Multi-Industry Electricity Load Forecasting Based on LSTM Neural Network Modeling

Accurate electric load forecasting is crucial for the stable operation and long-term planning of power systems. To address the complexity and long-term variations in load forecasting across different industries, we propose a long-term multi-industry load forecasting method based on the long short-term memory (LSTM) neural network modeling. An industry-specific feature calculation module is designed that explicitly models key indicators such as daily average load. By integrating these features with the LSTM network, the proposed model effectively captures both short-term and long-term dependencies in load data. Simulation results demonstrate that the proposed model outperforms temporal convolutional network (TCN) and transformer models in industrial, commercial, and residential load forecasting scenarios, exhibiting superior robustness while significantly improving prediction accuracy and stability.