The vertical wind in a tunnel will change the heat convection and smoke distribution in the tunnel and influence the temperature-sensing error. This study starts from the relevant state departments for the mandatory acceptance of tunnel fire detection (i.e., the alarm signal must be generated within 60 s after ignition). A fiber grating temperature sensing fire detection technology used to collect the temperature field of multiple ignition experiments under different wind speeds and ignition positions is then developed. The temperature field distribution and its dynamic changes under different working conditions are analyzed. Subsequently, an intelligent algorithm for correcting the fire point position based on the temperature field is proposed. The problem of the position error of the fire alarm caused by the longitudinal wind is corrected according to the temperature distribution and the dynamic change of the tunnel fire. In addition, the fire location can be intelligently determined within the specified time.