Application of immune clonal annealing algorithm for traffic diversion in traffic accident areas

To address the delayed adaptation of conventional traffic diversion methods in dynamic accident scenarios, this study proposes an immune clone annealing algorithm (ICAA)-based framework with three key innovations. First, a residual capacity analysis model is developed to pre-identify high-congestion-risk road sections (threshold: congestion probability CP ≥ 0.65) using directed graph theory, enabling proactive congestion management. Second, an enhanced support vector machine (SVM) classifier is introduced, integrating vehicle speed, density, and occupancy with cuckoo search optimization, achieving a congestion detection accuracy of 93.2%. Third, a throughput-maximizing objective function, resolved by the ICAA, is designed to optimize traffic flow redistribution. Simulation results under peak-hour accident conditions demonstrate that the proposed framework reduces average delays by 37.5% and reduces probability of congestion by 20%. These advancements highlight the framework’s robust adaptability in multi-accident scenarios, and make it a practical solution for real-time traffic management in complex urban networks.