A Parallel Short-Term Power Load Forecasting Method Considering High-Level Elastic Loads

This article proposes an electric load forecasting model for systems containing high-level elastic loads. The model consists of three structures: a one-dimensional (1-D) convolutional network structure, a parallel forecasting structure, and a deep residual network (ResNet) structure. The 1-D convolutional network structure is utilized to extract features from the input data. The parallel forecasting structure is employed to predict the basic and elastic components of the loads. The deep ResNet structure is designed to enhance the model's generalization ability and prevent vanishing gradients. The proposed model is tested on the Independent System Operator-New England (ISO-NE) and Malaysia datasets, and the results demonstrate that the proposed model outperforms existing models and has high generalization ability.