An improved adaptive response surface method for structural reliability analysis

The response surface method(RSM) is one of the main approaches for analyzing reliability problems with implicit performance functions.An improved adaptive RSM based on uniform design(UD) and double weighted regression(DWR) was presented.In the proposed method,the basic principle of the iteratively adaptive response surface method is applied.Uniform design is used to sample the fitting points.And a double weighted regression system considering the distances from the fitting points to the limit state surface and to the estimated design points is set to determine the coefficients of the response surface model.Compared with the conventional approaches,the fitting points selected by UD are more representative,and a better approximation in the key region is also observed with DWR.Numerical examples show that the proposed method has good convergent capability and computational accuracy.

An Intelligent Method for Structural Reliability Analysis Based on Response Surface

As water depth increases, the structural safety and reliability of a system become more and more important and challenging. Therefore, the structural reliability method must be applied in ocean engineering design such as offshore platform design. If the performance function is known in structural reliability analysis, the first-order second-moment method is often used. If the performance function could not be definitely expressed, the response surface method is always used because it has a very clear train of thought and simple programming. However, the traditional response surface method fits the response surface of quadratic polynomials where the problem of accuracy could not be solved, because the true limit state surface can be fitted well only in the area near the checking point. In this paper, an intelligent computing method based on the whole response surface is proposed, which can be used for the situation where the performance function could not be definitely expressed in structural reliability analysis. In this method, a response surface of the fuzzy neural network for the whole area should be constructed first, and then the structural reliability can be calculated by the genetic algorithm. In the proposed method, all the sample points for the training network come from the whole area, so the true limit state surface in the whole area can be fitted. Through calculational examples and comparative analysis, it can be known that the proposed method is much better than the traditional response surface method of quadratic polynomials, because, the amount of calculation of finite element analysis is largely reduced, the accuracy of calculation is improved, and the true limit state surface can be fitted very well in the whole area. So, the method proposed in this paper is suitable for engineering application.

Structural reliability analysis using enhanced cuckoo search algorithm and artificial neural network

The present study proposed an enhanced cuckoo search(ECS) algorithm combined with artificial neural network(ANN) as the surrogate model to solve structural reliability problems. In order to enhance the accuracy and convergence rate of the original cuckoo search(CS) algorithm, the main parameters namely, abandon probability of worst nests paand search step sizeα0 are dynamically adjusted via nonlinear control equations. In addition, a global-best guided equation incorporating the information of global best nest is introduced to the ECS to enhance its exploitation. Then, the proposed ECS is linked to the well-trained ANN model for structural reliability analysis. The computational capability of the proposed algorithm is validated using five typical structural reliability problems and an engineering application. The comparison results show the efficiency and accuracy of the proposed algorithm.

Reliability Sensitivity-based Correlation Coefficient Calculation in Structural Reliability Analysis

The correlation coefficients of random variables of mechanical structures are generally chosen with experience or even ignored,which cannot actually reflect the effects of parameter uncertainties on reliability.To discuss the selection problem of the correlation coefficients from the reliability-based sensitivity point of view,the theory principle of the problem is established based on the results of the reliability sensitivity,and the criterion of correlation among random variables is shown.The values of the correlation coefficients are obtained according to the proposed principle and the reliability sensitivity problem is discussed.Numerical studies have shown the following results：（1） If the sensitivity value of correlation coefficient ρ is less than（at what magnitude 0.000 01）,then the correlation could be ignored,which could simplify the procedure without introducing additional error.（2） However,as the difference between ρs,that is the most sensitive to the reliability,and ρR,that is with the smallest reliability,is less than 0.001,ρs is suggested to model the dependency of random variables.This could ensure the robust quality of system without the loss of safety requirement.（3） In the case of ｜Eabs｜ρ0.001 and also ｜Erel｜ρ0.001,ρR should be employed to quantify the correlation among random variables in order to ensure the accuracy of reliability analysis.Application of the proposed approach could provide a practical routine for mechanical design and manufactory to study the reliability and reliability-based sensitivity of basic design variables in mechanical reliability analysis and design.

SUPPORT VECTOR MACHINE FOR STRUCTURAL RELIABILITY ANALYSIS

Support vector machine （SVM） was introduced to analyze the reliability of the implicit performance function, which is difficult to implement by the classical methods such as the first order reliability method （FORM） and the Monte Carlo simulation （MCS）. As a classification method where the underlying structural risk minimization inference rule is employed, SVM possesses excellent learning capacity with a small amount of information and good capability of generalization over the complete data. Hence, two approaches, i.e., SVM-based FORM and SVM-based MCS, were presented for the structural reliability analysis of the implicit limit state function. Compared to the conventional response surface method （RSM） and the artificial neural network （ANN）, which are widely used to replace the implicit state function for alleviating the computation cost, the more important advantages of SVM are that it can approximate the implicit function with higher precision and better generalization under the small amount of information and avoid the ＂curse of dimensionality＂. The SVM-based reliability approaches can approximate the actual performance function over the complete sampling data with the decreased number of the implicit performance function analysis （usually finite element analysis）, and the computational precision can satisfy the engineering requirement, which are demonstrated by illustrations.

Structural Reliability Analysis of Ship Hulls Accounting for Collision or Grounding Damage

Classical structural reliability analysis of intact ship hulls is extended to the case of ships with collision or grounding damages.Still water load distribution and residual bending moment capacity are included as random variables in the limit state equation.The probability density functions of these random variables are defined based on random damage parameters given by the Marine Environment Protection Committee of the International Maritime Organization,while the proposed reliability formulation is consistent with international recommendations and thus may be valuable in the development of rules for accidental limit states.The methodology is applied on an example of an Aframax oil tanker.The proposed approach captures in a rational way complex interaction of different pertinent variables influencing safety of damaged ship structure.

Structural Reliability Modeling Based on Imprecise Probability Theory under Insufficient Data

Traditional structural reliability analysis methods adopt precise probabilities to quantify uncertainties and they are suitable for systems with sufficient statistical data.However,the problem of insufficient data is often encountered in practical engineering.Thus,structural reliability analysis methods under insufficient data have caught more and more attentions in recent years and a lot of nonprobabilistic reliability analysis methods are put forward to deal with the problem of insufficient data.Non-probabilistic structural reliability analysis methods based on fuzzy set,Dempster-Shafer theory,interval analysis and other theories have got a lot of achievements both in theoretical and practical aspects and they have been successfully applied in structural reliability analysis of largescale complex systems with small samples and few statistical data.In addition to non-probabilistic structural reliability analysis methods,structural reliability analysis based on imprecise probability theory is a new method proposed in recent years.Study on structural reliability analysis using imprecise probability theory is still at the start stage,thus the generalization of imprecise structural reliability model is very important.In this paper,the imprecise probability was developed as an effective way to handle uncertainties,the detailed procedures of imprecise structural reliability analysis was introduced,and several specific imprecise structural reliability models which are most effective for engineering systems were given.At last,an engineering example of a cantilever beam was given to illustrate the effectiveness of the method emphasized here.By comparing with interval structural reliability analysis,the result obtained from imprecise structural reliability model is a little conservative than the one resulted from interval structural reliability analysis for imprecise structural reliability analysis model considers that the probability of each value is taken from an interval.

Structural Reliability Analysis Based on Support Vector Machine and Dual Neural Network Direct Integration Method

Aiming at the reliability analysis of small sample data or implicit structural function,a novel structural reliability analysis model based on support vector machine(SVM)and neural network direct integration method(DNN)is proposed.Firstly,SVM with good small sample learning ability is used to train small sample data,fit structural performance functions and establish regular integration regions.Secondly,DNN is approximated the integral function to achieve multiple integration in the integration region.Finally,structural reliability was obtained by DNN.Numerical examples are investigated to demonstrate the effectiveness of the present method,which provides a feasible way for the structural reliability analysis.

Structural Reliability Analysis Method in Fuzzy Environment

In this paper, the fuzzy-set-based structural possibility theory is investigated, and this theory can be used to deal with the subjective uncertainties in the design of engineering structures. Furthermore, a comprehensive model of structural safety assessment, which can merge subjective uncertainties with objective uncertainties, is presented. In this model, the fuzziness of stress-strength inference model, safety margin functions of single or multiple limit-state, structural failure state and the final assessment result are taken into account. This continuous model can be transformed into an equivalent model of probability-based and solved by the present structural reliability analysis method and parallel algorithm. An example is given to show the main idea of the method presented in this paper.

Response-surface-based structural reliability analysis with random and interval mixed uncertainties

Traditional reliability analysis requires probability distributions of all the uncertain parameters.However,in many practical applications,the variation bounds can be only determined for the parameters with limited information.A complex hybrid reliability problem then will be caused when the random and interval variables coexist in a same structure.In this paper,by introducing the response surface technique,we develop a new hybrid reliability method to efficiently compute the interval of the failure probability of the structure due to the probability-interval hybrid uncertainty.The present method consists of a sequence of iterations.At each step,a response surface model is constructed for the limit-state function by using a quadratic polynomial and a modified axial experimental design method.An approximate hybrid reliability problem is created based on the response surface model,which is subsequently solved by an efficient decoupling approach.An updating strategy is suggested to improve the quality of the response surface and whereby ensure the reliability analysis precision.A computational procedure is then summarized for the whole iterations.Four numerical examples and also a practical application are provided to demonstrate the effectiveness of the present method.

Method of Estimating Human Error Probabilities in Construction for Structural Reliability Analysis Based on Analytic Hierarchy Process and Failure Likelihood Index Method

Human error(HE) is the most important factor influencing on structural safety because its effect often exceeds the random deviation.Large numbers of facts have shown that structural failures may be caused by the gross error due to HE.So it is essential to analyze HE in construction.The crucial work of human error analysis(HEA) is the estimation of human error probability(HEP) in construction.The method for estimating HEP,analytic hierarchy process and failure likelihood index method(AHP-FLIM),is introduced in this paper.The method also uses the process of expert judgment within the failure likelihood index method(FLIM).A numerical example shows the effectiveness of the methods proposed.

A Comprehensive Model for Structural Non-Probabilistic Reliability and the Key Algorithms

It is very difficult to know the exact boundaries of the variable domain for problems with small sample size,and the traditional convex set model is no longer applicable.In view of this,a novel reliability model was proposed on the basis of the fuzzy convex set(FCS)model.This new reliability model can account for different relations between the structural failure region and variable domain.Key computational algorithms were studied in detail.First,the optimization strategy for robust reliability is improved.Second,Monte Carlo algorithms(i.e.,uniform sampling method)for hyper-ellipsoidal convex sets were studied in detail,and errors in previous reports were corrected.Finally,the Gauss-Legendre integral algorithm was used for calculation of the integral reliability index.Three numerical examples are presented here to illustrate the rationality and feasibility of the proposed model and its corresponding algorithms.

A new structural reliability index based on uncertainty theory

The classical probabilistic reliability theory and fuzzy reliability theory cannot directly measure the uncertainty of structural reliability with uncertain variables, i.e., subjective random and fuzzy variables. In order to simultaneously satisfy the duality of randomness and subadditivity of fuzziness in the reliability problem, a new quantification method for the reliability of structures is presented based on uncertainty theory, and an uncertainty-theory-based perspective of classical Cornell reliability index is explored. In this paper, by introducing the uncertainty theory, we adopt the uncertain measure to quantify the reliability of structures for the subjective probability or fuzzy variables, instead of probabilistic and possibilistic measures. We utilize uncertain variables to uniformly represent the subjective random and fuzzy parameters, based on which we derive solutions to analyze the uncertainty reliability of structures with uncertainty distributions. Moreover, we propose the Cornell uncertainty reliability index based on the uncertain expected value and variance.Experimental results on three numerical applications demonstrate the validity of the proposed method.

A robust and efficient structural reliability method combining radial-based importance sampling and Kriging

Simulation based structural reliability analysis suffers from a heavy computational burden, as each sample needs to be evaluated on the performance function, where structural analysis is performed. To alleviate the computational burden, related research focuses mainly on reduction of samples and application of surrogate model, which substitutes the performance function. However,the reduction of samples is achieved commonly at the expense of loss of robustness, and the construction of surrogate model is computationally expensive. In view of this, this paper presents a robust and efficient method in the same direction. The present method uses radial-based importance sampling (RBIS) to reduce samples without loss of robustness. Importantly, Kriging is fully used to efficiently implement RBIS. It not only serves as a surrogate to classify samples as we all know, but also guides the procedure to determine the optimal radius, with which RBIS would reduce samples to the highest degree. When used as a surrogate, Kriging is established through active learning, where the previously evaluated points to determine the optimal radius are reused. The robustness and efficiency of the present method are validated by five representative examples, where the present method is compared mainly with two fundamental reliability methods based on active learning Kriging.

IMPROVED STRATIFIED SAMPLING MONTE CARLO METHOD TO ANALYZE RELIABILITY OF STRUCTURAL SYSTEMS

A stratified sampling Monte Carlo method to analyze the reliability of structural systems is presented. Introducing a small exploratory simulation, this method overcomes the difficulties for getting the systematic sampling probability of all the strata. Several useful and efficient stratification methods are given and the strategies of stratification and simulation are studied. A general conclusion has been presented corresponding to actual engineering structures. The strict theoretical proof has been given,and it is especially effective to solve probabilistic integration. Statistic error of evaluating failure probability is reduced obviously. Especially in highly non-linear and nonreonvex problems, it is more accurate than other methods. Compared with other variance reduction techniques, this method can obtain a more obvious variance reduction and an increased sampling efficiency. Moreover, without strict limiting condition, it is convenient to use. This method is especially suitable to solve the reliability problem of structural systems with multiple failure modes and highly non-linear safety margin equations.

The Exact Limits of Structural Reliability for Normal Stress and Normal Strength

Under the condition of normal strength and stress with unknown distribution parameters but getting a completed sample respectively, a comparison among the errors of some kinds of approximate limits for structural reliability has been made in this paper, basing on the exact limits presented. All results in this paper can be used with condition logical normal distribution conveniently.

Determination of Partial Safety Coefficients of Structural Reliability Based on Contaminated Distribution

Partial safety factors must be evaluated precisely for the given target reliability index to ensure the certain level of structural reliability due to uncertain factors.The current studies of partial safety factors do not consider human error in construction for structural reliability.A mathematically model should be improved to simulate the partial safety coefficient concerned uncertainty factors which concern the effect of human error in construction.We employ the contaminated distribution to obtain the realistic mean value and standard variance of variable of structural parameters which coexist with random error human error.The reasonable partial safety coefficient can be calculated based on the realistic value of structural parameters concerned the effects of random error and gross error.

Structural Interval Reliability Algorithm Based on Bernstein Polynomials and Evidence Theory

Structural reliability is an important method to measure the safety performance of structures under the influence of uncertain factors.Traditional structural reliability analysis methods often convert the limit state function to the polynomial form to measure whether the structure is invalid.The uncertain parameters mainly exist in the form of intervals.This method requires a lot of calculation and is often difficult to achieve efficiently.In order to solve this problem,this paper proposes an interval variable multivariate polynomial algorithm based on Bernstein polynomials and evidence theory to solve the structural reliability problem with cognitive uncertainty.Based on the non-probabilistic reliability index method,the extreme value of the limit state function is obtained using the properties of Bernstein polynomials,thus avoiding the need for a lot of sampling to solve the reliability analysis problem.The method is applied to numerical examples and engineering applications such as experiments,and the results show that the method has higher computational efficiency and accuracy than the traditional linear approximation method,especially for some reliability problems with higher nonlinearity.Moreover,this method can effectively improve the reliability of results and reduce the cost of calculation in practical engineering problems.

A new structural reliability analysis method in presence of mixed uncertainty variables

For structures with both random and fuzzy uncertainty,this paper presents a novel method for determining the membership function in fuzzy reliability with the Automatic Updating Extreme Response Surface(AUERS)method.In the proposed method,fuzzy variables are initially converted into a value domain under the given cut level and the extreme point in the domain where the reliability reaches its extreme value is considered.Second,the Particle Swarm Optimization(PSO)algorithm is used to determine the extreme point according to the extreme responses for different sets of random sample inputs.A kriging response surface is subsequently constructed between the random variables and the corresponding extreme points.An automatic updating strategy is then introduced based on the Relative Mean Square Predicted Error(RMSPE)before performing every iteration of reliability analysis.By adding new sample points,the approximate quality of the kriging response surface is improved.Finally,reliability analysis is used to determine the reliability bound under the given cut level.The proposed method assures the accuracy and computation efficiency of the mixed uncertainty reliability analysis results while it prevents the solution from becoming trapped in a local optimum,which occurs in classical optimization methods.Two example analyses are used to demonstrate the validity and advantages of the proposed method.

Genetic Algorithms of Structural Fuzzy Reliability Index

In this paper the simple generation algorithms are improved. According to the geometric meaning of the structural reliability index, a method is proposed to deal with the variables in the standard normal space. With consideration of variable distribution, the correlation coefficient of the variables and its fuzzy reliability index, the feasibility and the reliability of the algorithms are proved with an example of structural reliability analysis and optimization.