Application of Markov Decision Process in Generating Units Maintenance Scheduling

An important issue in power system planning is maintenance scheduling of generating units. In a traditional power system, the problem of maintenance scheduling is of high importance and has various technical and economical constraints. Different methods have been used to solve the problem. On the other hand, changes in legal environment of electricity industry poses new bound and requirements on the maintenance scheduling problem. This paper presents a new approach on maintenance scheduling of generating units in a generating company (GENCO). The proposed approach uses Markov decision process (MDP) to minimize total costs of unserved energy and reserve. The impacts on maintenance scheduling of different factors such as unit failure and variation in maintenance duration are determined. The proposed approach has less constraints compared to traditional maintenance scheduling methods. The method is then applied to a sample GENCO to investigate capabilities and limitations of proposed method