Scheduling generating units with ramping constraints using constructive dynamic programming

Individual thermal unit sub-problem with ramping constraints is very difficult to be solved because the generation levels of two consecutive hours are coupled. In this paper, an optimal method of solving this problem is presented. Compared to the standard dynamic programming, it does not need to discretize the state variable. The structure information of the system state space is fully used to obtain the recursive mapping relation of cost-to-go functions between two neighboring time intervals. Once cost-to-go functions across all hours are found in a backward sweep, the optimal strategy can be obtained in a forward sweep. Numerical tests show the efficiency and effectivity of this method.