The invention discloses an identification method for an offline secondary channel. The method comprises the steps of applying an impact force by means of a pneumatic impact hammer at an actuator mounting position; simultaneously measuring a force signal of the impact hammer and an acceleration signal of an error sensor, and identifying the dynamic characteristic of the structure by means of the two signals; generating a wideband white noise signal by means of a controller, driving an actuator by means of a filter and a power amplifier, and calculating the dynamic characteristic of the actuator through acquiring the acceleration signal on an actuator mover and the white noise signal transmitted from the controller; synthesizing the calculated dynamic characteristic of the controlled structure and the dynamic characteristic of the actuator for obtaining a weight coefficient of a secondary channel; and according to a control requirement, performing resampling on the calculated secondary channel weight coefficient, thereby obtaining the total system of the secondary channel. The signal of a controlled machine error response point is not affected by the noise signal which is generated by the controller, thereby realizing relatively high identification precision of the secondary channel. The invention further discloses an identification device for the offline secondary channel.