Aimed at the problem that topology updating is fast in construction and network communication is low in efficiency in drone cluster networking, a clustering algorithm MPC based on mobile prediction is proposed. On the basis of this, a new clustering routing protocol MPCR is proposed. At the base of the grayWNN joint forecasting model, the mobile state movement factor is obtained by applying the prediction results to the direction, speed, and position evaluation factors. The clustering algorithm selects cluster heads by calculating the mobility factor and link retention time comprehensively. Passive routing discovery mechanism between clusters and active routing discovery mechanism between cluster members are used by the routing protocol for decreasing routing overhead. This paper verifies the prediction accuracy, the clustering algorithm, and the routing protocol through the simulation. The results shows that WNN improves forecast accuracy and packet delivery rate by 3 to 4 percentage points. And the clustering algorithm extends the cluster head holding time and the average link holding time, improving the clustering stability. The corresponding routing protocol reduces the endtoend delay, compared with other same protocols, and increases packet delivery ratio of packets, improving the routing performance.