Artificial intelligence algorithms for short term scheduling of thermal generators and pumped-storage

The authors develop two algorithms for scheduling pumped-storage and thermal generators in a 24 hour schedule horizon based on the heuristic-guided depth-first search method. They differ primarily in the scheduling strategy for guiding the search process in determining the optimal schedule efficiently. In the first algorithm, the idea of the scheduling strategy is to schedule pumped-storage generations at peak load periods in such a way that the remaining load demand curve for thermal generation scheduling has a flattened peak region. In the second algorithm, the scheduling strategy allows pumped-storage generations to be scheduled at peak loads only when commitment of a thermal unit is required to meet the load demand. The operational constraints of the thermal and pumped-storage units together with the volume constraints of the reservoirs are fully taken into account in the algorithms. The effectiveness of the developed algorithms are demonstrated by applying them to a real life power system of 29 thermal units and two pump units.