For efficiently carrying out the “working procedure level” manufacturing resource scheduling from the perspective of cloud manufacturing service platform, considering not only the characteristics of distribution and diversity of manufacturing resource but also the constraints related to the service quality and the users’ subjective evaluation, the multi-objective optimization model was presented based on the four objectives: minimizing the makespan and the total cost, maximizing the quality of cloud manufacturing resource service and the users’ satisfaction. Then an improved non-dominated particle swarm optimization algorithm is designed for the model. In the algorithm, on the one hand the dynamic learning factor is introduced to guide the search scope and precision of the particles, so as to avoid falling into the local optimal solution, on the other hand the dynamic selection factor is introduced to improve the diversity of population, and so as to enable the algorithm converges to the optimal solution. Finally, the experiment results demonstrate the feasibility and effectiveness of the above model and algorithm.