Machine learning aided uncertainty quantification for engineering structures involving material-geometric randomness and data imperfection

Computer Methods in Applied Mechanics and Engineering, Год журнала: 2024, Номер 423, С. 116868 - 116868

Опубликована: Фев. 26, 2024

In real-world engineering, uncertainty is ubiquitous within material properties, structural geometry, load conditions, and the like. These uncertainties have substantial impacts on estimation of performance. Furthermore, information or datasets in real life commonly contain imperfections, e.g., noise, outliers, missing data. To quantify these induced by behaviours reduce effects data imperfections simultaneously, a machine learning-aided stochastic analysis framework proposed. A novel supervised learning technique, namely Capped Extended Support Vector Regression (CX-SVR) developed to effectively suppress outliers noise datasets. Its inherent convexity optimization capped strategy theoretically supports accuracy CX-SVR, especially handling with imperfections. Once effective surrogate model established, subsequent analyses, like sampling-based methods, can circumvent cumbersome physical model, which potentially nest computational burden errors engineering applications. The high robustness proposed approach be summarized four main aspects: unrestrictive selection system inputs their statistical information, 'perfect' 'imperfect' data, enough (including moments, probability density functions, cumulative distribution functions) outputs, problems from various fields.

Язык: Английский

---

Design and optimization of a mechanical metamaterial featuring dual tunability in auxeticity and bandgap modulation

Composite Structures, Год журнала: 2025, Номер unknown, С. 119050 - 119050

Опубликована: Март 1, 2025

Язык: Английский

---

Virtual model-aided reliability analysis considering material and geometrical uncertainties for elastic metamaterials

Mechanical Systems and Signal Processing, Год журнала: 2024, Номер 211, С. 111199 - 111199

Опубликована: Фев. 10, 2024

Elastic metamaterials (EMMs) have enormous potential to be employed in real-world engineering, including transport, aerospace, civil due their exceptional wave-manipulating capabilities. The system uncertainties raised by various factors EMMs would lead fluctuations wave attenuation performance. Without thoroughly considering the uncertainties, severe structural failure may occur. Thus, provide possible access quantifying reliability involving material and geometrical separately simultaneously, a virtual model-aided framework is proposed. A recently developed modelling technique, namely Extended Support Vector Regression (X-SVR), has been adopted generate models, as alternatives original physical relationships between parameters concerned bandgap characteristics for EMMs. By implementing sampling-based method on established sufficient statistical information of responses are estimated effectively efficiently. In addition, sensitivity analysis update can easily executed models. Furthermore, computational efficiency, accuracy, robustness proposed demonstrated numerical investigation. Convincingly, this advanced significantly benefit reliability-based analysis, design, fabrication multi-disciplinary engineering applications.

Язык: Английский

---

Investigating Static and Dynamic Behaviors in 3D Chiral Mechanical Metamaterials by Disentangled Generative Models

Advanced Functional Materials, Год журнала: 2024, Номер unknown

Опубликована: Сен. 18, 2024

Abstract Chiral mechanical metamaterials with engineered asymmetric structures have garnered significant interest owing to their potential for manipulating waves and unconventional properties. Although several design methodologies made strides in the discovery of chiral metamaterials, 3D designs that incorporate multiple characteristics remain largely unexplored static dynamic properties these not received sufficient attention. Here, an innovative approach is introduced inverse desired properties, leveraging streamlined shapes enable a versatile design. By employing conditional generative adversarial networks (c‐GANs), are targeted by focusing on both wave attenuation stress distribution under compression shear loading. additive manufacturing fabricated demonstrate performance proposed c‐GAN. Experimental investigations show generated exhibited wide range low concentrations, variety stress‐strain curve profiles including nonlinear stiffness, demonstrating effectiveness achieving characteristics. This research offers advancements on‐demand capabilities applications.

Язык: Английский

---

Full-field experiment-aided virtual modelling framework for inverse-based stochastic prediction of structures with elastoplasticity

Computer Methods in Applied Mechanics and Engineering, Год журнала: 2024, Номер 431, С. 117284 - 117284

Опубликована: Авг. 22, 2024

Язык: Английский

---

Dynamic Cluster Head Optimization for Multipath Routing in Mobile Ad Hoc Networks via Hybrid Stochastic Bandgap Optimization Mixstyle Neural Networks

International Journal of Communication Systems, Год журнала: 2025, Номер 38(7)

Опубликована: Март 14, 2025

ABSTRACT Mobile ad hoc networks (MANETs) are wireless that may be rapidly built and self‐organize. They ideal for military operations, disaster relief, outdoor events, communications in areas without radio infrastructure. In order to find more security flaws, it is advised employ intrusion detection, which controls the system. For further against unauthorized access prevention, monitoring essential. Depending on how long system lasts, a mobile node's capacity forward packets impacted by loss of its power supply. This research proposes use hybrid stochastic bandgap optimization (SBO) mixstyle neural (MNNs) optimize cluster head multipath routing networks. The proposed method combines both SBO MNNs. used choose optimum pathways, MNN select MANET. MATLAB platform build suggested solution, then assessed based several performance metrics, including detection rate, energy consumption, delay, throughput. outperformed other approaches like deep convolutional (DCNNs) bacteria aging algorithm (BFOA) with maximum rate 96% low consumption 0.12 mJ.

Язык: Английский

---

Robust Design Optimization of Chiral Rainbow Metamaterials Under Additive Manufacturing Defects

Опубликована: Янв. 1, 2025

Язык: Английский

---

Uncertainty quantification for locally resonant coated plates and shells

International Journal of Mechanical Sciences, Год журнала: 2024, Номер 282, С. 109587 - 109587

Опубликована: Июль 26, 2024

The effect of uncertainty in design parameters on the acoustic performance coated plates and shells for maritime applications is presented. locally resonant coatings are designed using a viscoelastic material with an impedance similar to water embedded layers inclusions. Both voids hard inclusions, which respectively exhibit monopole dipole resonance scattering, considered. Effective medium approximation theory employed characterise inclusions as homogenised effective geometric properties. coating then modelled multilayered equivalent fluid composed alternating homogeneous bonded rigid backing plate response calculated transfer matrix method. also externally applied elastic cylindrical shell. by expressing shell displacements pressures exterior domain terms Fourier series expansions, applying continuity equations at each interface between surface, surrounding water. Efficient stochastic models based non-intrusive polynomial chaos expansion (PCE) method developed transforming analytical into computationally efficient surrogate point collocation. Uncertainty dominant associated geometry properties examined. influence tuned different frequencies multiple reported. Strong designs occurs broad frequency range around local For all models, predominantly were observed yield greatest variation responses shells.

Язык: Английский

---

Stochastic Modeling of Periodic Beams Under Uncertain Boundary Conditions and Environmental Fluctuations

Опубликована: Янв. 1, 2024

Язык: Английский

---

Mode conversion approach for wave attenuation enhancement of 3D rainbow metamaterials

Engineering Structures, Год журнала: 2024, Номер 321, С. 118999 - 118999

Опубликована: Сен. 23, 2024

Язык: Английский

---