To solve the task scheduling problem of Cyber-Physical Systems (CPSs),which are typically distributed heterogeneous parallel computing architecture, the Directed Acyclic Graph (DAG) is adopted as the scheduling model,the minimum task implementation time is taken as the object,and multi-mutation adaptive genetic algorithm is used.The effectiveness of individual gene is ensured though dependency matrix,and improved crossover and mutation operators,while the diversity of individual gene and convergence of the algorithm are ensured by using multi-mutation and adaptive methods.Simulation results show that the algorithm is more efficient than list-scheduling algorithm.