Task status evaluation strategy for dynamic traffic police dispatching in accident response

Urban road traffic safety is facing severe challenges, with limited police resources leading to a large number of traffic accidents failing to receive timely response and rescue. The key to alleviating this contradiction lies in optimizing the utilization efficiency of police resources. This study investigates the police resource dynamic dispatching problem for traffic accident response under uncertainties in accident occurrence time, location, and police handling duration. A dynamic assignment optimization model for traffic police is constructed, and a quantile-based Status Evaluation Strategy (SE) for police dispatching is proposed to solve the model, which leverages historical accident handling duration quantiles to assess police status. Empirical analysis is conducted using accident data from Yinzhou District, Ningbo (January-July 2024; 151 days, 13,793 accidents). The Wilcoxon signed-rank test is employed to compare the proposed strategy with traditional Greedy Idle Immediate Response Strategy (IR) and Time-Window-based Adaptive Large Neighborhood Search Response Strategy (ALNS). Results show that the Status Evaluation Strategy using the 90th percentile handling duration (SE90) performs best, significantly reducing accident handling delay (SE90-IR: z= - 5.254, p < 0.001, r = 0.428; SE90-ALNS: z= - 3.394, p < 0.001, r = 0.276), and exhibits better performance stability compared to the same strategy using the 75th percentile handling duration (SE75). Further analysis of delay components reveals that SE90 outperforms others by better balancing the conflict between dispatching decision delay and total travel time of police. This study provides theoretical foundations and technical frameworks for developing intelligent dynamic police resource allocation systems in urban traffic accident response.