[en] In CNC shaping milling machine, the prediction of the state of tool wear has important application significance to improve productivity, reduce scrap rate and avoid security risks. In this paper, the detection and control system of disk milling cutter is set up by the current monitoring method, the input characteristic quantity and target characteristic quantity of BP neural network for tool wear diagnosis are measured, and the disk milling cutter wear condition prediction neural network is established based on the GA-BP algorithm. At last, the online prediction of milling cutter wear state is realized. The network test results show that the prediction rate of tool wear condition is more than 92.78%. (paper)