The invention discloses an intelligent traffic path planning method based on federated learning and digital twinning. The method comprises the following steps of: S1, carrying out vehicle registration in a vehicle system, and verifying the identity information of vehicles; S2, according to the local historical data of the vehicles, participating in federated learning, and training a local model; S3, aggregating the local models of all the vehicles to obtain an aggregation model; S4, judging whether the aggregation model converges to preset precision or exceeds a time limit or not, if yes, entering the step S5, and otherwise, returning to the step S2; S5, establishing an Internet-of-vehicles global digital twinning model; S6, periodically updating the Internet-of-vehicles global digital twinning model; and S7, initiating a path finding request to a roadside unit, and updating the optimal path and the local prediction model in real time. The planning method provided by the invention is applied to the field of Internet of vehicles, so as to solve the problems of low flow prediction and path planning accuracy, high time delay and privacy leakage risk in the current road traffic system.