Probabilistic optimal power flow calculation method considering sequential features and correlations of multi-uncertain source-load scenarios

In order to accurately evaluate the influence of the correlation and sequential feature of wind power, PV power, and load on the economic operation of a power system, a sequential probabilistic optimal power flow calculation method based on multi-uncertain source-load scenarios generated by a multi-generative adversarial network (MGAN) is proposed. First, the historical datasets of wind power, PV power, and load in adjacent areas were collected, and their probability distributions in 24 time periods were analyzed. The results showed that the probability distribution of different periods had obvious differences, which were defined as time sequence differences in this paper. Second, a novel neural network called MGAN for scenario generation was designed. Based on the historical data of each time period, the correlated wind power, photovoltaic (PV) power, and load scenarios were generated by MGAN. Finally, in the modified IEEE 118-bus system, in which two wind power sources, two PV power sources, and two loads with correlation were integrated, probabilistic optimal power flow calculations were carried out for 24 periods with the goal of the lowest power generation cost of thermal power units. The experimental results showed that the power generation cost fluctuation ranges that considered the time sequence difference were reduced by 60% on average compared with the nonconsidered. The proposed method can effectively reduce the power generation cost budget.