A reinforced learning control using iterative error compensation for uncertain dynamical systems