Optimal power flow (OPF) is generally a non-convex optimization problem. The integration of unified power flow controller(UPFC) increases the nonlinearity of the original OPF problem. As a result, it is hard to obtain a global optimal solution using the traditional interior point method. Tree growth algorithm (TGA) is proved to be efficient and robust in solving complex engineering problems. TGA can globally search for optimal solutions to solve the OPF model that considers UPFC. In this model, the power generation cost, active power loss, and voltage deviation are included in the objective function, while the secure operation constraints of the power system integrated with UPFC are considered. Finally, numerical results on the IEEE 30-bus system and an actual system of Nanjing west ring network are carried out. The optimal solutions obtained from TGA, particle swarm optimization and interior point method are compared. At the same time, different heuristic algorithms are calculated 50 times using Monte Carlo method. These results verify the effectiveness of proposed approach and better accuracy and robustness of TGA.