Measures of Component Importance in Nonrepairable and Repairable Multistate Strongly Coherent Systems

In Natvig and Gåsemyr (Methodol Comput Appl Probab 11:603–620, 2009) dynamic and stationary measures of importance of a component in a binary system were considered. To arrive at explicit results the performance processes of the components were assumed to be independent and the system to be coherent. Especially the Barlow–Proschan and the Natvig measures were treated in detail and a series of new results and approaches were given. For the case of components not undergoing repair it was shown that both measures are sensible. Reasonable measures of component importance for repairable systems represent a challenge. A basic idea here is also to take a so-called dual term into account. For a binary coherent system, according to the extended Barlow–Proschan measure a component is important if there are high probabilities both that its failure is the cause of system failure and that its repair is the cause of system repair. Even with this extension results for the stationary Barlow–Proschan measure are not satisfactory. For a binary coherent system, according to the extended Natvig measure a component is important if both by failing it strongly reduces the expected system uptime and by being repaired it strongly reduces the expected system downtime. With this extension the results for the stationary Natvig measure seem very sensible. In the present paper most of these results are generalized to multistate strongly coherent systems. For such systems little has been published until now on measures of component importance even in the nonrepairable case.     

New Results on the Barlow–Proschan and Natvig Measures of Component Importance in Nonrepairable and Repairable Systems

In this paper dynamic and stationary measures of importance of a component in a binary system are considered. To arrive at explicit results we assume the performance processes of the components to be independent and the system to be coherent. Especially, the Barlow–Proschan and the Natvig measures are treated in detail and a series of new results and approaches are given. For the case of components not undergoing repair it is shown that both measures are sensible. Reasonable measures of component importance for repairable systems represent a challenge. A basic idea here is also to take a so-called dual term into account. According to the extended Barlow–Proschan measure a component is important if there are high probabilities both that its failure is the cause of system failure and that its repair is the cause of system repair. Even with this extension results for the stationary Barlow–Proschan measure are not satisfactory. According to the extended Natvig measure a component is important if both by failing it strongly reduces the expected system uptime and by being repaired it strongly reduces the expected system downtime. With this extension the results for the stationary Natvig measure seem very sensible.     

Multidimensional dependency measures

The problem of dependency between two random variables has been studied throughly in the literature. Many dependency measures have been proposed according to concepts such as concordance, quadrant dependency, etc. More recently, the development of the Theory of Copulas has had a great impact in the study of dependence of random variables specially in the case of continuous random variables. In the case of the multivariate setting, the study of the strong mixing conditions has lead to interesting results that extend some results like the central limit theorem to the case of dependent random variables.In this paper, we study the behavior of a multidimensional extension of two well-known dependency measures, finding their basic properties as well as several examples. The main difference between these measures and others previously proposed is that these ones are based on the definition of independence among n random elements or variables, therefore they provide a nice way to measure dependency.The main purpose of this paper is to present a sample version of one of these measures, find its properties, and based on this sample version to propose a test of independence of multivariate observations. We include several references of applications in Statistics.     

Bias-reduced estimators for bivariate tail modelling

Ledford and Tawn (1997) introduced a flexible bivariate tail model based on the coefficient of tail dependence and on the dependence of the extreme values of the random variables. In this paper, we extend the concept by specifying the slowly varying part of the model as done by Hall (1982) with the univariate case. Based on Beirlant et al. (2009), we propose a bias-reduced estimator for the coefficient of tail dependence and for the estimation of small tail probabilities. We discuss the properties of these estimators via simulations and a real-life example. Furthermore, we discuss some theoretical asymptotic aspects of this approach.     

A load share model for non-identical components of a k-out-of-m system

We consider a k-out-of-m load sharing system when the lifetimes of the components are not necessarily identically distributed random variables. For such systems, a model for the load sharing phenomenon through the exponentiated conditional survival functions of ordered failure times is proposed. This model is more general than the load sharing model with identically distributed component lifetimes and leads to a different family of distributions for ordered random variables. A general expression for the reliability of the system is given. The computations of the reliability for a two component parallel load sharing system corresponding to the exponential and Weibull distributions are discussed. For illustrative purpose, we discuss the inference procedures for a two component parallel load sharing system corresponding to the exponential distributions. A simulation study is carried out to assess the proposed estimation and testing procedures. The applicability of the proposed load sharing model is shown through two data sets.     

基于成本优先级的部件维修策略分析

现实中,系统由于任务、环境等因素,无法实时对故障部件进行维修。因此需要在任务间隔期间或对故障部件进行维修的同时对系统各部件进行预防性机会维修。本文考虑系统期望维修成本,提出了基于部件维修优先级的预防性维修策略。首先把系统期望维修成本分为失效部件维修成本、失效部件导致系统故障的成本和预防性维修其他部件的成本,提出了基于成本的二态和多态系统部件维修优先级度量方法,并在两种场景下分析了如何选择预防性维修部件。其次针对多态系统,研究了基于成本重要度的部件最佳维修水平,并讨论了成本约束下的部件预防性维修策略。最后以某型预警机系统为例进行验证,结果表明,基于成本的预防性维修策略不仅与故障部件位置和相关成本有关,而且还与可用于预防性维修的其他部件重要性有关。     

Measures of the functional dependence of random vectors

In this work, we define a set of properties that any measure of functional dependence that exists between random vectors should possess. We also construct measures of functional dependence and show that they satisfy the properties mentioned above. Relationships between these measures and previously defined measures of functional dependence between random variables are discussed.     

Input-output behaviour of a stochastic Galerkin method for a low pass filter

We apply the stochastic Galerkin method to a linear dynamical system, which includes random variables to quantify uncertainties in physical parameters. The input-output behaviour of the stochastic Galerkin system is described by a transfer function in the frequency domain. The importance of each output component can be estimated by Hardy norms. We investigate a Hardy norm in the case of a linear dynamical system modelling the electric circuit of a low pass filter. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

On the residual dependence index of elliptical distributions

The residual dependence index of bivariate Gaussian distributions is determined by the correlation coefficient. This tail index is of certain statistical importance when extremes and related rare events of bivariate samples with asymptotic independent components are being modeled. In this paper we calculate the partial residual dependence indices of a multivariate elliptical random vector assuming that the associated random radius has distribution function in the Gumbel max-domain of attraction. Furthermore, we discuss the estimation of these indices when the associated random radius possesses a Weibull-tail distribution.     

On rank correlation measures for non-continuous random variables

For continuous random variables, many dependence concepts and measures of association can be expressed in terms of the corresponding copula only and are thus independent of the marginal distributions. This interrelationship generally fails as soon as there are discontinuities in the marginal distribution functions. In this paper, we consider an alternative transformation of an arbitrary random variable to a uniformly distributed one. Using this technique, the class of all possible copulas in the general case is investigated. In particular, we show that one of its members—the standard extension copula introduced by Schweizer and Sklar—captures the dependence structures in an analogous way the unique copula does in the continuous case. Furthermore, we consider measures of concordance between arbitrary random variables and obtain generalizations of Kendall's tau and Spearman's rho that correspond to the sample version of these quantities for empirical distributions.     

L 1-Convergence for Weighted Sums of Some Dependent Random Variables

In this article, the authors discuss the L ₁-convergence for weighted sums of some dependent random variables under the condition of h-integrability with respect to an array of weights. The dependence structure of the random variables includes pairwise lower case negative dependence and conditions on the mixing coefficient, the maximal correlation coefficient, or the ρ*-mixing coefficient. They prove that all the weighted sums have similar limiting behaviour.     

Asymptotic normality in random graphs with given vertex degrees

We consider random graphs with a given degree sequence and show, under weak technical conditions, asymptotic normality of the number of components isomorphic to a given tree, first for the random multigraph given by the configuration model and then, by a conditioning argument, for the simple uniform random graph with the given degree sequence. Such conditioning is standard for convergence in probability, but much less straightforward for convergence in distribution as here. The proof uses the method of moments, and is based on a new estimate of mixed cumulants in a case of weakly dependent variables. The result on small components is applied to give a new proof of a recent result by Barbour and Röllin on asymptotic normality of the size of the giant component in the random multigraph; moreover, we extend this to the random simple graph.     

Ordering Results for Risk Bounds and Cost-efficient Payoffs in Partially Specified Risk Factor Models

Motivated by the problem of sharp risk bounds in partially specified risk factor models and by the method of cost-efficient payoffs with given payoff structure we introduce and describe some stochastic odering problems for conditionally comonotonic resp. antimonotonic random variables. The aim is to describe the influence of the specified dependence of the components of the random vector X with a benchmark Z on the risk bounds in a risk portfolio resp. on the gain of cost efficiency of the optimal payoffs. We obtain in particular explicit results in dependence on distributional parameters for elliptical models in the case of risk bounds and for the multivariate Samuelson model in the case of cost efficient payoffs.     

Optimal replacement policies for a multicomponent reliability system

This paper studies a discrete time, infinite horizon, dynamic programming model for the replacement of components in a binary coherent system. Under quite general conditions, we show that it is optimal to follow a critical component policy (CCP), i.e., a policy specified by a critical component set and the rule: Replace a component if and only if it is failed and in the critical component set. We also discuss the problem of computing such policies.     

A cost-based importance measure for system components: An extension of the Birnbaum importance

In reliability engineering, component importance measures are used to prioritise components in a system for purposes such as reliability improvement and maintenance planning. Existing importance measures have paid little attention to the costs incurred by maintaining a system and its components within a given time period. Cost-effectiveness analysis, however, is critically important in increasingly competitive markets. This paper proposes a new cost-based importance measure which considers costs incurred by maintaining a system and its components within a finite time horizon. Possible extensions are discussed and examples are given to show the use of the new measure.     

Threshold Results for U-Statistics of Dependent Binary Variables

Large sample results for certain U-statistics, and related statistics, of binary dependent random variables are studied. The class of U-statistics include partial sums and polynomials of partial sums of a sequence of random variables. A very wide range of limit results are found. The form of the limit result can depend substantially on the magnitude of the appropriate normalizing sequence for the sum. Unexpectedly, the nature of the limit result also depends significantly on whether the degree of the U-statistic is even or odd. It is shown that dependence is a major factor contributing to this result. The limit results are illustrated with reference to a simple dynamic sequence of binary variables and a reinforced random walk.     

Orderings of coherent systems with randomized dependent components

Consider a general coherent system with independent or dependent components, and assume that the components are randomly chosen from two different stocks, with the components of the first stock having better reliability than the others. Then here we provide sufficient conditions on the component’s lifetimes and on the random numbers of components chosen from the two stocks in order to improve the reliability of the whole system according to different stochastic orders. We also discuss several examples in which such conditions are satisfied and an application to the study of the optimal random allocation of components in series and parallel systems. As a novelty, our study includes the case of coherent systems with dependent components by using basic mathematical tools (and copula theory).     

Reducing model risk via positive and negative dependence assumptions

We give analytical bounds on the Value-at-Risk and on convex risk measures for a portfolio of random variables with fixed marginal distributions under an additional positive dependence structure. We show that assuming positive dependence information in our model leads to reduced dependence uncertainty spreads compared to the case where only marginals information is known. In more detail, we show that in our model the assumption of a positive dependence structure improves the best-possible lower estimate of a risk measure, while leaving unchanged its worst-possible upper risk bounds. In a similar way, we derive for convex risk measures that the assumption of a negative dependence structure leads to improved upper bounds for the risk while it does not help to increase the lower risk bounds in an essential way. As a result we find that additional assumptions on the dependence structure may result in essentially improved risk bounds.     

Constraints on concordance measures in bivariate discrete data

This paper aims to investigate the constraints on dependence measures based on the concept of concordance when discrete random variables are involved. The main technical argument consists in a continuous extension of integer-valued random variables by convolution with unit support kernels.     

A computationally fast variable importance test for random forests for high-dimensional data

Random forests are a commonly used tool for classification and for ranking candidate predictors based on the so-called variable importance measures. These measures attribute scores to the variables reflecting their importance. A drawback of variable importance measures is that there is no natural cutoff that can be used to discriminate between important and non-important variables. Several approaches, for example approaches based on hypothesis testing, were developed for addressing this problem. The existing testing approaches require the repeated computation of random forests. While for low-dimensional settings those approaches might be computationally tractable, for high-dimensional settings typically including thousands of candidate predictors, computing time is enormous. In this article a computationally fast heuristic variable importance test is proposed that is appropriate for high-dimensional data where many variables do not carry any information. The testing approach is based on a modified version of the permutation variable importance, which is inspired by cross-validation procedures. The new approach is tested and compared to the approach of Altmann and colleagues using simulation studies, which are based on real data from high-dimensional binary classification settings. The new approach controls the type I error and has at least comparable power at a substantially smaller computation time in the studies. Thus, it might be used as a computationally fast alternative to existing procedures for high-dimensional data settings where many variables do not carry any information. The new approach is implemented in the R package vita.