A Cross-Domain Deep Learning Approach Using Sparrow and Tree Growth Algorithms for Cyber Threat Detection and Renewable Energy Forecasting

Abstract

The rapid proliferation of Internet of Things (IoT) devices and the simultaneous global expansion of renewable energy systems have introduced a critical need for intelligent, adaptive, and secure computational frameworks. This study proposes a cross-domain deep learning model integrating Sparrow Search Algorithm (SSA) and Tree Growth Algorithm (TGA) for dual objectives: detecting cyber threats in IoT environments and forecasting renewable energy generation. The hybrid model leverages deep learning’s nonlinear mapping capability with metaheuristic optimization’s adaptive exploration-exploitation dynamics, offering improved generalization across heterogeneous datasets. The SSA component enhances feature selection by identifying high-impact features that maximize classification performance in cyber threat detection, while TGA optimizes neural weights to improve prediction accuracy in renewable energy forecasting. Experimental evaluations on NSL-KDD and solar-wind energy datasets reveal the model’s superior accuracy, robustness, and convergence speed compared to conventional deep learning and standalone optimization methods. The proposed cross-domain approach not only enhances cybersecurity in IoT but also supports smart energy management, illustrating the power of hybrid intelligence in bridging two vital industrial sectors.