The Research Of Concrete Arch Dam Damage Localization Using Probabilistic Network

In China, thousands of dams have been constructed in past years. Damage may occur in the period of service of each concrete dam. It will influence the design function of hydropower station and seriously it will lead to the dam to destroy and bring disaster to the lives and properties of the people in the lower reaches, if these damages develop arbitrarily and can't be found timely. Therefore to these hydro-structure, it is necessary to analyse the damage and to assess safety condition.In this paper, the probabilistic neural network(PNN) is employed to locate the damage of concrete dam, which has overcome some drawbacks of the traditional method of safety monitoring. In this paper, expound the theory of probabilistic neural network, introduce the method and step of damage localization using PNN according to corresponding vibration frequency and shape, illustrate the foundation means of numerical simulation dam model and method of how to deal with the water and to simulate cracks, introduce the theory of how to calculate the vibration frequency and shape using finite-element method, then compare the vibration data of numerical calculation with actual measurement of Matikeng reservoir arch dam. Afterwards Xianghongdian concrete gravity-arch dam was analyzed with numerical simulation method and calculate the corresponding vibration frequency and shape on undamaged condition and damaged conditions, which has crack in different place with different damaged degree. For discussing the influence of noise, the identify results was gained by using probabilistic neural network from the data polluted by noise.The results show that it is possible to practically locate the damage in concrete arch dams by using PNN. Of course, the method is limited by environmental factor and precision of instrument when used in practice.