The characteristics of randomness, running style, and unpredictability of user requirements in the cloud environment, brings great challenges to task scheduling. Meanwhile, the scheduling efficiency of cloud task allocation is an important factor affecting cloud resource systems. Therefore, this paper takes into account the characteristics of tasks, systems and users, a many-objective task scheduling model was constructed in cloud computing. In order to better solve the proposed many-objective task scheduling model, a reference vector guided evolutionary algorithm based on angle-penalty distance of normal distribution (RVEA-NDAPD) is proposed, and compared with the existing standard many-objective evolutionary algorithms (MaOEAs). Simulation results show that the algorithm can effectively improve the performance of the proposed model in cloud computing and obtain a suitable task allocation strategy.

在云环境中，随机性，运行方式和用户需求的不可预测性等特征给任务调度带来了巨大挑战。同时，云任务分配的调度效率是影响云资源系统的重要因素。因此，本文考虑了任务，系统和用户的特点，在云计算中构建了多目标任务调度模型。为了更好地解决提出的多目标任务调度模型，提出了一种基于正态角惩罚距离的参考矢量制导进化算法（RVEA-NDAPD），并与现有的标准多目标进化算法（MaOEAs）进行了比较。 ）。