The invention discloses a user traffic prediction system in a wireless network, which comprises four subsystems, namely, basic feature extraction, user movement trajectory prediction, user traffic preference analysis and regional traffic prediction subsystems. Firstly, the user traffic is analyzed to extract user features initially. According to the user's moving trajectory data in the region, thenext step direction of the user is predicted based on the convolution neural network model. According to the traffic situation of users in the region, the next period traffic of users is predicted based on the circulating neural network model. Combined with user movement trajectory prediction and traffic prediction, the traffic distribution of the whole region in the next period is predicted. Theinvention predicts the network traffic of each wireless access point in the next period of the area by analyzing the wireless network connection and the traffic usage of the wireless network users inthe area. It can be used to solve the problems of bandwidth and channel assignment of multiple routers in wireless network. The problem of traffic usage prediction for specific users is solved. It can also be used to predict the trajectory of the user's movement.