基于Docker容器虚拟化技术的人工智能学习系统设计  

Research on application of short video recommendation model based on collaborative filtering algorithm

作　　者：李长亮 李浩峰[2] LI Changliang;LI Haofeng(Zhejiang Zhoushan Tourism and Health College,Zhoushan Zhejiang 316111,China;Guizhou University,Guiyang Guizhou 550025,China)

机构地区：[1]浙江舟山群岛新区旅游与健康职业学院,浙江舟山316111 [2]贵州大学,贵阳550025

出　　处：《自动化与仪器仪表》2023年第12期124-128,133,共6页Automation & Instrumentation

基　　金：市厅级浙江省高等教育学会2022年度高等教育研究课题立项资助《“元宇宙”视角下的HyFlex混合弹性教学模式创新研究》(KT2022346)。

摘　　要：为进一步提升学校学习系统的智能化管理水平,基于Docker容器虚拟化技术构建一种人工智能学习管理系统。其中,设计的管理系统总共包含监测、资源预测、自动伸缩以及资源调度四个模块,并主要对其中的资源预测方法以及应用部署进行了详细设计。实验仿真结果表明,与双层LSTM模型以及Attention-LSTM模型相比,本研究提出的GRU-LSTM预测模型整体上能够取得更好的预测效果,在MAE、MSE、RMSE三个指标上,进行容器CPU的使用率预测结果误差分别为1.345、2.870以及1.683,在进行容器内存使用率上的预测结果误差分别为0.111/0.035以及0.179;在模块功能的测试上,本研究设计的学习管理系统性能良好且运行稳定。综上,本研究设计的基于Docker容器虚拟化技术的人工智能学习管理系统能够进行准确且稳定的系统管理,能够帮助管理人员进一步准确地掌握系统的各项信息,具有较强的实用性。To further enhance the intelligent management level of school learning systems,an artificial intelligence learning management system is constructed based on Docker container virtualization technology.Among them,the designed management system consists of four modules:monitoring,resource prediction,automatic scaling,and resource scheduling.The resource prediction methods and application deployment are mainly designed in detail.The experimental simulation results show that compared with the double-layer LSTM model and Attention LSTM model,the GRU-LSTM prediction model proposed in this study can achieve better overall prediction performance.The predicted errors of container CPU usage on MAE,MSE,and RMSE indicators are 1.345,2.870,and 1.683,respectively.The predicted errors on container memory usage are 0.111/0.035,and 0.179,respectively;In terms of module function testing,the learning management system designed in this study has good performance and stable operation.In summary,the artificial intelligence learning management system designed in this study based on Docker container virtualization technology can achieve accurate and stable system management,which can help managers further accurately grasp various information of the system and has strong practicality.

关 键 词：Docker技术 管理系统 资源预测 LSTM 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]