## Journal of Central South University

 第51卷    第6期    总第310期    2020年6月

[PDF全文下载]    [Flash在线阅读]

(1. 中南大学 资源与安全工程学院，湖南 长沙，410083；
2. 东南大学 土木工程学院，江苏 南京，211189
)

An inference method for probability distribution of extreme value engineering parameters based on normal information diffusion principle

1. School of Resources and Safety Engineering, Central South University, Changsha 410083, China;
2. School of Civil Engineering, Southeast University, Nanjing 211189, China

Abstract:There are a lot of extremum-types of engineering parameters in the reliability analysis. In order to obtain the optimal probability distribution of such parameters, an inferential method for the probability distributions of extremum-type of engineering parameters based on the normal information diffusion(NID) principle was proposed. Considering the truncated interval combining “3 ” statistical principle and the effect of skewness, the annual maximum standard wind pressure and friction angle were used as the engineering examples. The NID distribution and extremum-type distribution were used to deduce the corresponding probability distribution,and the K-S method was used to test the goodness of fit of the corresponding distributions. In order to investigate the influence of sample numbers on the fitting accuracy for two distributions,the Monte-Carlo method was used to generate eight groups of simulated samples,and the sample numbers were 15, 20, 30, 50, 100, 200, 500 and 1 000,respectively. And the corresponding K-S test method was also used to test the inferred probability distributions at different sample numbers. Combining the fitting advantages of the NID distribution, it was applied to the prediction of the maximum corrosion depth for nuclear pipelines engineering. The results show that whether for engineering samples or simulated samples, both the NID distribution and extremum-type distribution pass the test. The K-S test values of the NID distribution are always lower than those of extremum-type distribution and show great advantages in convergence with the increase of sample numbers. The cumulative probabilities of the NID distribution are equal to 1 at the fitting error magnitude of 10-4, while these extremum-type distributions are less than 1 at the same of fitting error magnitude. The NID distribution is used as a predictive model for the corrosion depth of nuclear pipelines,which can predict the maximum corrosion depth more accurately.

Key words: geotechnical engineering; reliability; extremum-type parameters; optimal probability distribution; normal information diffusion(NID) 中南大学学报（自然科学版） ISSN 1672-7207 CN 43-1426/NZDXZAC 中南大学学报（英文版） ISSN 2095-2899 CN 43-1516/TBJCSTFT