construction through quantitative evaluation methods. The planning method for greenway route selection proposed in this research entails regular analysis with the help of existing greenway activity trajectories of residents, and is suitable for greenway route selection planning in central urban areas, especially in old urban areas, while further research is needed for the analysis of greenway in new areas and at large regional scales. In addition to being applied to the slow-traffic system, the aforesaid intelligent route selection method can be further extended to the research on the selection of other routes such as school roads and tourist routes, so as to plan routes that are more in line with real travel behavior and can help create a human-centered urban spatial environment. In the future, AI-based planning methods can be further promoted in the field of urban planning and construction and extended to other relevant fields such as land layout and public space configuration, so as to better measure the characteristics of different spatial elements and identify their change patterns, thus contributing to the building a system of intelligent planning methods.
DownLoad: