电信科学 ›› 2017, Vol. 33 ›› Issue (3): 163-167.doi: 10.11959/j.issn.1000-0801.2017058

• 电力信息化专栏 • 上一篇    下一篇

基于决策树的企业信息系统故障自动诊断分析方法

金鑫,闫龙川,刘军,张书林   

  1. 国家电网公司信息通信分公司,北京100761
  • 修回日期:2017-02-24 出版日期:2017-03-01 发布日期:2017-04-05
  • 作者简介:金鑫(1985-),女,国家电网公司信息通信分公司工程师,主要从事信息运维工作。|闫龙川(1979-),男,国家电网公司信息通信分公司高级工程师,主要从事信息运维工作。|刘军(1970-),男,国家电网公司信息通信分公司高级工程师,主要从事信息通信运维及管理工作。|张书林(1968-),男,国家电网公司信息通信分公司工程师,主要从事信息通信运维及管理工作。

Enterprise information system automatic fault diagnosis and analysis method based on decision tree

Xin JIN,Longchuan YAN,Jun LIU,Shulin ZHANG   

  1. State Grid Information & Telecommunication Branch, Beijing 100761, China
  • Revised:2017-02-24 Online:2017-03-01 Published:2017-04-05

摘要:

目前,大型企业信息系统规模和复杂度快速增长,但对故障的诊断分析仍主要依赖传统的人工经验,这不仅耗时、耗力,还影响对故障的及时处理。针对这一问题,创新性地提出了基于决策树的企业信息系统故障自动诊断分析方法,根据信息系统运行监控指标告警信息,实现对信息系统故障的自动诊断。利用某大型国有企业的实际生产运行数据,提取典型告警数据特征对该方法进行了验证,并在R语言环境下对决策树模型及其训练方法进行了仿真和对比分析。实验结果证明,该方法可以较为准确地实现故障自动快速诊断,有助于提高信息系统故障诊断分析效率。

关键词: 自动诊断分析, 信息系统故障, 决策树, R语言

Abstract:

With the rapid growth of the scale and complexity of enterprise information systems, traditional fault diagnosis and analysis methods relying on human experiences and manual operations cost more and more labor and time. To solve this problem, an automatic algorithm was proposed. The algorithm exploits information from system operation monitoring indicators and alarm data, based on decision tree, to automatically diagnose and analyze faults of enterprise information systems. The algorithm was verified, and the decision tree model and training method was simulated and analyzed comparatively under R language environment, using alarm data extracted from real operation data of a typical large-scale enterprise system. The experiment results show that this algorithm is able to achieve fast automatic fault diagnosis accurately, and is much helpful on improving efficiencies of information system fault processing.

Key words: automatic diagnosis analysis, fault of information system, decision tree, R language

中图分类号: 

No Suggested Reading articles found!