The invention discloses an FPGA (Field Programmable Gate Array) online predication control method based on a Kuhne macroscopic traffic flow model, which is used for solving a technical problem of poor real-time property of the traditional FPGA online predication control method. A technical scheme is as follows: through carrying out approximate discrete processing on the model, establishing a parallel processing flow and designing a dynamic data storage scheme, predication controls of a sealed road ramp entrance and a variable information board based on the Kuhne macroscopic traffic flow model are achieved by using the FPGA. Therefore, traffic flow density and running speed on a highway are effectively controlled in real time.