Abstract: To strengthen practical teaching of reinforcement learning in automation specialty and improve students’ understanding and application of its basic principles, a simulation experimental teaching system is designed for a linear first-order inverted pendulum based on reinforcement learning control. Based on the MWORKS platform, a reinforcement learning environment is constructed for the linear first-order inverted pendulum system, including the state space, action space, and reward function. Typical reinforcement learning algorithms, including DQN, PPO, and DDPG, are implemented, and designed a human-machine interaction interface. The system provides functions such as parameter configuration, agent training, training process visualization, training result analysis, and control response display. It can be used to compare and analyze the cumulative rewards, stable running steps, and control response curves of different algorithms. The system supports qualitative and intuitive understanding as well as quantitative analysis and evaluation of reinforcement learning control, helps students master the basic principles of reinforcement learning and enhance their learning interest, and provides a reference for experimental teaching in intelligent control courses.