Exergy fuel cell micro combined heat and power systems using COA-HDNN hybrid technique

This manuscript introduces the COA-HDNN technique, which combines Hamiltonian deep neural networks (HDNN) with the crayfish optimisation algorithm (COA). The COA is to optimise energy management systems (EMS) for micro-combined heat and power (μCHP) systems with fuel cells (FC) and thermoelectric devices (TED). HDNN predicts energy demand and system performance metrics. Implemented in MATLAB, the COA-HDNN method outperforms existing approaches, including the salp swarm algorithm (SSA), particle swarm optimisation (PSO), and heap based optimisation (HBO). The results demonstrate its efficiency at 98%, cost of 2.6 × 105$, and computation time of 2.19 s, highlighting significant improvements in performance and operational efficiency.