An improved cutting plane consensus algorithm is adopted to optimize the optimal power flow problem in the active distribution network with high penetration of photovoltaic. Firstly

the power flow equation is approximated linearly

the node voltage is expressed linearly by node injection power

and hence a quadratic programming model including network loss and penalty cost of the curtailment of photovoltaics is constructed. Then

using Lagrange dual relaxation technology to decouple the centralized model

the distribution network and the photovoltaic inverters are taken as independent agents. For each agent

a master problem is built to approximate the original problem and only the cutting plane constraints are transferred between agents. On the premise of satisfying the coupling constraint between any two agents

the optimal solutions can be obtained for each agent. The agents are independent of each other and with good confidentiality. They do not need an upper level coordinator

and can converge to the optimal solution when the cutting plane information is occasionally lost. Finally

an actual rooftop photovoltaic system and a 33-bus system are tested to demonstrate the correctness and effectiveness of the proposed method.