Atomatic Identification Of Bridge Structure Model Parameters Based On The Stochastic Subspace Identification

Bridge structural health monitoring system is an important security for the safe operation of existingbridge structures, as well as bridge structure modal parameter identification is one of the important parts of bridge health monitoring system.And the accurate identification of the parameters can response the inherent dynamic characteristics of the bridge structures, so it has a great sense to long-term health monitoring of bridge. However, the existing modal parameter identification algorithms cannot realize theadaptive decompositionof structural response signal and automatic identification of modalparameters, in that case, the existing ensembleempirical mode decomposition algorithm was improved, and the statistics of the hierarchical clustering algorithm were combined with stochastic subspace identification algorithm, in order to realize the automatic identification of modal parameters. The main work and achievements of thispaper are as follows.1. Summing up the main points of the basic theory and algorithms of the ensemble empirical mode decomposition algorithm, and making a systematic analysis and improvement accordingly of the deficiency existing in the algorithm. The main contents includes:the standard deviation of added white noise amplitude and the integration average times were determined on behave of the characteristics of un-decomposition signal themselvesautomatically; Introduction the statistics of statistical clustering analysis to prevent the modal aliasing phenomenon;establishing the new index of screening the effective IMF components in accordance with information entropy, energy density and the average period of each IMF component to realize the automatic sifting and signal reconstruction of the effective IMF components.2. Summing up the main points ofthe basic theory and algorithms of the stochastic subspace algorithm, put forward a corresponding improvement in accordance with the problems between system order selection and parameter identification which needs manul screening. The main contents includes:the order selection was finished by carryingon a logarithmetics treatment on the basis of the existing legal order singular values leap; achieved the purpose of modal parameters automatic identification by using of statistics in the hierarchical clustering algorithm for real modal filter; And the feasibility, effectiveness and applicability of automatic recognition algorithm was verified by using simulation examples of the simply supported girder bridge.3. Themodal parameters automatic identification of actual bridge structure should be adaptive decomposition and reconstruction byusing EEMDCAN decomposition algorithm, then identified its reconstructed signal, compared and analyzed the identified results with the results of actual parameters, in that case, the algorithm which proposed in this paper was verified that could be applied to the modal parameters automatic identification of real bridge structure.4.The corresponding modal parameters which in spring, summer, autumn and winter of a long-span cable-stayed bridge were identified respectively, and its specific change trends of frequency values in different seasons were analyzed, on the other hand, made a contrastive analysis of the differences of frequency values between day and night.