An alternative approach to the synthesis of the intelligent control of manufacturing systems understood to be the discrete-event dynamic systems (DEDS) is presented in this paper. The Petri nets (PN) are used to express analytically both the model of DEDS to be controlled and to represent knowledge about the control task specifications expressing goal criteria of control, constraints and further external conditions. Both the DEDS model and the knowledge base (KB) are simultaneously used in the procedure of the control system synthesis. The elementary control possibilities are generated (by means of the model) in any step of the procedure and tested with respect to existence conditions. When there are several control possibilities satisfying the conditions the most suitable one is chosen by means of the KB.