As one of the important parts of bridge health monitoring system, the bridge structure modal parameters identification has great signification for the long-term health monitoring of bridge. However, the existing modal parameter identification algorithms cannot realize the adaptive decomposition of structural response signal and automatic identification of modal parameters. In that case, this article provides a brief introduction to the basic principle of stochastic subspace algorithm firstly; And then the sliding window was introduced in order to realize the on-line identification of time-varying structure modal parameters; And the system order algorithm was put forward based on the singular entropy theory that is aimed at the problem of ordering difficulty, so as to realize the automation system order determine; Finally, it was the hierarchical clustering algorithm in statistics combined with random subspace algorithm that to realize the automation identification of bridge structure modal parameters. At last, the proposed algorithm in this article was used to identify the modal parameters of the Sutong Bridge in vertical and transverse direction, and the identify results were analyzed comparing with the real value, the results show that this proposed algorithm can not only realize the automatically determine of system order, also can realize the automatic identifying of modal parameters as well as real-time monitoring of the bridge structure parameters, so that to real-time monitoring the running status of the bridge structure.