The invention provides a D2D communication network slice 
allocation method based on deep 
reinforcement learning. The method comprises the following steps: S1, classifying communication services according to service types, and building a multi-service slice and D2D slice 
resource allocation model; s2, constructing a 
reinforcement learning model of slice 
resource allocation according to a Dueling DDQN 
algorithm; s3, defining a current state s, a next moment state s' and a current action a of a service slice for an agent in the Dueling DDQN 
algorithm, and constructing a reward r of a 
system by the states and the actions; and S4, learning of Dueling DDQN is carried out by using experience playback, and finally, an optimal solution of slice 
resource allocation is obtained. According to the method, resource allocation is carried out on the multi-service slices and the D2D slices, the multi-service slices and the D2D slices correspond to different uRLLC slices, mMTC slices, eMBB slices and D2D slices, a resource allocation model based on deep 
reinforcement learning is constructed in combination with a network 
slicing technology and a Duelling DDQN 
reinforcement learning algorithm, the slice resource allocation efficiency is improved, the communication requirements of various services are met, and the experience quality is optimal.