Short-Term Load Forecasting Based on the IPSO-VMD and IGWO-BiGRU

Short-term load forecasting is a crucial component of power system operation and planning, and its accuracy is of great significance for ensuring grid security and improving energy efficiency. This paper proposes a short-term load forecasting model based on optimized Variational Mode Decomposition (VMD) and Bidirectional Gated Recurrent Unit (BiGRU). The aim is to improve forecasting accuracy by deeply mining the temporal features and modal information in load data. During the model construction process, to address the issue of VMD easily falling into local optima during the decomposition stage, an Improved Particle Swarm Optimization (IPSO) algorithm is introduced for parameter optimization, effectively enhancing decomposition effectiveness and computational efficiency. Meanwhile, to enhance the learning and forecasting capabilities of BiGRU, an Improved Grey Wolf Optimizer (IGWO) that incorporates adaptive inertia weight, Singer chaotic mapping, and nonlinear convergence factors is introduced to optimize BiGRU. This enables efficient exploration of complex solution spaces and global optimization. Training and testing with actual grid load data have verified the significant advantages of the optimized VMD-IGWO-BiGRU model in terms of forecasting accuracy and stability.