Cusp Anomaly Recognition Method for Structural Monitoring Data

Due to abnormal equipment or environmental noise, cusp anomaly data usually occurs in structural monitoring data. Manual handing is time-consuming, and cusp anomaly data disturbs the accuracy of alarm. Based on moving average filter and 3σ criterion, dual-window sliding filter was used for jump value anomaly recognition in structural monitoring. Abnormal data was translated to reasonable data by reasonable representative value of abnormal data based on 3σ criterion. The phenomenon of the jump of construction data was considered in moving average filter. Thus, misjudgment was solved. The proposed method was verified by real monitoring data in Hangzhouxi Railway Station. The results indicated that dual-window sliding filter could identify the abnormal jump point efficiently and avoid misjudgment by the phenomenon of the jump of construction data.