实现的Sara强化学习算法，动作选择策略为贪心策略，且存在收敛条件和最终状态。(The implemented Sara reinforcement learning algorithm, the action selection strategy is a greedy strategy, and there are convergence conditions and final states.)