This article investigates the attitude control problem for a novel foldable wave-energy powered autonomous underwater vehicle (FWEPAUV) with model uncertainties and external disturbances. To find a trade-off between the control accuracy and control energy consumption, a robust performance-prescribed optimized backstepping control (RPOBC) scheme is designed for optimal tracking problem by combining the reinforcement learning (RL) technique. The performance-prescribed mechanism is devised to ensure the prescribed tracking accuracy. To make the controller robust against the model uncertainties and external disturbances, an auxiliary system with a novel cost function is introduced into the optimized backstepping (OB) framework. Besides, the theoretical analysis indicates that the RPOBC scheme can guarantee tracking errors converge to predefined performance with an optimal control performance. At last, the simulation results are presented and analyzed to demonstrate the effectiveness of the designed RPOBC strategy.