Abstract: Based on the combination of AI heuristic search optimization method and human expertise rules, this paper addresses the development of an expert system for production scheduling problem in a two-stage multi-product process. The knowledge representation scheme and the structure of the system are proposed. The system can readily deal with the qualitative factors involving in the problem, and speed up the system running by sufficiently using the background knowledge of the problem. An application of the proposed expert system to the lube oil system in a large oil refinery is presented as an example to illustrate the satistactory results.