Equivalent Linearization Design Method And Its Application In Seismic Isolation Systems

Earthquake disasters seriously threaten the safety of human life and property,and the base isolation system provides a very effective method of protecting structures against severe earthquake.The system usually shows strong nonlinear characteristics,and the calculation results of nonlinear time-history analysis are greatly affected by different seismic inputs.Using the response spectrum of the code to calculate can avoid the influence of different seismic waves on the calculation results.Seismic isolation systems usually show strong nonlinear characteristics.Because response spectrum of the code only considers the linear response of the structure,the equivalent linearization method is often used to simplify the calculation of the base isolation system.How to propose a reasonable equivalent linearization method and improve the calculation accuracy of this method has become the focus of research in this field.In this paper,the Bouc-Wen hysteresis model is selected for nonlinear time history analysis,and the parameter sensitivity ranking of each parameter under the harmonic response and seismic response is studied.Then the optimal equivalent linearization parameters(equivalent period and equivalent damping ratio)are obtained by using multiple population genetic algorithm.Next,through the training of artificial neural network,the equivalent linearization parameters prediction model of Bouc-Wen hysteresis model is established,and the accuracy of the prediction model is verified.Finally,the model is used to estimate the displacement response of base isolation bearing under earthquake.The specific research content and conclusions are as follows:(1)Using the global parameter sensitivity analysis method,considering the harmonic response and seismic response analysis results,the post-yield stiffness ratio a and the hysteretic loop shape control parameters β and γ are selected to calculate optimal equivalent linearization parameters.(2)Multiple population genetic algorithm is used to search for the equivalent period and equivalent damping ratio in two dimensions.The influence of single-degree-of-freedom system parameters and Bouc-Wen hysteresis model parameters on the optimal equivalent linearization parameters are studied.(3)The optimal equivalent linearization parameter prediction model is obtained by BP neural network,and the calculation results of the model are verified by the test sets.The optimal equivalent linearization parameters predicted by the model are brought into the linear system to calculate the maximum linear displacement,and compared with the maximum elastoplastic displacement calculated by nonlinear time history analysis,the error between the two is approximately within 10%.(4)The equivalent linearization method based on the displacement response spectrum is adopted.The seismic wave time history conversion spectrum and the response spectrum of the code are respectively used to calculate the maximum displacement of the base isolation system.The results show that the artificial neural network prediction model proposed in this paper has certain advantages in terms of accuracy and robustness.