With the high intensity and density of aerospace launch in China, aerospace project materials management faces various challenges, such as multiple aerospace projects concurrent, high supply timeliness and high quality level. Although several enterprises have successfully explored a versatile material management regulation for multiple aerospace projects, the low inventory turnover problem still exist due to the difficulty in versatile material recognition and inventory distribution. This paper studies the dynamic inventory management problem of aerospace project materials. Firstly, a numerical concept space model is established to describe the data characteristics of aerospace project materials. Then, we propose the variable-scale clustering algorithm based on the numerical concept space, which is utilized to automatically recognize versatile materials. Also, the obtained satisfy scale feature supports managers making inventory-related decision. Finally, a dynamic adjustment algorithm of inventory classification based on the variable-scale clustering is proposed, in order to dynamically maintain the inventory management plan of aerospace project materials. Experiments select the inventory data of aerospace project metal materials during Jan 1, 2015 to Mar 31, 2018 from the logistics center in China Academy of Launch Vehicle Technology. And experiment results illustrate that our proposed variable-scale clustering algorithm has high efficiency and practical application value in solving dynamic inventory management problem of aerospace project materials.