A Short-Term Load Forecasting Method Based on a Hybrid Model for Industrial Boiler Generator Sets

Abstract

Accurately predicting load variations is crucial for improving the energy efficiency and operational stability of industrial boiler generator sets. However, the current load scheduling of industrial boiler power plants heavily relies on manual experience, which hinders the optimization of system performance. This paper proposes a novel hybrid forecasting model, IFA-CNN-LSTM, which integrates data preprocessing, feature selection, and deep learning techniques. Firstly, the isolation forest algorithm is employed to detect and eliminate anomalies in the historical dataset. Then, key variables are identified by combining mechanistic analysis with spearman correlation. To address the variability of generator sets under diverse operating conditions, a hybrid neural network model combining convolutional neural networks and long short-term memory networks is constructed to capture both spatial and temporal features. The proposed model has been validated using real-world production data, and experimental results show that it outperforms traditional models in terms of forecasting accuracy.