Machine learning-based stability analysis and control strategy for multi-energy system with multi-port energy routers

With the increase of environmental pressure and rapid development of renewable energy technologies, countries around the world are trying to adjust their energy structures to reduce the dependence on traditional fossil fuels. The multi-energy system (MES) provides a new solution for optimizing energy supply, improving energy efficiency and ecological environment. Meanwhile, multiport energy routers (MER) have emerged within MES due to their capability to efficiently manage power in multi-directions across versatile energy systems. Hence, the control strategy of MER is essential for system’s operation. Furthermore, the large-scale applications of power electronics converter-based energy routers have significantly increased in MES, and it changed the characteristics of modern power system and presented significant challenges to the systems.

This project aims to propose a control strategy for multi-port energy router in multi energy system, where the deep reinforcement learning is used to deal with power coupling problem and enhance the capability of making use of renewable energy. Then the stability of proposed control strategy is analysed to ensure the system’s operation. Finally, ML based impedance identification method for MER is developed, which can recognize the controller characteristic and predict impacts of MER and control method on the stability of modern power system.