基于多源数据协同感知的电气火灾预警算法研究  

Research on Electrical Fire Warning Algorithm Based on Multi-source Data Collaborative Sensing

作　　者：赵月爱[1] 郭兴原 ZHAO Yueai;GUO Xingyuan(Department of Computer Science and Technology,Taiyuan Normal University,Jinzhong 030619,China;Mathematics Department,Taiyuan Normal University,Jinzhong 030619,China)

机构地区：[1]太原师范学院计算机科学与技术系,山西晋中030619 [2]太原师范学院数学系,山西晋中030619

出　　处：《太原理工大学学报》2021年第6期907-912,共6页Journal of Taiyuan University of Technology

基　　金：国家自然科学基金资助项目(61273294);国家社科基金资助项目(20BJL080);山西省重点研发计划项目(201803D121088)。

摘　　要：在考虑电气数据的时序性与不同传感数据之间相关性的基础上提出了基于多源数据协同感知的电气火灾预警算法。该算法首先将传感数据用“时间序列”形式表示,利用改进的传感数据时序连续性检测算法计算单源传感数据相对离群距离,进而进行异常值检测;用多视图低秩分析法分析多源传感数据之间的关系进而识别出异常值。最后将检测结果进行实时有效的融合处理,从而得到更加准确的异常数据检测结果。实验结果表明,该算法异常检测率更高。A electrical fire warning algorithm based on multi-source data collaborative sensing was proposed,which takes into account the temporal characteristics of electrical data and the correlation between different sensing data.The relationship between multi-view and low rank analysis is used to identify outliers.Finally,the detection results are fused in real time and effectively to obtain more accurate abnormal data detection results.The experimental results proved the higher anomaly detection rate of the algorithm.

关 键 词：电气火灾 时序数据 多源数据 多视图低秩分析 相关性 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]