计算机科学 ›› 2025, Vol. 52 ›› Issue (6A): 240700178-8.doi: 10.11896/jsjkx.240700178
李英健1, 王永生1, 刘晓君2, 任渊3
LI Yingjian1, WANG Yongsheng1, LIU Xiaojun2, REN Yuan3
摘要： 实时预测云平台监控收集的负载数据,有助于云运维中及早获取系统未来的性能趋势。但由于负载数据通常不具备明显的周期性或规律性,存在较多的噪声干扰,现有方法在特征学习规划上存在不足,需要依赖其他负载特征并且难以捕捉负载趋势的动量。为实现精准高效的负载数据预测,提出了一种基于时空图注意力网络的云平台负载数据预测方法。首先,运用改进经验小波变换对负载数据做时频域变换,降低噪声干扰并得到有效分解后的模态特征;为了提高模型处理尖峰和非周期性特征的能力,利用金融技术指标设计适合负载数据特性的关键性能因子;然后,将模态特征和关键性能因子与原始序列进行特征重构,构建图学习层;最后,利用图注意力网络动态捕获负载序列和特征之间的关系,并通过双向长短期记忆网络关注时间依赖信息。使用亚马逊和阿里云等负载数据集进行实验验证,结果表明,在4个数据集上,RMSE相比最优对比模型分别降低了13.44%,36.90%,7.41%和14.93%。
中图分类号:
| [1]BAO H Y,YIN K L,CAO L,et al.AIOps in Practice:Status Quo and Standardization[J].Journal of Software,2023,34(9):4069-4095. [2]DING Y,SANG N,LI X Y,et al.Prediction method of capacity data in telecom industry based on recurrent neural network[J].Journal of Computer Applications,2021,41(8):2373-2378. [3]WANG E X,WANG X H,ZHANG K,et al.Cloud computing load forecasting algorithm based on dual attention mechanism[J].Computer Engineering,2023,49(11):40-48. [4]CHEN L,WANG W,YANG Y,et al.A novel robust prediction algorithm based on REMD-MWNN for AIOps[J].Knowledge-Based Systems,2021,228:107038. [5]SHARIFIAN S,BARATI M.An ensemble multiscale wavelet-GARCH hybrid SVR algorithm for mobile cloud computing workload prediction[J].International Journal of Machine Lear-ning and Cybernetics,2019,10:3285-3300. [6]XIE T L,DENG L,YOU W L,et al.Analysis and Prediction of Cloud VM CPU Load Based on EMPC-BCGRU[J].Computer Science,2023,50(8):243-250. [7]YANG Z X,XIE X L,LI S W.Short Term Load PredictionModel for Cloud Resources Based on VMD-ISSA-LSSVM[J].Research and Exploration in Laboratory,2023,42(6). [8]DING J L,GONG Z H.Cloud KPI data prediction method based on combined attention model EAAT[J].Journal of Nanjing University of Information Science & Technology(Natural Science Edition),2023,15(6). [9]KARL M,MARTIN D,ENDA B,et al.Predicting host CPUutilization in the cloud using evolutionary neural networks[J].Future Generation Computer Systems,2018,86:162-173. [10]RAO S N,SHOBHA G,PRABHU S,et al.Time Series Forecasting methods suitable for prediction of CPU usage[C]//2019 4th International Conference on Computational Systems and Information Technology for Sustainable Solution(CSITSS).IEEE,2019:1-5. [11]TRAN N,NGUYEN T,NGUYEN B M,et al.A multivariatefuzzy time series resource forecast model for clouds using LSTM and data correlation analysis[J].Procedia Computer Science,2018,126:636-645. [12]CAO Z,DENG L,XIE T L,et al.Cloud Platform Task CPU Load Prediction Method Using n-LSTM[J].Journal of Chinese Computer Systems,2024,45(1):75-83. [13]HASAN SHUVO M N,SHAHRIAR MASWOOD M M ,ALHARBI A G.LSRU:A Novel Deep Learning based Hybrid Method to Predict the Workload of Virtual Machines in Cloud Data Center[C]//2020 IEEE Region 10 Symposium(TENSYMP). Dhaka,Bangladesh,2020:1604-1607. [14]HE X W,XU J J,WANG B,et al.Research on Cloud Computing Resource Load Forecasting Based on GRU-LSTM Combination Model[J].Computer Engineering,2022,48(5):11-17,34. [15]ZHOU H,ZHANG S,PENG J,et al.Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:11106-11115. [16]YANG Q L,JIANG L Y.Study on Load Balancing Algorithm of Microservices Based on Machine Learning[J].Computer Science,2023,50(5):313-321. [17]SHOORKAND H D,NOURELFATH M,HAJJIA.A hybridCNN-LSTM model for joint optimization of production and imperfect predictive maintenance planning[J].Reliability Engineering & System Safety,2024,241:109707. [18]GILMER J,SCHOENHOLZ S S,RILEY P F,et al.Neural message passing for quantum chemistry[C]//International Confe-rence on Machine Learning.PMLR,2017:1263-1272. [19]JOHANNES K,ALEKSANDAR B,STEPHAN G.Predict then propagate:graph neural networks meet personalized pagerank[C]//International Conference on Learning Representations.2018. [20]THOMAS N K,MAX W.Semi-supervised classification withgraph convolutional networks[C]//International Conference on Learning Representations.2017. [21]VELICKOVIC P,CUCURULL G,CASANOVA A,et al.Graph attention networks[C]//International Conference on Learning Representations.2018. [22]GUO H H,WU L Y,ZHAO Q S,et al.Day-ahead electricity price forecasting of power market based on trend index and long short term memory[J].Smart Power,2022,50(9):97-103. [23]ZHENG Y,GUAN S,GUO K,et al.Technical indicator en-hanced ultra-short-term wind power forecasting based on long short-term memory network combined XGBoost algorithm[J].IET Renewable Power Generation,2024:1-12. [24]WANG Q Q,DONG F,HE C H,et al.MACD algorithm-based pattern recognition of loading impact of PBAT waste water treated by AnMBR[J].Industrial Water Treatment,2024:1-7. [25]WANG J Z,BA R J,GE H,et al.Research on early-warning prediction model of critical slide of creep landslide based on the MACD index[J].Hydrogeology & Engineering Geology,2022,49(6):133-140. [26]Online.Trading Platform [DB/OL].(2022) [2024-01-10].https://github.com/Torchlight-ljj/AIOPSdataset. [27]Numenta.NAB Data Corpus [DB/OL].(2019) [2024-01-10].https://github.com/numenta/NAB. [28]Alibaba.Alibaba Cluster Trace Program [DB/OL].(2018)[2024-01-10].https://github.com/alibaba. [29]CloudWise.Cloud Wise OpenSource [DB/OL].(2021) [2024-01-10].https://github.com/CloudWise-OpenSource/GAIA-DataSet. |
|
||
首页