Aiming at the problem that the global optimal of the day-ahead schedule for micro-grid is hard to convergence, a new optimal operation of micro-grid based on Bayesian and PSO algorithm is proposed, which combines the Bayesian Network (BN) theory with the Particle Swarm Optimization (PSO) algorithm from the perspective of probability network. Initially, the mathematical models of microgrid and the system constraints are established by considering the probability distribution of wind power and PV system, then the renewable factor and unit power production costs are introduced to achieve the optimization goal which contains the lowest total cost under the condition of energy conservation and emissions reduction. Finally, the simulation analysis comes from a typical microgrid system. The result shows that the BN-PSO algorithm can effectively solve the optimal operation problems of microgrid which includes random probability event, it’s a new way to solve such problems; when compared with the current mainstream algorithms, the BN-PSO algorithm can overcome the defect of local optimum and achieve convergence rapidly. This work is supported by the “Chunhui Plan” of Ministry of Education (No. Z2014076).