The Power Load Forecasting Model of Combined SaDE-ELM and FA-CAWOA-SVM Based on CSSA

Abstract

The security and stability of the power grid are directly affected by the accuracy of power load forecasting. Additionally, it plays an important role in power system planning. In order to enhance forecasting accuracy, a combined forecasting model is proposed in this paper. Firstly, preprocessing of the original data is conducted through improved singular spectrum analysis. Subsequently, load data prediction is carried out by the adaptive evolutionary extreme learning machine (SaDE-ELM). Additionally, load data prediction is performed using the support vector machine model(SVM), which is optimized by the chaotic adaptive whale algorithm based on the firefly disturbance strategy (FA-CAWOA-LSSVM). In the final step, the weight coefficients of the two prediction models are calculated by the chaotic sparrow search algorithm (CSSA). The load prediction results are obtained through the weighted summation of the two predictions. Superior performance is demonstrated by the combined prediction model compared with other single prediction models. The data preprocessing method, based on improved singular spectrum analysis, effectively enhances prediction accuracy.

Keywords

• Singular spectrum analysis
• combinatorial forecasting model
• power load forecasting