Study On Extreme Value Of Random Process Of Non-stationary Non Gaussian Wind Load

With the improvement of national economic strength,the state has put forward higher requirements for infrastructure construction.Therefore,in order to effectively study the impact of extreme events and system risk assessment of stochastic dynamic systems,the extreme value research of stationary non Gaussian and even non-stationary non Gaussian stochastic processes has gradually attracted the attention of researchers.The extreme value analysis of various actions borne by engineering structures is an important basis for determining the action design value and the action evaluation value of service structures.For various load actions borne by the structure,including wind load,snow load,vehicle load,temperature action,earthquake action,etc.,the design code adopts the extreme value in the statistical sense as the design value of the action.For the safety evaluation of service structures,the actual action level should also be considered,and the possible extreme value should be estimated as the action evaluation value.In the random vibration analysis,the probability of extreme value of action and resistance is considered in the dynamic reliability analysis of random structure.However,the extreme value analysis of these actions is mainly based on the assumption that all kinds of actions are stationary Gaussian random processes,which is not necessarily consistent with the actual situation.Therefore,the main contents and research results of this paper are as follows:(1)According to the frequency relationship between fluctuating components and time-varying mean value in non-stationary wind speed and their cross-correlation characteristics,a judgment method of the best decomposition order of extracting time-varying mean value by wavelet transform is proposed,which avoids the uncertainty of selecting the best decomposition order for the measured random wind speed with unknown time-varying mean value.In this paper,the different frequency uncorrelation of random signals is proved theoretically.The wavelet transform method is used to decompose the time-varying mean value and pulsation components of the simulated non-stationary wind speed,and the uncorrelation of different frequencies is verified;Hilbert spectrum analysis,time-varying mean and pulsation component cross-correlation analysis and pulsation autocorrelation analysis are carried out for each decomposition result.By analyzing the value of the cut-off point of the attenuation rate of the autocorrelation curve of the pulsation component,the best decomposition order extracted by wavelet transform is found.The results show that each order decomposition of wavelet transform method first decomposes the pulsating components with high frequency and low amplitude of non-stationary random process.When the value of the cut-off point of the attenuation rate of the autocorrelation function of the separated pulsating components is closest to zero,the time-varying mean value extracted by the wavelet decomposition order is the best.Compared with the accurate time-varying mean value of simulated wind speed,the best decomposition order is the same,The accuracy of the discrimination method is verified.The method is applied to extract the time-varying mean of the measured wind speed,and an effective time-varying mean is obtained.(2)Due to the general characteristics of non-stationary non Gaussian bridge wind speed,in order to ensure the safe and stable operation of the bridge structure under the action of long-period wind speed,combined with the measured wind speed of the bridge,the non-stationary non Gaussian wind speed time history is transformed into stationary non Gaussian wind speed through the stationary transformation method,and then the Sadek simiu method is used to obtain the stationary non Gaussian extreme value under the specified guarantee rate,Finally,according to the inverse process of the stationary transformation method,the stationary non Gaussian extreme value is transformed into the non-stationary non Gaussian extreme value of wind speed.Considering whether to eliminate the time-varying mean or time-varying root mean square,the extreme wind speed of non-stationary non Gaussian bridge is studied.The results show that there is little difference between the wind speed extreme value obtained by eliminating the time-varying mean value and the original measured wind speed,and the wind speed extreme value obtained by eliminating the time-varying root mean square is mostly increased than the original wind speed extreme value.Therefore,the influence of the time-varying root mean square of the measured wind speed of the bridge on the wind speed extreme value of non-stationary non-Gauss bridge is much greater than the time-varying mean value.(3)In the study of extreme wind load,the field measured data is the most common.In this paper,the wind pressure data of two measuring point layout schemes of rectangular masonry structure are measured,and the non Gaussian characteristics of wind pressure data are analyzed.The peak factor method,improved peak factor method and Sadek simiu method are used to analyze the peak factor of non Gaussian wind pressure at each measuring point;Cook method and improved cook method are used to analyze the extreme wind pressure of measuring points arranged along the circumferential direction of masonry structure.The results show that the measured wind pressure measurement points of rectangular masonry structure show non Gaussian characteristics,and the non Gaussian characteristics of leeward surface show strong negative skewness;Compared with the peak factor method,the peak factor obtained by the improved peak factor method and Sadek simiu method changes more obviously with the non Gaussian characteristics of wind pressure time history,and the results are more valuable for reference;The results of multivariate wind pressure extremum calculated by cook method and improved cook method considering the joint distribution of wind pressure are larger than those of univariate wind pressure extremum.