A Hybrid Clustering Process for Improving the Computational Efficiency of Automatic Analysis Stability Diagrams

In order to solve the problem of low efficiency of automatic analysis and calculation of stability diagram and realize real-time online identification of bridge modal parameters, an automatic analysis method of stability diagram based on hybrid clustering process is proposed. Firstly, the stability diagram of bridge structure is obtained by using Covariance-Driven stochastic subspace method; then, according to the recognition accuracy requirement, the stability diagram data are evenly divided into several frequency bands, so that the frequency similarity of the small class data and the order of distance matrix can be effectively reduced; finally, the graph theory clustering algorithm is used to automatically analyze the small class data to get the stable axis of the stability diagram. The proposed method is applied to Nanjing Yangtze River Fourth Bridge to verify the efficiency of the proposed method. The results show that the method based on hybrid clustering process can improve the computational efficiency of stability graph automatic analysis. The computational efficiency is much higher than that of pedigree clustering and graph theory clustering, and the algorithm has strong robustness.