Research And Modeling For Intensity Nonstationary Characteristics Of Three-Dimensional Ground Motion

Intensity nonstationary characteristics of earthquakes have great influences on the response of structures, expressing intensity nonstationary characteristics accurately is significance on structural anti-seismic analysis. Based on the inherent characteristics of earthquakes, a nonstationary model was built through statistical analysis, according to the engineering characteristics the parameter values with average and statistical meaning were obtained, then fully nonstationary artificial earthquakes fitting the target response spectrum were generated based on the parameter values. The main contents of this paper include:1. Collected and arranged earthquake records. Chose 20850 three dimensional records with obvious intensity nonstationary characteristics and complete site information from Japanese KiK-net ground motion database, after baseline correction, they were categorized according to magnitude, epicentral distance and site classification.2. Built three-dimensional intensity nonstationary model. All the parameter values were identified through genetic algorithm by making the cumulative energy of nonstationary model and the cumulative energy of records have the best approximation. The parameter values with statistical meaning were obtained through distribution fitting; the parameter values with average meaning were obtained through correlation analysis and attenuation relationship regression. Then the model was proved to be accurate and applicative.3. Studied on PGA characteristics of three-dimensional earthquakes. Studyed the PGA of three-dimensional records (3475 groups) by statistic analysis, it is found that the ratio of the smaller PGA value and the bigger PGA value accord with extreme value distribution at horizontal direction, the mean value is 0.80 for soil and 0.76 for bedrock; the ratio of the PGA value at vertical direction and the bigger PGA at horizontal direction accord with generalized extreme value distribution, the mean value is 0.50 for soil and 0.54 for bedrock.4. Generated fully nonstationary earthquakes. The difference of initial response spectrum and target response spectrum were expressed by difference spectrum, then some fully nonstationary artificial earthquakes were generated by using stochastic spectral representation method, their mean response spectra fitted the target one, and their fully nonstationary characteristics are close to recorded earthquakes.The intensity nonstationary model and its parameter values proposed in this paper can be used to express intensity nonstationary characteristics of given conditions, the distribution of parameter values will provide random nonstationary ground motion input for some random methods, and the model can provide theoretical principle for generating fully nonstationary earthquakes.