Intelligent traffic management via personalized group consensus based on chimp optimization-guided random vector functional link and quantum theory: A perspective of randomization

Abstract

In urban traffic management, bike-sharing systems are crucial for green transportation. However, due to the uneven distribution of shared bikes and randomization of user behavior, the urban dockless bicycle sharing system (UDBSS) faces issues of randomization. Since rebalancing in UDBSS involves the opinion and preference of multiple stakeholders, it can be modeled as a group consensus problem. Nevertheless, mutual influence among users, changing preferences, and psychological inconsistencies, along with the absence of personalized strategies in traditional methods, adversely affect demand decisions for UDBSS. To address this issue, this paper innovatively combines random vector functional link (RVFL) networks, quantum theory (QT), and prospect–regret theory (P–RT), to construct a personalized two-stage group consensus framework. First, with the support of three-way decisions, an improved K-means++ algorithm based on Euclidean distances and Hausdorff distances is used for clustering, which reduces the uncertainty in the UDBSS problem. Additionally, to address the randomization issue, RVFL is used to calculate intragroup user weights, and the chimp optimization algorithm (CHOA) is applied for the hyperparameter optimization. Furthermore, considering users’ psychological behavior, a two-stage consensus reaching process (CRP) is designed, and a personalized adjustment mechanism based on QT, P–RT, and hesitation degrees is proposed. Finally, the proposed model is applied to a shared bicycle deployment scenario, with experimental analysis using data from the Citi Bike system and survey data to verify its effectiveness and feasibility.