Real Time Dynamic Threshold Detection Algorithm Based on Local Weight and Application in Bridge Field

摘要

Discovering bridge structural anomalies accurately and eliminating hidden risks timely are of great significance to ensure bridge safety. Therefore, this paper proposed the Real Time Dynamic Threshold Detection Algorithm based on Improved Distance (RDTD). First, the Extended Frobenius Norm based on Local Weight (EFN_lw) is used to calculate the distance value of the multidimensional time subsequences; then, the Threshold Mechanism based on First Order Difference (TMFD) is designed to update the Normal Deviation Range (NDR) in time, and the anomaly judgment mechanism is to find outliers. The algorithm has the advantages of simple principle and small training sample size, at the same time, it can well take into account the requirements of real-time and accuracy. In order to verify the performance of the algorithm, two bridge engineering data sets in Shanghai are selected for test in this paper. The experimental results show that the algorithm can effectively find anomalies in a very short time while it has a certain promotion value and application value.