An unknown topology network interdiction strategy based on reinforcement learning is proposed. When the model of the wireless Ad hoc network is established, the proposed interdiction strategy could fulfill the needs of interdicting information transmission by jamming multi nodes and enable the jammer to interdict communication in the underlying network in real time manner. By jamming multi nodes as an operation style and counting stopped network flows as action feedback, the proposed strategy could learn the better nodes to jam without a priori knowledge of the network topology. Simulation results on the established wireless Ad hoc network with various parameters show that, the proposed interdiction strategy has a better performance in accumulate stopped flows than existing algorithms, and still has excellent jamming capability under the proposed new reward standard.