A multi-level optimization design and intelligent control framework for fuel cell-based combined heat and power systems

Fuel cell systems have attracted significant attention in the field of residential energy due to their high efficiency and environmentally friendly characteristics. However, the inherent coupling of its thermoelectric output limits the flexibility of the system to meet diverse residential energy needs. This study proposes a combined heat and power system based on a proton exchange membrane fuel cell integrated with an organic Rankine cycle and heat pump, and builds a multi-level optimization design and intelligent control framework. Through this framework, current density and split ratio were identified as two key operational parameters affecting heat and power output. To enhance the precision and adaptability of system control, a neural network evaluation metric based on sensitivity weighting was introduced to optimize the hyperparameters of the Back Propagation neural network controller. This approach significantly improved the accuracy of the control model and system performance. Based on the optimized neural network controller, an intelligent control strategy oriented towards heat demand was realized, effectively meeting users' dynamic needs. Results show that under typical demand conditions, the system achieved significant performance improvement: maximum thermal efficiency of 47.48 %, maximum electrical efficiency of 36.73 %, maximum hydrogen consumption rate of 1.3 g/s, and minimum levelized cost of energy of 0.4183 $/kW & sdot;h- 1. This research provides valuable theoretical guidance for the optimization design and operations management of fuel cell-based combined heat and power systems.