Abstract: The paper first presents an objective model of task scheduling, and then based on the analysis of Q learning algorithm,the Markov decision process description of the scheduling problem is given.To solve the problem of slow update speed in Q learning, a multi-step Q learning scheduling algorithmis proposed, in which the value function is updated based on the infor mation in multiple steps.The application example indicates that this new approach can improve the convergence speed and effectively solve the task scheduling problem.