Page 84 - 《软件学报》2026年第2期
P. 84
软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
2026,37(2):563−583 [doi: 10.13328/j.cnki.jos.007424] [CSTR: 32375.14.jos.007424] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563
*
云计算中基于 SAC 的多视角工作负载预测集成框架
曾文瑄 1 , 应 时 1 , 李田港 1 , 田相波 1 , 姜宇虹 2 , 刘虎杰 3 , 郝诗魁 3
1
(武汉大学 计算机学院, 湖北 武汉 430072)
(中国地质大学 (武汉) 计算机学院, 湖北 武汉 430078)
2
3
(湖北省楚天云有限公司, 湖北 武汉 430076)
通信作者: 应时, E-mail: yingshi@whu.edu.cn
摘 要: 工作负载的准确预测对于云资源管理至关重要. 然而, 现有预测模型通常使用固化结构从不同视角提取序
列特征, 导致不同模型结构之间难以灵活组合以进一步提升预测性能. 提出一种基于软演员-评论家算法 (soft actor-
critic, SAC) 的多视角工作负载预测集成框架 SAC-MWF. 首先, 设计一组特征序列构建方法来生成多视角特征序
列, 该方法能够以低成本从历史窗口生成特征序列, 从而引导模型关注不同视角下的云工作负载序列模式. 其次,
在历史窗口和特征序列上分别训练基础预测模型和若干特征预测模型, 以捕获不同视角下的云工作负载模式. 最
后, 利用 SAC 算法集成基础预测模型和特征预测模型, 生成最终的云工作负载预测. 在 3 个数据集上的实验结果
表明, SAC-MWF 方法在有效性和计算效率方面表现优秀.
关键词: 云计算; 工作负载预测; 强化学习; 多视角工作负载
中图法分类号: TP311
中文引用格式: 曾文瑄, 应时, 李田港, 田相波, 姜宇虹, 刘虎杰, 郝诗魁. 云计算中基于SAC的多视角工作负载预测集成框架. 软件
学报, 2026, 37(2): 563–583. http://www.jos.org.cn/1000-9825/7424.htm
英文引用格式: Zeng WX, Ying S, Li TG, Tian XB, Jiang YH, Liu HJ, Hao SK. SAC-based Ensemble Framework for Multi-view
Workload Forecasting in Cloud Computing. Ruan Jian Xue Bao/Journal of Software, 2026, 37(2): 563–583 (in Chinese). http://www.jos.
org.cn/1000-9825/7424.htm
SAC-based Ensemble Framework for Multi-view Workload Forecasting in Cloud Computing
1
1
1
3
2
1
ZENG Wen-Xuan , YING Shi , LI Tian-Gang , TIAN Xiang-Bo , JIANG Yu-Hong , LIU Hu-Jie , HAO Shi-Kui 3
1
(School of Computer Science, Wuhan University, Wuhan 430072, China)
2
(School of Computer Science, China University of Geosciences, Wuhan 430078, China)
3
(Hubei ChuTianYun Co. Ltd., Wuhan 430076, China)
Abstract: Accurate workload forecasting is essential for effective cloud resource management. However, existing models typically employ
fixed architectures to extract sequential features from different perspectives, which limits the flexibility of combining various model
structures to further improve forecasting performance. To address this limitation, a novel ensemble framework SAC-MWF is proposed
based on the soft actor-critic (SAC) algorithm for multi-view workload forecasting. A set of feature sequence construction methods is
developed to generate multi-view feature sequences at low computational cost from historical windows, enabling the model to focus on
workload patterns from different perspectives. Subsequently, a base prediction model and several feature prediction models are trained on
historical windows and their corresponding feature sequences, respectively, to capture workload dynamics from different views. Finally, the
SAC algorithm is employed to integrate these models to generate the final forecast. Experimental results on three datasets demonstrate that
SAC-MWF performs excellently in terms of effectiveness and computational efficiency.
Key words: cloud computing; workload forecasting; reinforcement learning; multi-view workload
* 基金项目: 国家自然科学基金 (62472329, 62072342); 国家重点研发计划 (2022YFB3304300)
收稿时间: 2024-08-22; 修改时间: 2024-10-24, 2025-01-21; 采用时间: 2025-03-03; jos 在线出版时间: 2025-08-20
CNKI 网络首发时间: 2025-08-20
