CLF-LDP Driven CO₂ Emission Forecasting for Smart Electric Vehicles

Abstract

Electric Vehicle (EV) networks play a crucial role in advancing environmental sustainability; however, inefficient charging and energy distribution strategies can contribute to increased COâ‚‚ emissions. This study proposes a hybrid Crow–Lévy Fuzzy Optimization combined with Linear Discriminant Prediction (CLF-LDP) framework to enhance emission forecasting accuracy and support energy-efficient decision-making in EV systems. The proposed approach integrates the global search capability of the Crow Search Algorithm (CSA), enhanced by Lévy flight exploration and fuzzy logic–based feature refinement, followed by discriminant-based predictive modeling. Experimental evaluation demonstrates that the CLF-LDP model achieves approximately 97% predictive accuracy, with a test R² of 0.90 under weighted training and 0.966 after regularization-based comparison. In contrast to high-capacity deep learning architectures that exhibited instability on unseen data, the proposed optimization-driven framework maintains consistent predictive robustness while preserving computational efficiency. Furthermore, integration of the predictive equation into a Hybrid Electric Vehicle (HEV) Simulink environment validates its applicability beyond offline regression analysis. By embedding COâ‚‚ prediction within the vehicle energy management system, the framework enables anticipatory emission-aware control rather than conventional rule-based adjustment. Overall, the CLF-LDP framework provides a scalable and computationally efficient solution that bridges predictive intelligence with real-time energy management, contributing to the development of sustainable and emission-aware smart transportation systems.