An Advanced Intrusion Detection and Energy Prediction Model Using TJO-Enhanced Feature Selection and Hybrid LSTM-DBN Architectures

Abstract

The increasing integration of Internet of Things (IoT) and Cyber-Physical Systems (CPS) has led to a rise in both cyber threats and energy inefficiencies, necessitating intelligent frameworks capable of handling intrusion detection and energy forecasting simultaneously. This paper introduces an advanced dual-function framework that leverages the Tree Growth Optimization (TJO) algorithm for enhanced feature selection and integrates a hybrid architecture combining Long Short-Term Memory (LSTM) networks and Deep Belief Networks (DBN). The TJO algorithm ensures the extraction of the most relevant and discriminative features for both anomaly detection and energy pattern prediction, thus reducing computational complexity while maintaining high accuracy. The hybrid LSTM-DBN model exploits the temporal dependencies of sequential IoT data and the hierarchical feature learning ability of DBN to achieve superior detection and forecasting performance. Experiments were conducted on benchmark IoT datasets, including NSL-KDD and an open-source Smart Grid dataset, to validate the model’s performance in terms of accuracy, precision, recall, F1-score, Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The proposed system achieved significant improvement over existing models, with intrusion detection accuracy reaching 99.12% and energy prediction MAE reduced by 17%. This study establishes a foundation for unified IoT security and energy management systems through the combination of intelligent feature selection and hybrid deep learning approaches.