In view of the Fiber Bragg Grating (FBG) sensor network monitoring the static loading on the aircraft wing box test piece, self-repairability of the FBG structural health monitoring system based on multi-agent technology is studied. When partial sensors signal can′t be acquired in certain agent, firstly, the support vector regression algorithm is used to predict the external loading damage position with the acquired sensor signals in each agent, and then the cooperative agent is used to cooperate each agent predicting result, so as to compensate the invalid sensor signals, thus the final predicting results are obtained for the structural health monitoring system. The research results indicate that the average predicting accuracy of the external loading position by the multi-agent technique is higher than that of any one agent, the invalid sensor signals can be compensated to some extent by the multi-agent technology.