PURPOSE: To ensure precise in-tunnel ventilation control by predicting a heavy traffic condition in the next cycle in accordance with the real values of a traffic density and vehicle speeds and performing switching between a steady-state model and a heavy-traffic model, accordingly.

CONSTITUTION: The real values of a traffic density and vehicle speeds are found by a traffic counter 1 and a traffic density and vehicle speeds for each of given cycles are predicted by a traffic density predicting means 2 and a vehicle speed predicting means 3, accordingly. Whether heavy traffic occurs in a tunnel in the next cycle or not is predicted by a heavy traffic predicting means 4, accordingly. Switching between a steady-state model 5 and a heavy- traffic model 6 is performed by a model switching means 7, accordingly. The air quantity of a ventilator and the reference value of a number corresponding to the applied model is determined by a ventilator air quantity and number reference value determining means 8. Outputs from a VI meter 9, a CO meter 10 and a wind direction and speed meter 11 are input to a feedback control means 12, a modified value is found for the reference value and it is input to a ventilator air quantity and number instruction value determining means 13 to determine a final instruction value.