Fault tree analysis is a well-known technique in reliability engineering and risk assessment, which supports decision-making processes and the management of complex systems. Traditionally, fault tree (FT) models are b...
详细信息
Fault tree analysis is a well-known technique in reliability engineering and risk assessment, which supports decision-making processes and the management of complex systems. Traditionally, fault tree (FT) models are built manually together with domain experts, considered a time-consuming process prone to human errors. With Industry 4.0, there is an increasing availability of inspection and monitoring data, making techniques that enable knowledge extraction from large data sets relevant. Thus, our goal with this work is to propose a data-driven approach to infer efficient FT structures that achieve a complete representation of the failure mechanisms contained in the failure data set without human intervention. Our algorithm, the FT-MOEA, based on multi-objective evolutionary algorithms, enables the simultaneous optimization of different relevant metrics such as the FT size, the error computed based on the failure data set and the Minimal Cut Sets. Our results show that, for six case studies from the literature, our approach successfully achieved automatic, efficient, and consistent inference of the associated FT models. We also present the results of a parametric analysis that tests our algorithm for different relevant conditions that influence its performance, as well as an overview of the data-driven methods used to automatically infer FT models.
This paper presents a new evolutionary algorithm, called Routes Generation evolutionary Algorithm with Knowledge (RGEAwK), for determining routes with optimal travel time in graph which models the public transport net...
详细信息
This work describes a new methodology for robust identification (RI), meaning the identification of the parameters of a model and the characterization of uncertainties. The alternative proposed handles non-linear mode...
详细信息
This work describes a new methodology for robust identification (RI), meaning the identification of the parameters of a model and the characterization of uncertainties. The alternative proposed handles non-linear models and can take into account the different properties demanded by the model. The indicator that leads the identification process is the identification error (IE), that is, the difference between experimental data and model response. In particular, the methodology obtains the feasible parameter set (FPS, set of parameter values which satisfy a bounded IE) and a nominal model in a non-linear identification problem. To impose different properties on the model, several norms of the IE are used and bounded simultaneously. This improves the model quality, but increases the problem complexity. The methodology proposes that the RI problem is transformed into a multimodal optimization problem with an infinite number of global minima which constitute the FPS. For the optimization task, a special genetic algorithm (epsilon-GA), inspired by Multiobjective evolutionary algorithms, is presented. This algorithm characterizes the FPS by means of a discrete set of models well distributed along the FPS. Finally, an application for a biomedical model which shows the blockage that a given drug produces on the ionic currents of a cardiac cell is presented to illustrate the methodology. (C) 2008 Elsevier Ltd. All rights reserved.
This paper describes a new approach, HIerarchical DEcision Rules (HIDER), for learning generalizable rules in continuous and discrete domains based on evolutionary algorithms. The main contributions of our approach ar...
详细信息
This paper describes a new approach, HIerarchical DEcision Rules (HIDER), for learning generalizable rules in continuous and discrete domains based on evolutionary algorithms. The main contributions of our approach are the integration of both binary and real evolutionary coding;the use of specific operators;the relaxing coefficient to construct more flexible classifiers by indicating how general, with respect to the errors, decision rules must be;the coverage factor in the fitness function, which makes possible a quick expansion of the rule size;and the implicit hierarchy when rules are being obtained. HIDER is accuracy-aware since it can control the maximum allowed error for each decision rule. We have tested our system on real data from the UCI Repository. The results of a 10-fold cross-validation are compared to C4.5's and they show a significant improvement with respect to the number of rules and the error rate. (C) 2003 Elsevier Inc. All rights reserved.
Machine learning support for medical decision making is truly helpful only when it meets two conditions: high prediction accuracy and a good explanation of how the diagnosis was reached. Support vector machines (SVMs)...
详细信息
Machine learning support for medical decision making is truly helpful only when it meets two conditions: high prediction accuracy and a good explanation of how the diagnosis was reached. Support vector machines (SVMs) successfully achieve the first target due to a kernel-based engine;evolutionary algorithms (EAs) can greatly accomplish the second owing to their adaptable nature. In this context, the current paper puts forward a two-step hybridized methodology, where learning is accurately performed by the SVMs and a comprehensible emulation of the resulting decision model is generated by EAs in the form of propositional rules, while referring only those indicators that highly influence the class separation. An individual highlighting of the medical attributes that trigger a specific diagnosis for a current patient record is additionally obtained;this feature thus increases the confidence of the physician in the resulting automated diagnosis. Without loss of generality, we aim to model three breast cancer instances, for reasons of both high incidence of the disease and the large application of state of the art artificial intelligence methods for this medical task. As such, the prediction of a benign/malignant condition as well as the recurrence/nonrecurrence of a cancer event are studied on the Wisconsin corresponding data sets from the UCI Machine Learning Repository. The proposed hybridization reached its goals. Rule prototypes evolve against a SVM consistent training data, while diversity among the different classes is implicitly preserved. Feature selection eventually leads to a resulting rule set where only the significant medical indicators together with the discriminating threshold values are referred, while individual relevance of attributes can be additionally obtained for each patient. The gain is thus dual: the EA benefits from a noise-free SVM preprocessed data and the resulting SVM model is able to output rules in a comprehensible, concise format for the
Metameric problems are variable-length optimization problems whose representations take on an at least partially segmented structure. This is referred to as a metameric representation. Frequently, each of these segmen...
详细信息
Metameric problems are variable-length optimization problems whose representations take on an at least partially segmented structure. This is referred to as a metameric representation. Frequently, each of these segments defines one of a number of analogous components in the solution. Examples include the nodes in a coverage network or turbines in a wind farm. Locating optimal solutions requires, in part, determining the optimal number of components. evolutionary algorithms can be applied but require modifications to the traditional fixed-length operators. This study proposes a new selection operator for metameric problems: length niching selection. First, the population is partitioned into several niches based on solution length. A window function determines at which lengths a niche is formed. Local selection is then applied within each niche independently, resulting in a new parent population formed by a diverse set of solution lengths. A coverage and a wind farm problem are used to demonstrate the effectiveness of the new operator.
The unavailability of excitation measurements poses challenges of application of many structural identification methods due to dealing with two typical types of inverse problems of parameter and force identification s...
详细信息
The unavailability of excitation measurements poses challenges of application of many structural identification methods due to dealing with two typical types of inverse problems of parameter and force identification simultaneously. To address this issue, four different identification methods are proposed based on correlation function to identify structures subjected to multiple unknown ambient excitations, namely gradient search, genetic algorithm, particle swarm optimization (PSO), and effective combination of PSO and gradient search. Numerical studies on a cantilever beam and an eight-story frame, experiments verification on the ASCE benchmark frame are carried out to test the performance of proposed methods. In addition, effect of selection of the reference point, number of data points, unknown initial conditions and modelling errors on accuracy of identification results are also investigated. The numerical and experimental results show that the proposed methods are capable of accurately identifying the unknown structural parameters. In particular, the hybrid method of PSO and gradient search, with approach of producing solutions close to the optimal by PSO and then taking as initial values in gradient search to quickly identify structural unknown parameters, achieves the best performance for overall consideration of identification accuracy and computational efficiency.
Nonlinear bioreactors are considered essential technology in chemical and biochemical industries. This paper presents a proposal of a robust model based fault diagnosis in a nonlinear bioreactor, formulated as the sol...
详细信息
Nonlinear bioreactors are considered essential technology in chemical and biochemical industries. This paper presents a proposal of a robust model based fault diagnosis in a nonlinear bioreactor, formulated as the solution of an inverse problem. The optimization problem is solved by using four different evolutionary strategies: Particle Swarm Optimization (PSO), Differential Evolution (DE), Covariance Matrix Adaptation Evolution Strategy (CMA-ES) and Particle Swarm Optimization with Memory (PSO-M), with DE resulting the best according to the evaluated quantitative indicators. The results obtained with this approach indicate advantages in comparison to other methods of fault diagnosis (FDI) present in literature. (C) 2016 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Hybridizing evolutionary algorithms with local search has become a popular trend in recent years. There is empirical evidence for various combinatorial problems where hybrid evolutionary algorithms perform better than...
详细信息
Hybridizing evolutionary algorithms with local search has become a popular trend in recent years. There is empirical evidence for various combinatorial problems where hybrid evolutionary algorithms perform better than plain evolutionary algorithms. Due to the rapid development of a highly active field of research, theory lags far behind and a solid theoretical foundation of hybrid metaheuristics is sorely needed. We are aiming at a theoretical understanding of why and when hybrid evolutionary algorithms are successful in combinatorial optimization. To this end, we consider a hybrid of a simple evolutionary algorithm, the (1+1) EA, with a powerful local search operator known as variable-depth search (VDS) or Kernighan-Lin. Three combinatorial problems are investigated: Mincut, Knapsack, and Maxsat. More precisely, we focus on simply structured problem instances that contain local optima which are very hard to overcome for many common metaheuristics. The plain (1+1) EA, iterated local search, and simulated annealing need exponential time for optimization, with high probability. In sharp contrast, the hybrid algorithm using VDS finds a global optimum in expected polynomial time. These results demonstrate the usefulness of hybrid evolutionary algorithms with VDS from a rigorous theoretical perspective.
We consider the usage of evolutionary algorithms for multiobjective programming (MOP), i.e. for decision problems with alternatives taken from a real-valued vector space and evaluated according to a vector-valued obje...
详细信息
We consider the usage of evolutionary algorithms for multiobjective programming (MOP), i.e. for decision problems with alternatives taken from a real-valued vector space and evaluated according to a vector-valued objective function. Selection mechanisms, possibilities of temporary fitness deterioration, and problems of unreachable alternatives for such multiobjective evolutionary algorithms (MOEAs) are studied. Theoretical properties of MOEAs such as stochastic convergence with probability 1 are analyzed. (C) 1999 Elsevier Science B.V. All rights reserved.
暂无评论