The stabilization of the Line of Sight (LOS) of an Electro-Optical System (EOS) locating on a moving platform contributes significantly to the image quality. A portion of perturbation inherited by the base motion can be eliminated by numbers of algorithms with powerful processors boards. However, it is difficult to implement these algorithms in embedded systems because of memory capacity and processing speed limitation. This paper introduces a method for identifying gimbal parameters and feedforward compensators. The key parameters including the friction force, cross-coupling effect and misalignment compensator are investigated using feedforward compensator theory and verified by practical experiments. The effectiveness in the stabilization loop is validated through many scenarios of disturbance. The result shows that the good improvement for the stabilization level of inertial stabilization platform has been achieved, and the reduction of LOS RMS errors is up to 40 per cent.