To optimize the trajectories of the sun tracking system, the dual-axis sun tracking control method for a photovoltaic power system is discussed in this paper. The optimization goal is maximization of an electric energy production in the photovoltaic system considering the tracking system’s consumption. Firstly, the elevation angle and azimuth angle trajectories are described as a nonlinear and bounded optimization problem. Then the predictable trajectories of elevation angle and azimuth angle are derived by using the known formulas and knowledge. However, many factors affect the tracking trajectories, such as machine accuracy, formula error, the partial shading due to the shadow of clouds, trees, birds, dirt, and buildings, etc. Firstly, a novel system structure is introduced to improve the electrical energy consumption. Then, the perturbation and observation (P&O) control method is proposed to improve the tracking trajectories by decreasing the errors of machine and formula by combining with the conventional predicted method. The tracking efficiency of the dual-axis sun tracking system using the predicted method is compared with that using the predicted method combined with the P&O method. The simulation results presented in the paper show that the proposed method has excellent tracking characteristics compared with the fixed title, single-axis tracking, and dual-axis tracking using the single predicted method.