Archives of Computational Methods in Engineering最新文献

Pre-processing techniques such as low-light image improvement have a wide variety of practical uses. Enhancing optical acuity and the caliber of photos taken in low-light are the objectives. Techniques for improving low-light images simultaneously boost the brightness, contrast, as well as noise reduction of the image. Self-learning tools, however, have accelerated a lot of this field advancements. Many deep neural networks have been created or put into use as a result. As such, this paper gives a quick summary of the state of the art in low-light image improvement, encompassing techniques related to the controversial open subject. We present a summary of deep learning techniques that are currently carried out to low-light settings. A clear overview of traditional methods for improving low-light primary images. We provide enhanced techniques based on deep learning algorithms and neural structure topologies. Specifically, the current state of deep learning -based low-light picture improvement technologies may be broadly categorized into four sections: visually-based approaches, unobserved learning, unsupervised learning, and observational learning technologies. After then, a database of dimly lit photos is gathered and examined. Furthermore, we present an overview of several quality evaluation standards for enhancing low-light images.

The Cuckoo Search Algorithm (CSA) is an optimization algorithm inspired by the brood parasitism behavior of cuckoo birds. It mimics the reproductive and breeding tactics of cuckoos to tackle optimization challenges. To better handle multi-objective optimization problems (MOPs), a variation called the multi-objective CSA (MOCSA) has been developed. MOCSA is designed to uncover a spectrum of solutions, each providing a balance between various objectives, thereby allowing decision-makers to choose the optimal solution according to their specific preferences. The literature has witnessed a notable increase in the number of published MOCSAs, with MOCSA research papers recorded in the SCOPUS database. This paper presents a comprehensive survey of 123 distinct variants of MOCSAs published in scientific journals. Through this survey, researchers will gain insights into the growth of MOCSA, the theoretical foundations of multi-objective optimization and the MOCSA algorithm, the various existing MOCSA variants documented in the literature, the application domains in which MOCSA has been employed, and a critical analysis of its performance. In sum, this paper provides future research directions for MOCSA. Overall, this survey provides a valuable resource for researchers seeking to explore and understand the advancements, applications, and potential future developments in the field of multi-objective CSA.

In this work, the challenges of the computational evaluation of localized structural failure are discussed and a set of challenging problems for the numerical modeling of quasi-brittle structural failure is presented for the assessment of the performance of models aiming at reproducing the phenomenon. The selected set of challenging problems includes numerical benchmarks and experimental tests reported in the literature, covering several localized structural failure conditions: bending, twisting, merging and branching cracks. The present work focuses on the critical issues when computing localized structural failure faced by present models including: the need to employ a method that produces mesh bias objective results, the requirement to reproduce experimental results in terms of bearing capacity, force–displacement curves, mechanical dissipation, structural size effect, collapse mechanisms with accuracy, the need to perform 3D calculations in a computationally efficient manner to address engineering applications, or the ability to accommodate a broad range of material constitutive behaviors including isotropic and orthotropic crack models with several failure criteria. In the present work, these points are addressed with the use of mixed strain/displacement finite element formulations, which guarantee the local convergence of the computed strains and displacements. This approach is general enough to solve the issues discussed including the spurious mesh bias dependence of computed results in localized structural failure, the aptness to reproduce structural size effect in the computations and the inclusion of orthotropic damage constitutive behavior. To ensure the computational efficiency of the Mixed Finite Element Method, the simulations are performed with adaptive formulation refinement (AFR) and adaptive mesh refinement (AMR) capabilities.

Mammography is the first choice for screening of breast tissue for women aged 38 and above. There are two types of mammographic images, i.e. digitized screen film mammograms and direct digital mammograms. The accurate delineation and segmentation of breast masses from digitized screen film mammograms is considerably challenging task even for experienced radiologists keeping in-view the wide variations in appearances of breast masses buried in different background densities like fatty, fatty glandular and dense tissues. This study presents exhaustive exploration of deep learning based segmentation methods applied to original as well as preprocessed mammographic images from benchmark digital database for screening mammography (DDSM) images. The methods have been characterized as (a) instance segmentation models (b) semantic-segmentation models and (c) hybrid segmentation models. The judicial selection of data augmentation methods used for segmenting breast masses has been highlighted keeping in view the significance of preserving the shape/margin characteristics for diagnosis of breast masses. The shape characteristics being important for differential diagnosis and the significance of preserving the aspect ratio has also been highlighted. Various segmentation performance assessment measures have also been described. The challenges, proposed solutions and future recommendations in the design of DL based segmentation models for DDSM images have also been identified.

The durability, functionality, and performance of machined components are greatly affected by the surface textures created during the machining process. This paper systematically analyses the generation of surface texture in machining operations. The methods of surface modelling in different machining operations are critically reviewed and only models based on machining theories are taken into consideration. In addition, the combined effects of a large number of influential factors are considered and reviewed. The research findings indicate that there is a need to improve the precision of surface modeling analyses and during the evaluation of modeling accuracy, it is crucial to evaluate not just the height parameters but also the functional, hybrid, and spatial parameters. Therefore, it is worthy to mention that this review will help to obtain the suitable surface roughness model and to maximize the performance of the machining system.

This paper explores advancements in multiscale computational models for understanding arterial mechanics and diseases. Arteries, as dynamic structures, must adapt to constant blood flow and pressure, with their layered composition playing a crucial role in maintaining functionality. Recent research highlights the importance of both macroscopic properties and microstructural elements, such as collagen fibers, elastin, smooth muscle cells, and the extracellular matrix. Multiscale modeling bridges these scales, providing insights into how microstructural changes influence arterial behavior under various conditions, including hypertension, atherosclerosis, and aneurysms. This paper emphasizes the utility of these models in simulating arterial conditions, predicting disease progression, and designing medical devices. Key challenges, such as computational complexity, biological integration, and the need for advanced imaging, are addressed alongside suggestions for future directions, including real-time simulations and nanoscale modeling. By combining biological and mechanical perspectives, multiscale approaches offer a comprehensive framework for advancing both scientific understanding and clinical applications in arterial health.

Anti-cancer peptides (ACPs) represent promising candidates for cancer therapy because they can target cancer cells selectively while leaving healthy cells unaffected. ACPs offer a multifaceted approach to cancer treatment by combining targeted cytotoxicity, immune system activation, and the potential to overcome drug resistance. Their development is aided by computational tools that expedite the discovery of promising candidates. As a result, they have received significant attention and broadly studied by many researchers. Currently, numerous peptide-based drugs are undergoing evaluation in preclinical and clinical trials. Accurately identifying ACPs has become a major focus of research, leading to the construction of diverse methods for their detection in silico. These methods implemented different training/testing datasets, classifiers, feature engineering, and feature selection techniques. Thus, it is indispensable to highlight the strengths and weaknesses of current methods and provide insights to improve novel computational tools for identification of ACPs. To address this, we conducted a comprehensive investigation of 26 available existing methods for ACPs, examining their feature engineering methods, classification learning algorithms, performance validation parameters, and availability of web servers. Subsequently, we performed a thorough performance assessment to examine the robustness of these studies using different benchmark datasets. Based on our findings, we offer potential strategies for enhancing model performance and effectiveness.

Lung cancer remains a critical global health challenge, with its prognosis heavily dependent on the timing of diagnosis. This literature review critically examines Artificial Intelligence and Computer-Aided Diagnosis (CADx) systems for lung cancer detection using Computed Tomography (CT) images, guided by seven pivotal research questions. Adhering to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 standards and focusing on high-impact studies from 2013 to 2023, we provide an exhaustive assessment of current methodologies, underscore the variety and efficacy of algorithms and datasets, and evaluate preprocessing and performance evaluation strategies. Our findings reveal significant advancements in integrating machine learning and deep learning techniques, highlighting the importance of machine learning and deep learning methods and scrutinizing their goals, strengths, and limitations. Through a comprehensive meta-analysis, we offer insights into the state-of-the-art in lung cancer CADx, emphasizing data handling, model robustness, and avenues for enhancing diagnostic accuracy and reliability. This review not only critically relates varied methodologies and validates them against established metrics but also offers insights into future research trajectories aimed at enhancing early and accurate lung cancer diagnosis, thereby markedly improving patient outcomes. Targeting broad audiences, from experts in biomedical engineering to those across engineering and clinical sciences, we pave the way for future innovations in this vital domain.

This review paper examines the evolution of shape factors for the bearing capacity of shallow foundations, with a specific focus on rectangular and circular footings. Through a critical examination of methodologies from early empirical approaches to the sophisticated analyses enabled by recent technological advancements, this paper highlights the transformative impact of computational modeling on the field. Specifically, the review utilizes 3D finite element and finite difference analyses to validate and recalibrate shape factors against modern and reliable data. The quantitative findings confirm the reliability of the shape factors developed by Zhu and Michalowski in 2005 through classical finite element analysis in Abaqus. Their ({s}_{gamma }) factor, for example, was validated using Flac3D. Particularly notable is the finding that shape factors for circular footings can be effectively expressed by adjusting those for square footings using a simple geometric ratio, (4/pi ). This adjustment, based on the perimeter or area ratios of the two shapes, suggests a more efficient approach that challenges the necessity for distinct shape factors for different footing types. Additionally, the review highlights historical gaps such as non-documented factors from early empirical research, limitations due to the scale effects of small-scale tests, and assumptions supporting shape factors derived from limit analysis. It also emphasizes that depending on the aspect ratio of the footing and the friction angle of the soil, the percentage error in bearing capacity calculations using non-acceptable shape factors, including those adopted by various design standards, could be several tens of percentage units. Additionally, the review identifies a gap in current research regarding large-scale experimental validation of these computational models, pointing to future directions in experimental research.

Particle Swarm Optimization (PSO) is a key tool in Artificial Intelligence, is well-known to the public for its effectiveness in addressing complex and diverse problems. It possesses strong global search capabilities and robustness, serving as a powerful tool for problem-solving. PSO can handle multiple solutions simultaneously, accelerate problem-solving processes through parallel computing, and dynamically adjust search strategies based on the complexity and variability of problems, thereby adapting to different types of problems. As an efficient swarm intelligence-based algorithm, PSO has been a highly regarded Swarm Intelligence (SI) model since its establishment in 1995, undergoing numerous modifications and innovations to address various complex real-world problems. This article extensively investigates the variants and applications of PSO. Conducted based on a Systematic Review (SR) process, it delves deep into the research papers published in recent years, encompassing different algorithms, a wide range of application domains, potential issues, and future prospects. Specifically, this article reviews existing research methods and their applications, focusing on single-objective algorithms published from 2018 to the present, including but not limited to multiple swarms or multiple samples, learning mechanisms, hybrid algorithms, and their applications in various interdisciplinary fields such as mechanical engineering, civil engineering, power system, energy, and Internet of Things (IoT). Each paper contains practical guidance and inherent limitations, prompting discussions on their applications and outlining potential challenges of PSO, as well as guiding future research directions.