Inversion Of Dynamic Material Parameters Based On Particle Swarm Optimization

With the development of Chinese water conservancy and hydropower engineering, the height of many dams which have been built or are being built reach 200m or even 300m level. Most of these high dams are located in strong earthquake belt of southwest region. Once some accident happens to those dams in earthquake, disastrous consequences will be caused. Therefore, seismic safety is one of the main problems to be solved in the design of earth rock dam. At present, the seismic safety analysis of earth rock dam is usually based on the finite element method. And the dynamic model parameters are the main factors affecting the results of finite element analysis. The dynamic parameters of dam are determined by indoor dynamic tri-axial test or dynamic single shear test. But the dam material density in the construction is influenced by the construction technology, construction method and construction quality, which has difference with the density obtained in laboratory. And the density will affect the soil dynamic characteristics, so the dynamic parameters determined in laboratory are different from actual dynamic parameters. If the seismic analysis of dam is conducted with dynamic parameters obtained in laboratory, actual situation cannot be necessarily presented.Inversion analysis method provides a new way to identify the dynamic parameters of the dam which makes full use of the response information of dam in the earthquake. The material parameters are determined by the numerical calculation. However, there are many shortcomings about dynamic parameter inversion. For example, only a part of dynamic parameters are inversed and inversion research is based on two-dimensional finite element model without considering the difference between two-dimensional finite element model and the actual situation. In view of the above problems, acceleration response spectrum is adopted as the objective function in this paper. The material parameters are inversed by particle swarm optimization in 2D and 3D finite element model, which include dynamic shear modulus coefficient K2, dynamic shear modulus parameter K1, dynamic shear modulus index n and the maximum damping ratio. The influence of modulus decay curve and damping ratio on the dynamic characteristics is explored. MPI parallel particle swarm optimization improves the calculation speed and efficiency that ensures the feasibility of 3D model retrieval in time. Besides, the study of the effect of 2D and 3D stress state of Liyutan on the inversion results is achieved.