A parametric study of a deep reinforcement learning control system applied to the swing-up problem of the cart-pole