To study the random scheduling problem of dispersed manufacturing resources in the context of cloud manufacturing, the service time is set as an uncertain production parameter, a random scheduling model is established, and an improved grey wolf algorithm combined with reinforcement learning is used to solve it to optimize the scheduling scheme. Starting from the perspectives of cloud platforms, users, and service providers, a multi-objective mathematical model based on uncertain environments is proposed. At the same time, the Grey Wolf Algorithm is combined with reinforcement learning, Q-learning is used to adjust parameters, and dual populations are explored and mined separately to prevent falling into local optima and to solve medium to large-scale instances. By conducting comparative expe-riments to evaluate the performance of the algorithm, it was found that the algorithm showed the best performance in all 16 cases, which can improve processing efficiency and increase stakeholder benefits. Sensitivity analysis was also conducted on the algorithm parameters to verify their rationality and applicability.