The information knowledge base can improve production efficiency and reduce resource consumption in industrial production, but it frequently meets with data failure. Furthermore, erasure coding strategies for multi-node failure reconstruction are limited, and the link relationship between participating nodes during the reconstruction of failed data has not been fully considered, resulting in low reconstruction efficiency. Aiming at the data failure of the information knowledge base, a multi-node failure reconstruction method was proposed. According to the data processing capacity of the node, the new node with the highest data processing capacity was selected as the routing node; then according to the link bandwidth of the routing node, the candidate supply node and the remaining idle nodes, the supply node and the new node were determined, thereby constructing the data reconstruction network topology and improving the efficiency of reconstruct failed data. Experimental results show that this method has shorter reconstruction time and higher reconstruction success rate when compared with traditional erasure code reconstruction methods.