An Enhanced Proximal Policy Optimization-Based Reinforcement Learning Method with Random Forest for Hyperparameter Optimization