摘要
软件在运行过程中会消耗资源,此过程存在两个问题,一是难以确定资源消耗发生变化的拐点,二是同一软件同一模块运行不同任务时,所产生的数据样本数量级差异过大。基于此,提出一种结合FCM和LSTM的算法研究软件运行资源变化规律的方法,利用FCM算法实现不同数量级样本间的群聚,接着把处理后的样本放入LSTM模型进行训练,进而得到资源消耗变化曲线。实验表明,通过资源消耗变化曲线能够确定拐点的类型和其出现的位置区间,进而找到软件运行资源变化规律。另外,通过对比分析,FCM-LSTM模型在解决此问题上的准确率高于其他同类型的传统算法。
Software consumes resources in the process of consumption.There are two problems in this process:one is that it is difficult to determine the inflection point of resource consumption change;the other is that when the same software and the same module run different tasks,the data samples generated are too different in order of magnitude.Based on this,a method combining FCM and LSTM algorithm is proposed to study the variation rule of software running resources.FCM algorithm is used to realize clustering among samples of different orders of magnitude,and the processed samples are put into LSTM model for training,so as to obtain the variation curve of resource consumption.The experiment shows that the type of inflection point and its position interval can be determined by the curve of resource consumption change,and the rule of software running resources change can be found.In addition,through comparative analysis,the accuracy of FCM-LSTM model is better than other traditional algorithms of the same type in solving this problem.
作者
李春生
王胡景
张可佳
富宇
LI Chunsheng;WANG Hujing;ZHANG Kejia;FU Yu(School of Computer&Information Technology,Northeast Petroleum University,Daqing 163319,China)
出处
《微型电脑应用》
2024年第3期1-6,共6页
Microcomputer Applications
基金
国家自然科学基金项目(51774090,61502094)
黑龙江省省属本科高校基本科研业务费东北石油大学引导性创新基金(2021YDL-12)
黑龙江省青年创新人才培养计划(UNPYSCT-2020144)。
