Abstract: At present, the industrial robot is used as the executive mechanism of the X-ray inspection system for the welding seam of the rocket tank, which usually follows the “teach-reproduce” trajectory movement and thus cannot adapt to the flexible and intelligent production line with small batch and variety. In this paper, servo mechanism, image processing technology and intelligent robot were combined, and the image-based visual servo control (IBVS) method was used to feed back the current image feature information, which was compared with the expected image feature to form the image feature deviation. The obtained feature error was based on SSA-BP control method, which was used to compensate the weld position error in the process of detecting the welding seam of rocket tank, so as to improve the working accuracy of the X-ray inspection system. Finally, the experiment showed that this method had good adaptability and robustness.