基于ARIMA与GGACO算法的ETL任务调度机制研究  

Research on ETL Task Scheduling Mechanism Based on ARIMA and GGACO Algorithm

作　　者：周金治[1] 刘艺涵 吴斌[1] ZHOU Jinzhi;LIU Yihan;WU Bin(School of Information Engineering,Southwest University of Science and Technology,Mianyang 621000,China)

机构地区：[1]西南科技大学信息工程学院,四川绵阳621000

出　　处：《控制工程》2025年第2期208-215,共8页Control Engineering of China

摘　　要：随着抽取-转换-加载(extraction-transformation-loading,ETL)系统的ETL任务量增多,任务复杂度和波动性也随之提升,现有的ETL任务调度机制难以满足调度需求,如时间片轮转法受限于弹性调度能力弱、效率低下等缺点。为研究如何提升ETL任务调度机制的弹性调度能力以及执行效率,提出了一种基于整合移动平均自回归(autoregressive integrated moving average,ARIMA)模型与贪心-遗传-蚁群优化(greedy-genetic-ant colony optimization,GGACO)算法的ETL任务调度机制。初期,建立ARIMA模型并弹性地结合贪心算法计算初始解;中期,利用遗传算法的全局快收敛的特性结合初始解圈定最优解的大致范围;最后,利用蚁群优化算法的局部快速收敛性进行最优解搜索。实验结果表明:该调度机制能够弹性地指导任务调度尽可能地找到最优解,减少任务的执行时间,以及尽可能实现更高效的负载均衡。As the extraction-transformation-loading(ETL)task of the ETL system increases,the difficulty is more complicated and the fluctuation increases.The existing ETL task scheduling mechanism is limited by the shortcomings of the time film rotation method,which is difficult to meet the scheduling needs.In order to study how to improve the elastic scheduling capacity and execution efficiency of the ETL task scheduling mechanism,an ETL task scheduling mechanism based on the autoregressive integrated moving average(ARIMA)model and greedy-genetic-ant colony optimization(GGACO)algorithm is proposed.Initially,an ARIMA model and elastic combination of greedy algorithm are established to calculate the initial solution.In the mid-term,the global rapid convergence characteristic of the genetic algorithm is utilized to refine the initial solution and delineate the approximate range of the optimal solution.Finally,the local rapid convergence of the ant colony algorithm is employed to precisely search for the optimal solution.Experimental results demonstrate that the proposed scheduling mechanism can flexibly guide task scheduling to achieve near-optimal solutions,reduce task execution time,and realize more efficient load balancing.

关 键 词：弹性调度 ARIMA 贪心算法 遗传算法 蚁群优化算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]