Skip to Main Content
In a space spanned by generation parameter, the different properties of loading margin surface and generation limit surface are investigated at first in this paper. Then, the shortage of taking the generation limit surface as limit boundary for generation rescheduling (GR) is elaborated and a developed GR model is presented. Additionally, in order to take the uncertainty of load pattern into consideration, a bilevel programming model with the opposite structure to each other is proposed. In this model, the upper level is used to optimize generation pattern (GP) for maximizing the loading margin while the lower level is used to determine the worst load pattern for checking the adaptability of the GP to the variation of the load pattern. A Step-by-step Approximation based Primal Dual Interior Point method (SAPDIP method) and Successive Linear Programming method (SLP method) are used to solve the two sub-models, respectively. The GP optimized by the proposed model reflects the balance between the “variation” and “adaption”, and is less sensitive to the changes in load pattern. Finally, the proposed method is tested on multiple standard IEEE test systems as well as a real power system in China.