基于动态预测的深水爆炸试验容器可靠性分析

Reliability analysis of deepwater explosion test vessel based on dynamic prediction

为了确保深水爆炸试验容器在服役期间的安全性,提出了一种基于智能预测的随机-区间动态可靠性模型,通过动态测试数据建立了容器响应的广义回归神经网络(general regression neural network,GRNN)预测模型,获得了容器的最大应变区间变量,同时考虑容器结构的随机特性,开展了现役深水爆炸试验容器的可靠性分析,并分别采用3种方法进行了可靠性指标计算。分析结果表明,对于深水爆炸试验容器这类高可靠性且缺乏样本数据的结构,建立基于动态预测的混合可靠性模型,并通过区间计算可靠性指标的方法简便、可行;模型的区间变量随着结构动态测试数据的变化而变化,且对结构的不确定性分析也是动态的,因此得到的容器可靠性也随着其服役过程不断推进,具有动态特性,可以更好地反映容器在服役期间的性能变化,为容器的使用维护提供决策依据。

A deep-water explosion test vessel is an important test equipment which is filled with water to simulate different water depth environment by loading different hydrostatic pressure,and it can be used to study deep water explosion theory and engineering technology based on the similarity principle.In order to ensure the safety of vessels used in deep-water explosion test,a random-interval dynamic reliability model based on intelligent prediction is proposed in this paper.A GRNN prediction network of vessel response is established through dynamic test data,and the maximum strain interval variable of the vessel is obtained.Considering the random characteristics of the vessel structure,the reliability analysis of the in-service deep-water explosion test vessel is carried out.During the period,three methods are used to calculate the reliability index,and the analysis shows that for the vessel structure with high reliability and lack of sample data,the hybrid reliability model based on dynamic prediction is established by the calculation of interval reliability index.The method is simple and feasible.At the same time,the interval variables of the model change with the structural dynamic test data,and the uncertainty analysis of the structure is also dynamic.Therefore,the reliability of the container obtained is also changing with the service process,and has dynamic characteristics,which can better reflect the performance changes of the container during the service period and provide the basis for the use and maintenance decision of the container.