In view of the existed problems of knowledge acquisition in intelligent systems, a dynamic knowledge extraction method based on Chinese natural language sentence-clustering is put forward, and the corresponding software prototype system is implemented. First of all, the proposed method is introduced in the paper by the way to give its outline. In order to demonstrate an important role of the proposed method, we make a complete case study via the intelligent design of certain machine tool. The design background of such a product is presented and the implementation steps is given in detail to show the whole design process. Through the practical case, we have succeeded in extracting knowledge from natural language text and the effectiveness of the proposed method is verified.