In order to solve the problems of real-time and energy consumption for embedded multi-core system, an energy-saving scheduling algorithm based on particle swarm optimization is proposed. According to the analysis of task scheduling and energy consumption for multi-core processor system, a new coding strategy and the corresponding target function are established. Three processes of task partition, task scheduling and voltage selection of energy-efficient scheduling for multi-core processor system are integrated into a simple iterative optimization cycle of the particle swarm algorithm, and compatible with the system real-time, the original issues are transformed into new ones of constrained particle swarm optimization, and a constraint processing technology based on feasibility rules is introduced. Thus a new overall energy-saving scheduling algorithm is gained based on constrained particle swarm optimization. The simulated results show that the proposed algorithm can rapidly converge to the optimal scheduling solution of less energy consumption.