Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (1): 255-267.DOI: 10.3778/j.issn.1002-8331.2007-0361

• Engineering and Applications • Previous Articles     Next Articles

Research on Accident Prediction Method of Special Equipment Based on Accident Characterization and CBR

GU Mengyao, LI Guanghai, DAI Zhixi   

  1. 1.College of Quality and Safety Engineering, China Jiliang University, Hangzhou 310018, China
    2.China Special Equipment Inspection and Research Institute, Beijing 100029, China
  • Online:2022-01-01 Published:2022-01-06

融合事故表征和CBR的特种设备事故预测研究

谷梦瑶,李光海,戴之希   

  1. 1.中国计量大学 质量与安全工程学院,杭州 310018
    2.中国特种设备检测研究院,北京 100029

Abstract: To better predict special equipment accident, the special equipment accident prediction research based on accident characterization and CBR(case-based reasoning) is proposed. Firstly, the general accident characterization technology for all kinds of special equipment is described in detail, which includes the structure model, standardization method and coding rule of accident characterization information. Then, the accident prediction method based on accident characterization and CBR is introduced. And in order to retrieve similar cases more accurately, an attribute similarity calculation method for different attribute types, an improved AHP method based on expert confidence, a case similarity calculation method considering time attenuation effect and a case similarity threshold determination function are proposed. Finally, this method is verified by a crane case of a company in Shanghai. The results show that the method can not only provide the fault tree of special equipment based on the accident precursory information, but also obtain the occurrence probability of accident characterization information like the accident characteristic and accident location, and the preventive measures to be taken.

Key words: special equipment, accident characterization, accident prediction, case-based reasoning

摘要: 为了更好地预测特种设备事故,提出了融合事故表征和CBR(case-based reasoning)的特种设备事故预测研究。详细说明了面向全类特种设备的通用型事故表征技术,包括事故表征信息的结构模型、规范化方法和编码规则。介绍了融合事故表征和CBR的特种设备事故预测方法,其中为了更精准地检索相似案例,提出了针对不同属性类型的属性相似度计算方法、基于专家置信度的改进型AHP法、考虑时间衰减效应的案例相似度计算方法以及案例相似度阈值确定函数。通过上海某公司的汽车起重机案例验证了该方法。结果表明该方法不但能根据事故前兆信息,给出特种设备的事故原因故障树,以了解事故发展趋势,还能提供如事故特征、事故发生部位等事故表征信息的发生概率,以及应采取的预防措施。

关键词: 特种设备, 事故表征, 事故预测, 案例推理