Recently, reinforcement learning (RL) has been widely applied to solving complex control problems. A major challenge in RL is balancing exploration and exploitation to maximize expected cumulative rewards. In highly complex control systems, both the action space and the state space often grow exponentially. Therefore, classical RL faces challenges in identifying optimal control strategies, resulting in reduced learning efficiency and an increased dependence on extensive computational resources and high-performance hardware. To address this problem, this study employs the PennyLane library to simulate quantum variational circuits (QVC), establishing a quantum reinforcement learning (QRL) model. Compared with classical RL models, QRL models leverage fewer trainable parameters and unique quantum physics characteristics, allowing these models to achieve reward maximization with fewer training episodes. The above advantage enhances the training efficiency for dynamic control tasks. In this study, the proposed algorithm is validated through the CartPole environment. The learning process is accelerated by designing a framework that combines the Proximal Policy Optimization (PPO) algorithm with a QVC. Furthermore, by incorporating the ε-greedy algorithm, the QRL model could achieve a more stable convergence behavior.