Due to the complexity and uncertainties of the manufacturing system,both the knowledge modeling and the data mining modeling are restricted by incomplete knowledge and data. In this paper,in order to reduce the uncertainties by making full and effective use of the known information,an modeling idea of integrating the knowledge with the data mining is proposed to embed the knowledge into the rough set model,with the function relations of the knowledge established. Then,based on the indiscernibility relations and the function relations,a novel knowledgebased rough set model ( KBRSM) is developed and the knowledge classification and inference are investigated. As compared with the original rough set model,KBRSM is of high classification accuracy,excellent performance of knowledge discovery as well as of a generalized form. Experimental results show that the proposed decision model is effective and practically flexible.