Predictive multi-energy management and optimization of renewable-integrated smart buildings with coupled electrical–thermal loads and hybrid energy storage

The integration of renewable energy sources with electrical and thermal energy storage provides an effective pathway toward improving the energy efficiency and sustainability of smart buildings. However, the coordinated management of electrical and thermal loads, intermittent renewable generation, and heterogeneous storage systems remains challenging because of their strong interactions, variable operating conditions, and associated economic and environmental constraints. This study develops a predictive multi-energy management and coordinated optimization framework for renewable-integrated smart buildings incorporating photovoltaic and wind generation, fuel cells, battery energy storage, and phase change material-based thermal energy storage. A Progressive Graph Convolutional Network (PGCN) is employed to capture spatial–temporal dependencies among interconnected energy components and generate short-term predictions of electrical demand, thermal demand, renewable generation, and storage states. The predicted energy profiles are subsequently integrated with the Black-Winged Kite Algorithm (BwKA), which optimizes the PGCN network parameters and supports coordinated power and thermal energy dispatch under the defined operating constraints. This combined prediction–optimization strategy enables the electrical and thermal subsystems, renewable sources, storage units, fuel cell, boiler, and utility grid to be coordinated within a unified energy management framework. The proposed approach is evaluated using MATLAB-based simulations under multiple system configurations and benchmarked against established approaches, including GA–BP, DRL–BO, GAN, and SVM. The results show that progressive integration and coordinated operation of renewable generation and storage substantially reduce utility-grid dependence while improving overall energy utilization. The fully integrated configuration achieves a system efficiency of 98.4%, an electricity cost reduction of 84.5%, an operational cost of 1.1 $, an LCOE of 0.4 $/kWh, and emissions of 42.6 ppm. Comparative results further demonstrate the effectiveness of the PGCN–BwKA approach in improving energy balancing, storage utilization, peak-load management, and coordinated electrical–thermal operation. Overall, the proposed framework provides a scalable simulation-based approach for predictive and coordinated multi-energy optimization in renewable-integrated smart buildings.