The invention discloses a multi-intelligent-agent system collaborative fault-tolerant control method based on finite time control. The method comprises the steps of firstly, selecting an intelligent agent as a leader node in all the intelligent agents, constructing a multi-intelligent-agent system communication topology and expressing by an undirected graph through communication links among the multiple intelligent agents, calculating a weighted adjacent matrix, establishing a finite time controller based on the state error based on a finite time theory for the established multi-intelligent-agent system with the leader nodes, establishing different finite time controllers for each intelligent agent, and obtaining a global dynamic equation of the multi-intelligent-agent system, so that thesystem can achieve cooperative fault-tolerant control within a specified time from the time a fault occurs. The defects of a non-finite-time closed-loop control system are overcome, and the robustnessand the disturbance resistance of a multi-intelligent-agent system are improved.