To solve the imbalance between efficiency and safety caused by traditional static speed limit management in road construction zones (which fails to adapt to dynamic traffic flow), this study proposes a dynamic speed limiting method under the vehicle-road collaboration environment. A multi-lane improved Nagel-Schreckenberg (NaSch) model based on Cellular Automata (CA) was constructed to simulate vehicle car-following and lane-changing behaviors in construction zones. With "maximizing traffic efficiency and minimizing speed fluctuations" as objectives, a dynamic speed limit control strategy was designed using the Genetic Algorithm and ε-constraint method. VISSIM was used to build simulation scenarios, comparing three schemes: static speed limit (Scheme A), traditional time-period dynamic speed limit (Scheme B), and vehicle-road collaborative dynamic speed limit (Scheme C). Results show that Scheme C outperforms A and B: in peak periods (2000 pcu/h), average delay decreases by 28.25% (vs A) and 16.67% (vs B), speed fluctuation standard deviation drops by 25.23% (vs A); in off-peak periods (1200 pcu/h), average delay reduces by 21.51% (vs A), and average speed increases by 11.99% (vs A). The method effectively improves construction zone traffic efficiency and safety, with more significant advantages in peak periods.