Generator maintenance scheduling via successive approximations dynamic programming

Host currently used thermal-generator maintenance scheduling methods schedule each unit at a time, leading in many cases to a poor local optimum or to an unfeasible solution. Considering simultaneously all units in the scheduling process, a computationally impractical method results. In this paper a piece-wise compromising method of solution is presented, which conveniently groups generating units and uses successive approximations dynamic programming. The optimal maintenance scheduling problem is formulated by employing a suitable state-space representation; most common constraints used in practice are described and a multiple objective function is presented. The effectiveness of the optimizing method is illustrated by solving a problem with data obtained from a real system. The grouping criterion presented appears to be the cornerstone of the superior performance of the proposed method.