Due to computational limitations, generation expansion planning often selects representative days to ensure computational efficiency. The optimality of the generation expansion planning solution greatly depends on the accuracy of the selected representative days. At present, the selection of representative days mostly adopts historical data. This method is essentially to separate the selection of representative days from the planning. We get the planning results according to the selected representative days, and then operate the simulation under the representative days. This method is feasible without considering the renewable energy expansion planning, because the net load curve does not change before and after the planning. However, in the context of the increasing penetration of renewable energy, this method of selecting representative days cannot represent the actual operating situation of the system. The net load data before and after the planning changes, and the representative days before the planning cannot reflect the actual running situation after the planning. Based on the above questions, this paper has proposed a representative day selection method based on forward–backward sweep that provides three advantages: the target level of net-load curve is captured, the representative days can reflect the future situation in reality, and a linear relaxation method to enhance computational effectiveness is designed. This method can effectively reduce the error of traditional methods and the effectiveness has been demonstrated in XJTU-ROTS test system.
Full Text Finder