As the tidal current energy has become a significant replacement for conventional energy, the efficiency of the tidal current turbine is highly stressed. An optimal design of the turbine blades can greatly boost the energy conversion efficiency of the turbine. This study focused on the key technologies of hydraulic turbine geometric modeling and reconstruction. The blade element momentum theory was adopted to obtain the distribution of the twist angle and chord of blade, based on which a parameterized model of the original rotor was established. Further, the artificial neural network and the Genetic Algorithm (GA) were utilized to optimize the performance of the original blades. Meanwhile, the approximate model trained by Artificial Neural Network (ANN) was used as the objective function of GA optimization, which greatly improved the iterative speed. Then, an error reactive mechanism was established to ensure the accuracy of the performance prediction. According to numerical simulations based on Computational Fluid Dynamics (CFD), the energy conversion efficiency of the optimized rotor was higher than that of the original rotor under the blade-tip speed ratios considered. The efficiency was improved by 8.5% at most. And the thrust coefficient increased significantly in the optimized model.