SSP-DMGTimeNet is a physics-constrained learning framework for spatiotemporal trajectory prediction of vehicle platoons. The model combines multi‑scale temporal representations with cross‑vehicle interaction features to capture complex time‑varying platoon dynamics. A propagation‑delay‑aware causal attention explicitly models upstream‑to‑downstream disturbance propagation by learning response delays between adjacent vehicles and accumulating them along the platoon. Time‑domain and frequency‑domain string‑stability losses suppress disturbance amplification across adjacent vehicles and arbitrary sub‑platoons during training. Experiments on the HighD dataset show an unstable‑window rate of 0.65% for five‑vehicle platoons and a maximum head‑to‑tail amplification of 0.898 on the ground‑truth excitation subset, while maintaining competitive trajectory prediction accuracy. Zero‑shot evaluation on NGSIM US‑101 and I‑80 yields velocity MAEs of 1.316 m/s and 1.252 m/s, with unstable‑window rates of 3.90% and 4.10%, respectively. These results demonstrate that incorporating platoon‑level physical constraints can balance prediction accuracy and disturbance propagation stability.
Review