Multi-timescale optimization scheduling of integrated energy systems based on high-accuracy predictions

The fluctuation and randomness of energy present significant challenges to the secure and reliable operation of energy supply systems. To address this issue, a coordinated dispatch framework for multi-timescale energy prediction and optimization scheduling is proposed. Firstly, the predictive approach based on differential statistics and machine learning is employed to forecast the meteorological parameters and load at different stages, and suitable signal decomposition methods are selected for load forecasting in the advance-day and within-day stages. Subsequently, a multi-timescale optimization scheduling model for the integrated energy system, with time steps of 1 h, 15 min, and 5 min, is developed based on predicted data. Deviations between the costs of optimization scheduling based on forecasted values and actual circumstances are compared. Finally, the advantages of multi-timescale optimized scheduling schemes are investigated. The results show that the RMSE and R2 of the actual-time forecast results for the summer solar irradiance test set are 61.36 W/m2 and 0.973, respectively, and the MAPE of the load forecasts are close to 0. On a summer typical day, the multi-timescale optimal scheduling results in an increase in the proportion of renewable energy in the system by 8.11 %, and a reduction in the total cost by 5.10 %.