结构性能试验建模中的统计不定性及其影响

Statistical uncertainty and its impact on the establishment of structural performance model by testing

结构性能的概率模型是设计中结构性能分析模型的基础,一般采用经典统计学中的矩法推断其中未知的计算模式不定性系数的概率特性.就一般的结构性能建模而言,其推断结果会受到显著的统计不定性的影响,并直接影响对结构性能概率特性的推断,存在因过高估计结构性能而导致额外失效风险的可能.论文根据一定置信水平下推断结果的相对误差,系统揭示了计算模式不定性系数推断中的统计不定性及其对结构性能推断结果的影响,并根据推断结果相对误差与样本容量之间的关系,提出需考虑统计不定性影响的条件以及试验建模中的最小试件数量.研究结果阐明了改进目前结构性能试验建模方法的必要性.

The probability model is the basis of performance analysis in structural design. In general, the classical statistical matrix method is used to infer the probability characteristic of unknown model uncertainty coefficient. In terms of structural performance modeling, the inference results can be significantly influenced by statistical uncertainty. Therefore the evaluating of probability characteristic of the structural performance is also influenced, which will lead to additional failure risk due to overestimation of the structural performance. In accordance with the relative error of inference results under a certain confidence level, this paper reveals the statistical uncertainty and its impact on the structural performance, points out the conditions that the statistical uncertainty need to be considered and the minimum number of specimens in the process of establishing the structural performance probability model. The conclusion shows it is necessary to improve current the modeling method in structural performance test.