A Study On Data Level And Feature Level Fusion Method Of Flood Discharge Structure’s Damage Identification

In recently years, hydraulic flood discharge structure with large flow with high water-pressure draw high attention from hydraulic circles. Flood discharge structure influenced by the flow loading chronically during its service time. The vibration problems caused by it lead to the defect and damage of structure on different levels and sometimes worse enough affect on the normal running of hydro-junction. The establishment of the system of damage identification and structured behavior monitoring of dam flood discharge structure based on multi-source is conductive to monitor the catastrophic evolution of flood discharge structure systematically and keep the safe operation of dam. In the field of damage identification about hydraulic flood discharge structure which base on the information fusion method, the information fusion method fuse vibration response signs of several senors into signs which can reflect the structure dynamic characteristic more comprehensive and precise base on a certain rules and lays the foundation for extract the sensitive characteristics of flood structure damage. Moreover, the feature level fusion method focuses on the sensitive characteristics of several different kinds of damage. It fused in the feature level to gain the conforming description about the same damage and achieve the evaluation of the actual condition of flood structure. At present, study on information fusion method of flood structure’s damage identification is rare.Therefore, this paper mainly focuses on the study of two aspects as follow:Firstly, the study on the data level information fusion method of flood discharge multi-senors vibration signal. Aim at the feature level fusion’s highly demand on the similarity of vibration response signal and possess the fixity of fused coefficients and the he impact of environmental noise and measuring-point arrangement to the sensor test precision during the flood discharge vibration test of high arch dam structure, this paper put forward a level fusion method of high arch dam data multi-sensor vibration response which base on variance contribution rate. This method can keep the fused coefficients of each points at the same energy levels during dynamic allocation through identify the vibratory energy of each points’ signal at different time andamend the percentage of it. This paper also verify the feasibility of this method in practical engineering through numerical modeling test of a arch dam and the prototypical field tests of Er Tan high arch dam. With the comparative analysis with the fusion algorithm of correlation function, this method has obvious advantage on the information extraction of high arch dam vibration measurement and capable of better dynamic integration and applicability.Secondly, the study on the feature level information fusion method of flood discharge. Base on the flexibility and sensitivity dynamic damage identification method, this paper take the frame structure of hydraulic gate as research objects,identify the damage of single position, unsymmetrical position and symmetric position. It explored that this method has deviations and defects on the unit misjudgment and damage degree when it use to identify unsymmetrical position of symmetrical structure. Moreover, use the static damage identification method which base on the strain change rate still could not manage a perfect effectiveness. This thesis use the D-S theory, Bayes theory fuse sensitive characteristic of dynamic and static damage. Through the comparison validation, it shows that multi-source can reflect the overall characteristic more precisely and roundly than single-source.What’s more, the feature level information fusion method can apply to damage identification of hydraulic structure properly. Meanwhile, the rustle of fusion of D-S theory and Bayes theory are consistent with practical damage condition.