Short‐term load forecasting of the integrated energy system considering the peak‐valley of load correlations

Abstract

Abstract At different times, the load correlation of the integrated energy system (IES) is different and its changes over time are regular. However, the regularity of such changes is rarely considered in the current IES prediction research. In view of the above reasons, an IES short‐term load forecasting method based on load‐correlation peaks and valleys is proposed. Inspired by the concept of load peaks and valleys, the concept of load‐related peaks and valleys is put forward. Furthermore, based on the above concept, a method of establishing different prediction models within a day based on the correlation threshold is proposed. In order to implement the initial correlation threshold reasonably, a marine predator algorithm with an integrated gray wolf optimizer (MPAIGWO) is proposed. After the initial parameters are selected, use the self‐learning ideas in situational awareness to change the parameters. In the case study, the power, thermal, and cold loads of IES were predicted. The comparison results show that the prediction results of this method have high accuracy. In addition, this method can also overcome the influence of irrelevant input variables on prediction. The algorithm comparison proves that the MPAIGWO has the highest optimization performance under the same conditions.