Research On High Arch Dam Structural Damage Identification Based Unscented Kalman Filter Method

At present,the extensive attention has been paid in the health monitor of oversize hydraulic structures.The essential task of health monitor is to analyze the change of stiffness and bearing capacity by measured structural response information,then the damage in structure can been identified.The difficulties are how to improve the efficiency and reliability of algorithm in this kind of inverse problem.This is just focus interested us.In this paper,it is aimed at developing a algorithm that can been effect applied in damage identification for hydraulic bam structure under model of Ertan high arc dam.The three-dimensional finite element model of Ertan arch dam is established.The nonlinear relationship between dam load and response is simulated by neural network,and the structure is damaged by unscented Kalman filter.(1)According to Ertan arch dam engineering data and related literature,the necessary parameters and boundary conditions of the model are determined.the finite element model of the dam is established through the ANSYS software.The static and dynamic response of the dam under combined load is analyzed,and the reliability of the model is verified.(2)Based on the neural network proxy model and 1 norm regularization method,an unscented kalman filter algorithm is proposed for high arch dam and other large hydraulic structures damage identification.The measure points are reasonably selected according to sensitivity The dam is divided into some sub-regions,and the reduction of elastic modulus in each sub-region is treated as damage parameter.The measure points are reasonably selected to the sensitivity of static displacement to damage parameters.The nonlinear relationship between temperature,reservoir water level,elastic modulus and static response of the structure is fitted by the proper neural network model.Finally,the damage state can be obtained by mean of optimization from unscented kalman filter algorithm.In order to solve the problem of ill-posedness in the process of the inverse problem,the regularization method is added to optimize the algorithm.Neural network has excellent nonlinear fitting ability.The proxy model by neural network is introduced to avoid the repeated calls of the finite element calculation program in the process of damage identification,So the computational efficiency can been sharply improved(3)Considering the change of temperature and the water level during the operation of the dam,the their influence to algorithm and results is analyzed.Numerical results show that the neural network model can accurately fit the nonlinear relationship between the dam response and the structural parameters.The unscented Kalman filter method based on the neural network can be effectively applied to the damage identification of the high arch dam and has superior stability and robustness.