A new directional simulation method for system reliability. Part II: application of neural networks

A challenge in directional importance sampling is in identifying the location and the shape of the importance sampling density function when a realistic limit state for a structural system is considered in a finite element-supported reliability analysis. Deterministic point refinement schemes, previously studied in place of directional importance sampling, can be improved by prior knowledge of the limit state. This paper introduces two types of neural networks that identify the location and shape of the limit state quickly and thus facilitate directional simulation-based reliability assessment using the deterministic Fekete point sets introduced in the companion paper. A set of limit states composed of linear functions are used to test the efficiency and possible directional preference of the networks. These networks are shown in the tests and examples to reduce the simulation effort in finite element-based reliability assessment.     

Adaptive radial importance sampling under directional stratification

We establish radial importance sampling under directional stratification and construct its easy-to-implement algorithm for estimating the probability of failure in structural reliability analysis. The proposed algorithm is expected to run in a fully adaptive manner for averaging the increasing realizations towards the unknown stratum probability of failure, along with updating parameterized importance sampling and adjusting the allocation of computing budget only occasionally yet effectively, fully considering the decreasing stratum variances, all on a single set of replications. The formulation does not require a monotonicity condition on the radial distance in the polar coordinate system to justify a deterministic numerical procedure, such as the root finding of directional simulation in its standard form. A wide variety of numerical results were provided for illustrating the applicability and effectiveness of the proposed framework and algorithm.     

基于贝叶斯理论的非线性结构模型修正及其动力可靠度分析EI北大核心CSCD

提出一种基于贝叶斯推理的非线性结构模型修正方法,同时考虑激励的随机性,建立了复合随机振动系统的动力可靠度分析方法。利用实测结构动力响应主分量的瞬时特征参数作为非线性指标构建似然函数,结合拒绝延缓自适应(Delayed Rejection and Adaptive Metropolis,DRAM)算法和高斯过程替代模型实现了非线性结构模型修正及其参数的不确定性量化。根据首次超越破坏准则,利用广义概率密度演化方法,分别对仅考虑激励随机性的确定性模型和同时考虑结构参数与激励不确定性的复合随机振动模型进行动力可靠度分析,并利用蒙特卡洛随机抽样方法验证了计算结果的准确性。研究结果表明:基于振动响应瞬时特征参数的贝叶斯推理方法能够快速、准确地实现结构的非线性模型修正及其参数的不确定性量化。与具有初始设计参数名义值的确定性模型相比,考虑参数不确定性的复合随机模型的动力可靠度总体偏低,因此,在结构安全评估中应考虑非线性模型参数不确定性的影响,使评估结果更加安全、可靠。     

A load space formulation for probabilistic finite element analysis of structural reliability

A load space formulation for calculating the failure probability of complex structures for which the limit state functions are implicit is described in this paper. This formulation is used in conjunction with probabilistic finite element (PEE) analysis and employs a directional simulation to calculate the structural reliability. Apart from the advantage that a lower order space is used, the main advantage of the load space formulation proposed in this paper is that the number of inversions of the structural stiffness matrix and/or its gradients with respect to the material property random variables is reduced dramatically when compared with the usual Monte Carlo simulation (MCS) method. When used in a finite element reliability analysis, this procedure can save significant amounts of CPU time. Numerical examples are presented to show the efficiency and accuracy of the proposed approach.     

Assessment of reliability intervals under input distributions with uncertain parameters

Structural and mechanical reliability analysis often face the problem that probability distributions of the input variables are known with imprecision. This latter is normally specified by intervals of variation of their parameters. Leaving aside a crude Monte Carlo simulation consisting this case in estimating the failure probability for several sets of random realizations of the input distributions, there are no parsimonious methods for solving this problem in the general case of several interval parameters per distribution. In this paper a method intended to fill this gap is proposed. It is based on a property of the reliability plot recently proposed by the author [Hurtado, Dimensionality reduction and visualization of structural reliability problems using polar features. Probabilistic Engineering Mechanics, 29 (2012) 16–31], namely the fact that the order statistics of any function of the input random variables, used for building a limit state function, is concealed in the plot. This property, which is demonstrated herein, is used for the development of numerical methods for interval or reliability analysis, as well as for their combination for the estimation of the reliability interval. The ordering property of the plot assures that the lowest and largest values of the failure probability derives from samples contained in two small sets of realizations of the input distribution parameters located in specific plot sectors. The application of the proposed methodology is illustrated with examples that demonstrate its rigorousness, simplicity and accuracy.     

考虑场地效应的建筑群动力可靠度PDEM评估

城市建筑群的动力可靠度评估对于区域防灾减灾具有重要意义。场地效应会导致地震动的空间变异性和土-结构相互作用效应,进而显著影响区域建筑群的非线性地震响应,而确定性区域震害模拟方法仍不足以准确反映随机建筑群动力系统的整体性态。该研究引入实测地震动场和土-结构相互作用效应,发展了建筑群系统非线性地震响应时域求解方法,并遵循概率守恒假定,建立基于概率密度演化方法(PDEM)的建筑群系统动力可靠度评估框架。以某高层框架结构建筑群为例,进行了场地效应下确定性建筑群非线性地震响应分析,进而完成了随机建筑群动力可靠度评估,并评估了场地效应对建筑群系统动力可靠度影响,得到了有益结论。     

Stochastic Subset Optimization for optimal reliability problems

Reliability-based design of a system often requires the minimization of the probability of system failure over the admissible space for the design variables. For complex systems this probability can rarely be evaluated analytically and so it is often calculated using stochastic simulation techniques, which involve an unavoidable estimation error and significant computational cost. These features make efficient reliability-based optimal design a challenging task. A new method called Stochastic Subset Optimization (SSO) is proposed here for iteratively identifying sub-regions for the optimal design variables within the original design space. An augmented reliability problem is formulated where the design variables are artificially considered as uncertain and Markov Chain Monte Carlo techniques are implemented in order to simulate samples of them that lead to system failure. In each iteration, a set with high likelihood of containing the optimal design parameters is identified using a single reliability analysis. Statistical properties for the identification and stopping criteria for the iterative approach are discussed. For problems that are characterized by small sensitivity around the optimal design choice, a combination of SSO with other optimization algorithms is proposed for enhanced overall efficiency.     

An active learning Kriging model combined with directional importance sampling method for efficient reliability analysis

In this paper, the active learning Kriging model (ALK), which has been studied extensively in recent years, has been expanded by combining with the directional importance sampling (DIS) method. The directional sampling method can reduce the dimensionality of the variable space by random sampling or interpolation in the direction of vector diameter, which can improve the efficiency of reliability analysis. It is especially suitable for the surfaces whose limit state is spherical or near-spherical. By introducing the control coefficient and constructing the directional importance sampling density function, the sampling efficiency can be further improved in the design point domain. A novel reliability analysis method called ALK-DIS method is proposed. The greatest advantage of the proposed method is its ability on great computational efficiency and dealing with small failure probability problem In addition, due to the excellent performance of directional sampling method in dealing with multi-failure model reliability problems, the ALK-DIS method has the advantage of being applied to system reliability analysis in this paper successfully. The applicability, feasibility and efficiency of the proposed method are proved on examples which contain linearity equation, non-linear numerical example, non-linear oscillator and system reliability engineering problems.     

Dynamic response and reliability analysis of structures with uncertain parameters

An original approach for dynamic response and reliability analysis of stochastic structures is proposed. The probability density evolution equation is established which implies that incremental rate of the probability density function is related to the structural response velocity. Therefore, the response analysis of stochastic structures becomes an initial‐value partial differential equation problem. For the dynamic reliability problem, the solution can be derived through solving the probability density evolution equation with an initial value condition and an absorbing boundary condition corresponding to specified failure criterion. The numerical algorithm for the proposed method is suggested by combining the precise time integration method and the finite difference method with TVD scheme. To verify and validate the proposed method, a SDOF system and an 8‐storey frame with random parameters are investigated in detail. In the SDOF system, the response obtained by the proposed method is compared with the counterparts by the exact solution. The responses and the reliabilities of a frame with random stiffness, subject to deterministic excitation or random excitation, are evaluated by the proposed method as well. The mean, the standard deviation and the reliabilities are compared, respectively, with the Monte Carlo simulation. The numerical examples verify that the proposed method is of high accuracy and efficiency. Moreover, it is found that the probability transition of structural responses is like water flowing in a river with many whirlpools, showing complexity of probability transition process of the stochastic dynamic responses. Copyright © 2004 John Wiley & Sons, Ltd.     

脉动风速连续随机场的降维模拟

基于标准正交随机变量的波数谱表示,通过定义标准正交随机变量集的随机函数形式,建立了连续时空随机场模拟的波数谱-随机函数方法。同时,引入快速傅里叶变换（FFT）的算法,极大地提高了波数谱-随机函数方法的模拟效率。在波数谱-随机函数模拟方法中,仅需两个基本随机变量即可在概率密度层次上描述时空随机场的概率特性,并利用数论方法选取基本随机变量的代表性点集,实现对连续时空随机场模拟的降维表达。数值算例表明,当模拟相同数量的样本时,综合考虑模拟的效率和精度两方面,该文方法与传统的波数谱表示方法不分伯仲,但该文方法所需的基本随机变量最少,生成的代表性样本数量少且构成一个完备的概率集,从而可结合概率密度演化理论实现结构随机动力反应及动力可靠度的精细化分析。最后,结合Kaimal风速谱及Davenport空间相干函数模型,模拟了水平向脉动风速连续随机场,验证了该文方法的有效性和优越性。     

Adaptive directional stratification for controlled estimation of the probability of a rare event

Within the structural reliability context, the aim of this paper is to present a new accelerated Monte-Carlo simulation method, named ADS, Adaptive Directional Stratification, and designed to overcome the following industrial constraints: robustness of the estimation of a low structural failure probability (less than 10⁻³), limited computational resources and complex (albeit often monotonic) physical model. This new stochastic technique is an original variant of adaptive accelerated simulation method, combining stratified sampling and directional simulation and including two steps in the adaptation stage (ADS-2). First, we theoretically study the properties of two possible failure probability estimators and get the asymptotic and non-asymptotic expressions of their variances. Then, we propose some improvements for our new method. To begin with, we focus on the root-finding algorithm required for the directional approach: we present a stop criterion for the dichotomic method and a strategy to reduce the required number of calls to the costly physical model under monotonic hypothesis. Lastly, to overcome the limit involved by the increase of the input dimension, we introduce the ADS-2⁺ method which has the same ground as the ADS-2 method, but additionally uses a statistical test to detect the most significant inputs and carries out the stratification only along them. To conclude, we test the ADS-2 and ADS-2⁺ methods on academic examples in order to compare them with the classical structural reliability methods and to make a numerical sensitivity analysis over some parameters. We also apply the methods to a flood model and a nuclear reactor pressurized vessel model, to practically demonstrate their interest on real industrial examples.     

A semi-analytical simulation method for reliability assessments of structural systems

A semi-analytical simulation method is proposed in this paper to assess system reliability of structures. Monte Carlo simulation with variance-reduction techniques, systematic and antithetic sampling, is employed to obtain the samples of the structural resistance in this method. Variance-reduction techniques make it possible to sufficiently simulate the structural resistance with less runs of structural analysis. When resistance samples and its moments determined, exponential polynomial method (EPM) is used to fit the probability density function of the structural resistance. EPM can provide the approximate distribution and statistical characteristic of the structural resistance and then the first-order second-moment method can be carried out to calculate the structural failure probability. Numerical examples are provided for a structural component and two ductile frames, which illustrate the method proposed facilitates the evaluation of system reliability in assessments of structural safety.     

Reliability estimation and prediction of multi-state components and coherent systems

This paper discusses the multi-state coherent system composed of multi-state components. First, using the min cut sets or min path sets, we present our simulation algorithm, instead of the general structure function, to calculate the probability that the system is in a specified state. Second, we check the components per period, e.g. one check per year, to obtain the state sequences of all components. When the state sequences are Markovian chains, we can predict the reliability of the components in several periods, such as the probability that the components are in specified states. Also, we give two methods to compute the system reliability in a number of periods: one employs the states of the components in these periods, which can be predicted by the state transition probability matrixes of the components; the other uses the state transition probability matrix of the system obtained by the simulated states of the components.     

Robust compliance topology optimization based on the topological derivative concept

In this paper, we present an approach for robust compliance topology optimization under volume constraint. The compliance is evaluated considering a point‐wise worst‐case scenario. Analogously to sequential optimization and reliability assessment, the resulting robust optimization problem can be decoupled into a deterministic topology optimization step and a reliability analysis step. This procedure allows us to use topology optimization algorithms already developed with only small modifications. Here, the deterministic topology optimization problem is addressed with an efficient algorithm based on the topological derivative concept and a level‐set domain representation method. The reliability analysis step is handled as in the performance measure approach. Several numerical examples are presented showing the effectiveness of the proposed approach. Copyright © 2015 John Wiley & Sons, Ltd.     

Compromise design of stochastic dynamical systems: A reliability-based approach

This paper presents a procedure for obtaining compromise designs of structural systems under stochastic excitation. In particular, an effective strategy for determining specific Pareto optimal solutions is implemented. The design goals are defined in terms of deterministic performance functions and/or performance functions involving reliability measures. The associated reliability problems are characterized by means of a large number of uncertain parameters (hundreds or thousands). The designs are obtained by formulating a compromise programming problem which is solved by a first-order interior point algorithm. The sensitivity information required by the proposed solution strategy is estimated by an approach that combines an advanced simulation technique with local approximations of some of the quantities associated with structural performance. An efficient Pareto sensitivity analysis with respect to the design variables becomes possible with the proposed formulation. Such information is used for decision making and tradeoff analysis. Numerical validations show that only a moderate number of stochastic analyses (reliability estimations) has to be performed in order to find compromise designs. Two example problems are presented to illustrate the effectiveness of the proposed approach.     

Local estimation of failure probability function by weighted approach

In the reliability-based design of engineering systems, it is often required to evaluate the failure probability for different values of distribution parameters involved in the specification of design configuration. The failure probability as a function of the distribution parameters is referred as the ‘failure probability function (FPF)’ in this work. From first principles, this problem requires repeated reliability analyses to estimate the failure probability for different distribution parameter values, which is a computationally expensive task. A “weighted approach” is proposed in this work to locally evaluate the FPF efficiently by means of a single simulation. The basic idea is to rewrite the failure probability estimate for a given set of random samples in simulation as a function of the distribution parameters. It is shown that the FPF can be written as a weighted sum of sample values. The latter must be evaluated by system analysis (the most time-consuming task) but they do not depend on the distribution. Direct Monte Carlo simulation, importance sampling and Subset Simulation are incorporated under the proposed approach. Examples are given to illustrate their application.     

Natural disaster risk analysis for critical infrastructure systems: An approach based on statistical learning theory

Probabilistic risk analysis has historically been developed for situations in which measured data about the overall reliability of a system are limited and expert knowledge is the best source of information available. There continue to be a number of important problem areas characterized by a lack of hard data. However, in other important problem areas the emergence of information technology has transformed the situation from one characterized by little data to one characterized by data overabundance. Natural disaster risk assessments for events impacting large-scale, critical infrastructure systems such as electric power distribution systems, transportation systems, water supply systems, and natural gas supply systems are important examples of problems characterized by data overabundance. There are often substantial amounts of information collected and archived about the behavior of these systems over time. Yet it can be difficult to effectively utilize these large data sets for risk assessment. Using this information for estimating the probability or consequences of system failure requires a different approach and analysis paradigm than risk analysis for data-poor systems does. Statistical learning theory, a diverse set of methods designed to draw inferences from large, complex data sets, can provide a basis for risk analysis for data-rich systems. This paper provides an overview of statistical learning theory methods and discusses their potential for greater use in risk analysis.     

基于计算智能技术的结构系统可靠性优化设计

利用随机摄动和Edgeworth级数方法,将非正态随机参数可靠性优化设计中的概率约束转化为等价的确定性约束,并运用粒子群算法迅速准确地获得结构系统可靠性优化设计的初始点。针对多失效模式的结构系统,提出了随机模拟-小波神经网络方法(MCS-WNN),有效解决了结构系统的可靠性仿真。并提出了一种便于逆映射的小波神经网络模型,实现了设计参数的可靠性优化。实验结果表明上述方法是行之有效的。     

基于概率密度演化理论的结构抗震可靠性分析

运用随机过程的正交展开方法,将地震动加速度过程表示为由10个左右的独立随机变量所调制的确定性函数的线性组合形式。结合概率密度演化方法和等价极值事件的基本思想,研究了非线性结构的抗震可靠度分析问题。以具有滞回特性的非线性结构为例,对某一多自由度的剪切型框架结构进行了抗震可靠性分析。结果表明：按照复杂失效准则计算的结构抗震可靠度较之结构各层抗震可靠度均低。这一研究为基于概率密度函数的、精细化的抗震可靠度计算提供了新的途径。     

Partition of the probability-assigned space in probability density evolution analysis of nonlinear stochastic structures

Based on a partition of probability-assigned space, a strategy for determining the representative point set and the associated weights for use in the probability density evolution method (PDEM) is developed. The PDEM, which is capable of capturing the instantaneous probability density function of responses of linear and nonlinear stochastic systems, was developed in the past few years. The determination of the representative point set and the assigned probabilities is of paramount importance in this approach. In the present paper, a partition of probability-assigned space related to the representative points and the assigned probabilities are first examined. The error in the resulting probability density function of the stochastic responses is then analyzed, leading to two criteria on strategies for determining the representative points and a set of indices in terms of discrepancy of the point sets. A two-step algorithm is proposed, in which an initial uniformly scattered point set is mapped to an optimal set. The implementation of the algorithm is elaborated. Two methods for generating the initial point set are outlined. These are the lattice point sets and the Number-Theoretical nets. A density-related transformation yielding the final point set is then analyzed. Numerical examples are investigated, where the results are compared to those obtained from the standard Monte Carlo simulation and the Latin hyper-cube sampling, demonstrating the accuracy and efficiency of the proposed approach.