大规模联合软件异常监测模型算法仿真  

Simulation of Anomaly Monitoring Model Algorithm for Large-Scale Joint Software

作　　者：张秋红[1] 杨东芳 路新华[3] 

机构地区：[1]南阳理工学院计算机与信息工程学院,河南南阳473004 [2]黄河交通学院机电工程学院,河南焦作454950 [3]郑州大学信息工程学院,河南郑州450002

出　　处：《计算机仿真》2015年第6期356-359,共4页Computer Simulation

摘　　要：对联合软件异常区域监测的过程中,大量的软件衔接处,会出现故障交叉,混合、无特征关联等问题,传统的软件故障检测针对不同软件设计了交叉区,以区别故障软件种类。如果此区域过大,无法对联合软件进行准确的区域划分,从而对异常区域的故障检测准确率不高,提出一种基于SVM的联合软件异常监测方法,采用SVM分类技术,依靠信息几何优化,对软件的重叠区域进行数据依赖性的计算,对软件混合故障区进行重新定义,增强SVM用于联合软件环境下的软件异常区域分类效果,生成一个联合软件环境下的软件异常定义函数,融入松弛变量,使得训练过程中允许一定的误差存在,调整嵌入位数集合值的大小及其元素构成,调整软件异常区域监测方法的检测率及误测率。实验结果表明,提出方法具有较好的准确率。In the paper, an anomaly monitoring method for joint software was proposed based on SVM. Relying on the information geometry optimization, the overlapping area of the software was calculated to obtain data dependency. The mixed fault region of software was redefined to enhance the effect of SVM used in software anomaly area classifi- cation under joint software environment, and generate an abnormal defined function of software under joint software environment, in which the slack variables wee fused, to allow the existence of a certain error in training process. The size and element construction of embedded digits set value were adjusted, and the detection rate and mis - detect rate of the monitoring method for software anomaly region were also adjusted. The experimental results show that the pro- posed method has high accuracy.

关 键 词：联合软件环境 软件异常区域 监测模型 

分 类 号：TP221[自动化与计算机技术—检测技术与自动化装置]