基于大数据的基桩高应变法检测合格率贝叶斯可靠度分析

Bayesian Reliability Analysis for Qualified Rate of Foundation Pile High Strain Dynamic Testing Based on Big Data

根据贝叶斯理论,本文提出基桩检测合格率概率分布模型,推导出其先验分布和后验分布均服从Beta分布,并引入基桩检测合格率可靠度指标作为评估基桩质量的标准。从由4305个工程、128651根桩的检测信息组成的历史检测大数据中,选取150个典型工程作为样本集对深圳某工程的高应变检测合格率进行可靠度分析,得到的可靠度指标大于目标可靠度指标,认为此桩基工程高应变检测合格。可靠度分析的引入,充分利用了历史检测大数据,克服了现有检测方法仅用一次检测数据进行质量评估的缺点,使得桩基工程质量评估更加科学合理,可为今后的工程实际和规范修编提供参考。

According to the Bayesian theory, this paper proposes a probability distribution model for the qualified rate of foundation pile testing. It deduces that both the prior distribution and posterior distributions are subject to the Beta distribution. The reliability index of the qualified rate of pile detections is introduced as the criterion of the quality assessment of the piles. From the historical big data comprising of 4305 engineering projects and 128651 pile tests, 150 typical projects are selected as the sample set to analyze the reliability of the qualified rate of high strain dynamic testing in a project in Shenzhen. It indicates that the reliability index is higher than the expected, hence the high strain test is qualified. Introducing the reliability analysis makes the most of the big data of historical detections and overcomes the shortcomings of the existing testing methods using only one test data for quality assessment, thus making the quality assessment of pile foundations more scientific and reasonable and furthermore providing some guidance for the future engineering practice and specification revision.