Ultra-short-term Cold Load Prediction Based on VMD-PSO-RBF

Abstract

Accurate prediction of cooling load can improve energy utilization, reduce energy consumption and reduce electricity costs, and has a significant impact on energy system planning and scheduling, as well as energy conservation and emission reduction. In view of the variable and nonlinear characteristics of cooling load data, the traditional BP neural network and RBF neural network methods cannot achieve better prediction results. Therefore, the method of combining the Variational Modal Decomposition method and the PSO-RBF method is used: Firstly, this paper uses the VMD. The load is decomposed, and then the PSO Optimized RBF Algorithm is used to predict the short-term load, and compared with the prediction results of BP and RBF neural network, the final Variational Modal Decomposition-Particle Swarm Optimization- Radial Basis Function has the best prediction effect and the smallest error. Therefore, the Variational Modal Decomposition-Particle Swarm Optimization- Radial Basis Function has a better effect on short-term power load and can effectively improve the accuracy of cooling load forecasting.