Estimating Bounded Uncertain Model for Stability-Certified Reinforcement Learning