文章摘要
黄晓辉* **,崔莉**,黄希**.基于监督学习的规则触发执行预测方法研究[J].高技术通讯(中文),2021,31(2):113~121
基于监督学习的规则触发执行预测方法研究
Research on the rule trigger execution prediction method based on supervised learning
  
DOI:10. 3772/j. issn. 1002-0470. 2021. 02. 001
中文关键词: 针对物联网(IoT)规则引擎在实际应用中,由于外部、内部用户以及系统本身的不安全因素引起的控制及执行安全问题,提出了一种基于监督学习理论的规则触发执行设备控制预测方法(EasiPRED)。首先,制定了物联网规则引擎的规则统一描述标准,便于规则的设置和管理;然后,基于主成分分析(PCA)方法提出了一种特征提取算法,分别对规则样本集进行降维以及规则特征提取,不仅保留了原有的特征信息,同时也降低了计算复杂度;最后,基于监督学习方法对规则样本集进行训练学习,预测规则在触发执行后的控制类别,并根据预测结果与实际规则的输出做一致性验证,以此判断控制行为的安全性。实验结果表明,EasiPRED能够有效地识别出不安全的控制规则,提高系统的整体安全性能,避免发生控制及执行安全事故。
英文关键词: rules engine of Internet of Things (IoT), insecurity factor,control security, monitor learning, control category
基金项目:
作者单位
黄晓辉* **  
崔莉**  
黄希**  
摘要点击次数: 1769
全文下载次数: 1246
中文摘要:
      物联网(IoT)规则引擎; 不安全因素; 控制安全; 监督学习; 控制类别
英文摘要:
      Aiming at the control and execution security problems caused by external and internal users and the system itself in the practical application of Internet of Things (IoT) rule engine, a control prediction method of rule trigger execution device (EasiPRED) based on supervised learning theory is proposed. First of all, this paper has developed the unified description standard of Internet of Things rule engine, which is convenient for the establishment and management of rules. Then, a feature extraction algorithm based on principal component analysis (PCA) is proposed to reduce the dimensionality of rule sample sets and extract rule features, which not only preserves the original feature information, but also reduces the computational complexity. Finally, based on the supervised learning method, the rule sample set is trained and learned, the control category of the rule is predicted after the execution is triggered, and the consistency of the predicted result and the output of the actual rule is verified, so as to judge the safety of the control behavior.Experimental results show that EasiPRED can effectively identify unsafe control rules, improve the overall safety performance of the system, and avoid control and execution safety accidents.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮