A Novel Unsupervised Clustering Algorithm for Monitoring and Evaluating Bridge Structural Health
In recent years,accidents of urban elevated bridge collapse have occurred frequently due to vehicle over-loading,structural design defects,construction quality issues and other problems.Therefore,an efficient method has been proposed that can monitor the operation status of elevated bridges in real time,conduct real-time detection of structural damage by improved K-means clustering algorithm,and detect data-driven structural damage for the first time.This method mainly collects vibration data of steel structure bridge models under intact structural condi-tions.The effective structural damage sensitive feature values are extracted from these data by deep research.Finally,an improved unsupervised clustering algorithm is used to train the singular value detection model.The experimental results show that damage sensitive characteristic values under intact bridge structures serve as training data to train the mathematical model.It can effectively detect and identify the test results of bridge structures under various damage conditions.This new detection method can monitor real-time the structural health of urban elevated bridges during long-term operation.