Research on Data Analysis and Dispatching Strategies of Shared Bicycles based on artificial intelligence technology

In the context of the increasing popularity of shared bicycles, optimizing their dispatch during peak hours has become a crucial issue in urban traffic management. This study addresses the optimization of shared bicycle dispatch during peak hours, using data from Beijing. This paper began by cleaning and decoding a week’s dataset, then defined block activity values for different time periods to identify popular blocks during rush hours. Due to the small size and large number of blocks, this paper applied the K-means algorithm to group these into 34 larger blocks, then calculated activity during peak periods to identify popular large blocks. This paper selected six key attributes for feature extraction: user ID, starting point coordinates, departure time, day of the week, and bike type. This paper used KNN regression and a fully connected neural network to predict the destination's coordinates. After performing K-fold cross-validation, the KNN model showed better performance, with an average validation loss of 0.0293 compared to the neural network's 0.0308. To address bike distribution issues, this paper proposed a manual adjustment strategy to balance bike availability across blocks. Using the 30th popular large block during the morning rush hour, we constructed a periodic adjustment method with 15-minute intervals. This paper simplified the problem into a constrained open traveling salesman problem and used enumeration and greedy algorithms to find the optimal adjustment strategy. This approach aims to stabilize bike distribution across blocks through spatial transfers. In conclusion, this study offers a framework for optimizing bike dispatch during peak hours and provides a scalable solution for managing bike distribution in larger urban areas during both rush hours and non-working days.