电子学报 ›› 2021, Vol. 49 ›› Issue (9): 1799-1808.DOI: 10.12263/DZXB.20190396

• 学术论文 • 上一篇    下一篇

复杂软件系统健康状态智能感知与诊断模型

王森1, 王煜2, 宁德军2   

  1. 1.同济大学电子与信息工程学院,上海 201804
    2.中国科学院上海高等研究院,上海 201210
  • 收稿日期:2019-04-15 修回日期:2020-12-31 出版日期:2021-09-25
    • 作者简介:
    • 王 森 男,1966 年8 月出生于上海. 教授级高工,主要研究方向为工业互联网、智能制造. E‑mail:wangsen@baosight.com
      王 煜 女,1994 年 1 月出生于山东临沂. 研究生,主要研究方向为软工数据智能、数据挖掘.E‑mail:wangyu02@sari.ac.cn
      宁德军(通信作者) 男,1972 年7 月出生于黑龙江省. 教授级高级工程师,CCF 会员. 长期从事下一代软件工程技术、大数据智能技术和海云协同计算研究. E‑mail:ningdj@ sari.ac.cn
    • 基金资助:
    • 上海市经济和信息委员会工业互联网创新发展专项资金项目 (2020-GYHLW-02010); 上海化学工业区公共事务中心上海化学工业区智慧决策平台项目 (E0420S1); 竞技体育高水平运动队人工智能辅助训练系统项目 (DQ200966-00)

Intelligent Perception and Diagnosis Model for Health Status of Complex Software System

WANG Sen1, WANG Yu2, NING De-jun2   

  1. 1.School of Electronics and Information Engineering, Tongji University, Shanghai 201804, China
    2.Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201210, China
  • Received:2019-04-15 Revised:2020-12-31 Online:2021-09-25 Published:2021-09-25
    • Supported by:
    • Program supported by Shanghai Municipality Commission of Economy and Informalization Industrial Internet Innovation and Development Project (2020-GYHLW-02010); Shanghai Chemical Industrial Zone Public Affairs Center - Shanghai Chemical Industrial Zone Intelligent Decision Platform Project (E0420S1); Program of AI Aided Sports Training System for High-level Competetive Sports Team (DQ200966-00)

摘要:

随着工业物联网和人工智能技术的迅猛发展,各种复杂软件系统(Complex Software System, CSS)日趋盛行,成为最重要的软件系统开发范式之一,其固有的成长性构造和适应性演化性质要求CSS必须能够实时感知和诊断自身的健康状态,确保其适应性演化过程中的质量. 本文采用特征工程和存储库数据挖掘技术,对影响开源CSS健康状态的特征进行分析,建立了一个数据驱动的实时、客观地反映开源CSS健康状态的自感知模型,并进一步借鉴质量控制图的思想,定义了能够辅助开源CSS故障诊断的自诊断模型. 最后,通过对比实验,证明了本文提出的模型因为全面综合了软件开发过程的绝大多数特征,能够更加全面和有效地评价软件的健康状态.

关键词: 复杂软件系统, 存储库数据挖掘, 自感知模型, 自诊断模型

Abstract:

With the rapid development of Industry Internet of Things and AI technology, various complex software systems (CSS) are becoming more and more popular, and becoming one of the most important software development paradigms. Its inherent growth construction and adaptive evolution require CSS to be able to perceive and diagnose its own health status in real time, so as to ensure the quality of its adaptive evolution. The paper uses feature engineering and mining software repositories (MSR) technology to analyze the features that affect the health of open source CSS, and establishes a data driven perception model that can reflect the health status of open source CSS in real time and objectively. Furthermore, a self-diagnosis model that can assist open source CSS fault diagnosis is defined with reference to the quality control chart. Finally, through the model comparison experiment, it is proved that our model can evaluate the health of software more comprehensively and effectively because it integrates most feature of software development process.

Key words: complex software system, mining software repositories, self-perception model, self-diagnosis model

中图分类号: