In order to solve the problem of accelerated degradation of cell performance in the photovoltaic electrolytic cell (PV-EC) direct coupling hydrogen production system due to frequent switching of cell array structure and high working current density

this study proposes an operation optimization strategy based on deep reinforcement learning (DRL). Firstly

the PV-EC direct coupling system model including photovoltaic cells and proton exchange membrane electrolyzers is established

and the energy utilization rate and the hydrogen production rate of the system are taken as the optimization objectives. Considering the attenuation characteristics of the electrolyzers and the uncertainty of photovoltaic power generation

the control problem is formalized as a Markov decision process (MDP). Secondly

a simulation platform including light irradiance prediction

photovoltaic power generation

and the electrolytic cell array is built in Python 3.7 environment

and the deep deterministic strategy gradient (DDPG) algorithm is used to train the agent and learn the optimal operation strategy of the system. The research results show that

compared with the existing control methods

the proposed strategy not only ensures the high energy utilization and hydrogen production rate of the system

but also effectively reduces the frequency of the electrolytic cell working at high current density

and reduces the number of array structure switching. This method has significant advantages in improving the comprehensive performance of the PV-EC direct coupling hydrogen production system and prolonging the life of the electrolytic cell.