Neural network-based adaptive control and energy management system of a direct methanol fuel cell in a hybrid renewable power system

Abstract

Due to fluctuations in electricity demand and weather patterns, the solar energy-based system should be integrated with energy storage for efficient energy management. However, using only a battery cannot cope with the long-term power demand. Direct methanol fuel cells (DMFCs) have been considered one of the most promising backup power supplies for household applications. Therefore, the photovoltaic and solar thermal collector system incorporated with battery and DMFC is proposed along with the energy management system. In this study, a neural network-based adaptive control is developed to control the DMFC to meet persistent residential power demand and follow the technical requirements. Simulation results show that the neural network-based control outperforms the proportional-integral-derivative control in set-point change, disturbance, and model mismatch cases. Implementing the controlled DMFC in the hybrid system results in lower power requirements from the grid, which is less than twice that of a hybrid system without the DMFC. Furthermore, when solar power was insufficient to supply the load demand, the battery lifecycle in the system integrated with the DMFC seems more extended than that of a system without the DMFC.