| Performance-based seismic engineering theory is widely used in the practice of structural seismic design,which is based on the probability framework and taken the seismic risk as the goal of study.The main contents of this theory are hazard analysis,demand analysis,damage analysis and loss analysis.Ground motion intensity measures are the key link between seismic hazard and demand analysis,and also the key parameter to quantify the damage potential of earthquake ground motions.How to select a reasonable ground motion intensity measure is a very important research topic.From the view of uncertainty analysis,the selection of intensity measures with high correlation with engineering demand parameters will help to reduce the uncertainty of structural seismic demand prediction,so as to rationally quantify the uncertainty of RTR(record-to-record)by selecting a small number of seismic records,thereby reducing the computational resources needed for structural seismic demand analysis.Based on the above research background,the PEER NGA-West 2 ground motion database were selected as inputs,classification of ground motion records was carried by cluster technique,and the responses of single degree of freedom systems in terms of maximum displacement and hysteretic energy of the systems under each group of earthquakes were obtained.The correlation between ground motion intensity measures and engineering demand parameters were constructed by using the principal component analysis and canonical correlation analysis method in the multivariate statistical analysis.On this basis,the adaptive compound intensity measures were constructed by partial least square regression method.Based on the proposed compound ground motion intensity measures,probabilistic seismic hazard analysis,fragility analysis and risk analysis were carried out in Xi'an region.The main contents of this paper are as followings:(1)Cluster analysis based on genetic simulated annealing algorithm was used to group a suite of 2157 pairs of horizontal bidirectional ground motion records obainted from the PEER strong motion database.The preliminary ground motion records were clustered into four groups: the first group is large magnitude M long distance R;the second group is small magnitude M long distance R;the third group is small magnitude M short distance R and the fourth group is large magnitude M short distance R.Finally,the ground motion records of each group were classified according to the site type,and a comparative study of the acceleration response spectra of site ?,site ? and site ?? with respect to different clustering groups was conducted.A method of preliminary selection of ground motion records based on magnitude and distance clustering was proposed with consideration of site classification.(2)Based on the principal component analysis method,the single degree of freedom systems were used as the study objects,the ground motion records of different clustering groups were selected as inputs,and the six ground motion intensity measures were selected,the correlation between principal components of multivariate earthquake intensity measures and engineering demand parameter and its effect rule was carried out in order to help comprehensively and accurately understand the correlation between them.The results show that compared to the correlation between a single earthquake intensity measure and engineering demand parameter,the correlation between principal components of earthquake intensity measures and engineering demand parameter is more stable since it considers multivariate earthquake intensity measures.(3)In order to consider the effects of multivariate earthquake intensity measures and multivariate engineering demand parameters on the damage potential of earthquake ground motions,multivariate damage potential estimation of earthquake ground motions was proposed based on canonical correlation analysis.The canonical variable of earthquake intensity measures is then developed through maximumizing the correlation coefficient between earthquake intensity measures and engineering demand parameters,which can well predict the damage potential of earthquake ground motions and provide a new idea to quantify the damage potential of earthquake ground motions.The results show that the multivariate earthquake intensity measures can better estimate the damage potential of earthquake ground motions than a single earthquake intensity measure.(4)The partial least squares regression method was adopted to construct the compound ground motion intensity measures by the logarithmic linear combination of multivariate ground motion intensity measures,and the probabilistic seismic demand analysis was studied by considering compound intensity measures and engineering demand parameters.On this basis,the statistical performance evaluation of the compound intensity measures were investigated in order to help comprehensively and accurately predict structural responses.The results show that compared to a single earthquake intensity measure,the compound intensity measure is more efficient and more proficient,and the sufficient and robust of compound intensity measure is better.(5)Based on the adaptive compound spectral accelerations,Xi'an region was the study objects depend on the information of the Xi'an seismological bureau,the ground motion prediction model of compound spectral accelerations in Xi'an region was obtained,and the probabilistic seismic hazard analysis in Xi'an region was calculated by Monte Carlo simulation.On the basis of the deaggregation of probabilistic seismic hazard,the scenario earthquakes were obtained by the weighted method for the three structural periods at 0.2s,1.0s and 3.0s.The ground motions selected from the PEER strong motion database which are suitable to the site conditions of Xi'an region.The conditional target spectrum and a method of preferred selection of ground motion records based on the adaptive compound spectral accelerations were proposed.(6)The preferred selection of ground motion records were applied into the single degree of freedom system for incremental dynamic analysis.The displacement ductility coefficient was taken as the engineering demand measure,and the fragility curves at different displacement ductility coefficient levels were calculated.The probabilistic seismic risk analysis was obtained by convolution of probabilistic seismic hazard analysis and fragility analysis,and the probabilistic seismic risk analysis of adaptive compound spectral accelerations was realized based on the risk curves of displacement ductility coefficients.The results show that for probabilistic seismic risk analysis,the risk level of the compound spectral acceleration is less than a single spectral acceleration. |
PDF Full Text Request