A knowledge representation has been proposed using the state-space theory of Artificial Intelligence for Dynamic Programming Model, in which a model can be defined as a six-tuple M= (1,G,O,T,D,S). A building block modeling method uses the modules of a six-tuple to form a rule-based solution model. Moreover, a rule-based system has been designed and set up to solve the Dynamic Programming Model. This knowledge-based representation can be easily used to express symbolical knowledge and dynamic characteristics for Dynam-ic Programming Model, and the inference based on the knowledge in the process of solving Dynamic Program-ming Model can also be conveniently realized in computer.