Short-Term Power Load Forecasting Based on the NGO-Transformer-BiLSTM Model

Accurate short-term power load forecasting is critical for efficient power grid management, yet existing methods struggle with the complex and dynamic nature of electricity consumption patterns. Traditional models, including statistical and machine learning approaches, often fail to capture long-range dependencies and complex non-linear relationships. This paper introduces a novel approach by combining the Transformer architecture and BiLSTM with the North Gray Wolf Optimization (NGO) algorithm to improve forecasting accuracy. The proposed NGO-Transformer-BiLSTM model leverages the strengths of both the Transformer and BiLSTM to model long-range and bidirectional dependencies, while optimizing hyperparameters through NGO to enhance prediction performance. The model’s key innovations include the integration of Transformer and BiLSTM architecture, the application of NGO for hyperparameter optimization, and its ability to handle noisy and incomplete data. Experimental results demonstrate that the NGO-TransformerBiLSTM outperforms traditional and recent models in terms of MAE, MAPE, and RMSE, showing superior robustness and accuracy.