Abstract
This paper proposes a robust scheme for optimizing the power flow in a photovoltaic system. The scheme utilizes distributed saddle point dynamics and a decentralized approach to solve the power flow problem. It converts the convex optimization problem of the dynamic system control into the asymptotically stable dynamic systems and employs a linear approximation of power flow equations; specifically, a quadratic programming model is deployed with the aim of minimizing real-power losses to guarantee a globally optimal solution. Then, the photovoltaic inverters and electric networks are analyzed independently in a decentralized manner to exchange injection power among nodes while maintaining their independence to support the plug-and-play feature. A case study and the experimental results show that the proposed scheme achieves higher optimization accuracy and are more economical than the existing state-of-the-art schemes.