Multi-objective optimization method for power supply and demand balance in new power systems

The large amount of source and load uncertainty in new power systems poses challenges to the optimization of power supply and demand balance. The traditional optimization methods have not fully considered the uncertainty characteristics of different sources and loads. In this regard, a supply–demand balance optimization method based on ISAO-BiTCN-BiGRU-SA-IPBLS is proposed. Firstly, the ISAO algorithm is introduced into the hyperparameter optimization of BiTCN-BiGRU-SA, and the source and load interval prediction method based on LINMAP selection is proposed. Afterwards, a multi-objective optimization method for power supply and demand balance based on two-stage robust optimization is proposed. The first stage takes the daily planned output of adjustable power sources as the optimization variable, with daily operating cost, renewable energy delivery rate, and maximum loss in extreme scenarios as the optimization objectives. The second stage takes the daily operation of energy storage as the optimization variable and minimizes the maximum loss in the extreme scenario as the optimization objective. Finally, the method is applied to the county-level new power system in Hunan Province, China. The results show that the MAPE of the load and PV point prediction results in this work decreases by 13.43 % and 16.93 % after introducing the ISAO, respectively. Compared with the traditional Gaussian method, the Euclidean distance of error indicators between the load/PV interval prediction results in this work and the ideal results at an 85 % confidence interval decreases by 53.19 %/100 %. Compared with the traditional optimization method only considering economy, the work’s method improves the renewable energy delivery rate by 0.10 and 0.02 respectively, and reduces the maximum loss in extreme scenarios by 76.75 % and 3.62 % respectively on the maximum load day and maximum renewable energy output day.