Archives of Computational Methods in Engineering最新文献

The optimal treatment selection for cancer patients is extensive, and pharmacogenetic prediction is done using genetic cohort, chemical structure, and target information. Though previous studies sought to characterise pharmacological reactions, there were limits in categorization. Due to the development of various solutions, existing feature selection techniques such as statistical combinations suffer from drawbacks such as local optima, lack of heuristics, and so on. This further leads to a low convergence rate which affects the classification rate. To address this, the current study describes a hybrid approach that is based on machine learning and deep learning, as well as a comparison of the localization heuristic-based Harris Hawk intelligence method and Gravitational Optimization methods with Machine Learning (ML) and Deep- Learning (DL) algorithms. The study suggests the use of Conditional Generative Adversarial Network (CGAN) to obtain better feature selection with less volatility in order to improve data quality and minimise intrinsic variation. In this study, the possible associations between cell lines and drugs are deduced using the CCLE- Cancer-Cell Line Encyclopaedia and Genomics of Drug Sensitivity in Cancer- GDSC datasets, and the study proposes a hybrid Bi-Residual Dense Attention Network for cell line categorization. The proposed method shows better prediction performance based on precision, accuracy, F1-score, Area under curve (AUC), Area under the receiver operating characteristic curve (AUROC), specificity and recall. For the GDSC dataset, the BRDAN-HH framework achieved an accuracy of 0.9675, recall of 0.9795, specificity of 0.975, precision of 0.9785, F1-score of 0.9799, AUC of 0.97, and AUROC of 0.9705. Similarly, for the CCLE dataset, it demonstrated robust performance with an accuracy of 0.9655, recall of 0.986094, specificity of 0.975, precision of 0.975, F1-score of 0.986, AUC of 0.966, and AUROC of 0.9758. The results highlight the efficacy of the BRDAN-HH framework in delivering superior classification metrics, making it a valuable tool for analysing large-scale biomedical datasets.

The influence of tribology has broadened across diverse fields, witnessing substantial and immense growth in different research-related activities over the last decade. This exciting domain drives innovation in lubricant material for extending the lifetime of machinery and contributing to the conservation of energy. Molecular dynamics (MD) simulations play an important role in tribological studies and provide useful insights into atomic-level interactions among sliding surfaces. MD simulations allow researchers to design the model and track the interactions and movements of individual molecules and atoms. This degree of accuracy offers a better understanding of the basic mechanism, including the response of material to different loads and different environmental circumstances. Two-dimensional materials showcase remarkable tribological characteristics. The ultrathin nature and unique atomic arrangement of these materials offer various advantages in the reduction of wear and friction by making them ideal candidates for numerous applications in coatings and lubrication. This review paper explores the MD simulations on tribology in recent years, with a focus on both traditional two-dimensional materials (such as graphene, hexagonal boron nitride, and molybdenum disulfide) and emerging materials (such as MXenes and phosphorene). Our investigation covers the complexity of frictional force at both macroscopic and microscopic scales, the wear mechanism, and the role of adding lubrication for preventing wear and minimizing friction. The main aim is to offer engineers, researchers, and scientists a cherished resource for a better understanding of the complicated ingredients of tribology and direct them to future developments in this critical domain.

Reservoir numerical simulation is crucial for advancing research and development in petroleum engineering. To obtain high-precision spatial and temporal simulation results, a great amount of time and computational resources are needed. Parallel computing addresses this problem by distributing computational workloads and memory requirements across multiple processors. It enables large-scale and high-fidelity simulations and reduces time costs. In this paper, we review existing parallel computing for large-scale reservoir numerical simulation. The paper is achieved by conducting a systematic literature review published between 1990 and 2024. Using the PRISMA guideline, 134 supporting studies are selected for detailed extraction. The key contributions of this paper are threefold: (1) classification and analysis of numerical methods (including discretization methods, nonlinear methods, and linear iterative solvers and preconditioner methods); (2) an in-depth discussion on parallel techniques in high-performance computing (HPC), such as parallel programming models, load balancing, communication optimization, and GPU acceleration; and (3) an outline of software implementations, particularly solvers and reservoir simulators. In conclusion, developing efficient, robust, and scalable linear solving tools is key to reservoir simulation. We compare available preconditioner options and summarise the current state of the art in linear solving tools. Meanwhile, CPU and GPU parallel acceleration techniques have been rapidly developed. These emphases will provide a theoretical foundation and practical guidance for optimizing linear solution processes in the future.

This paper presents a comparative analysis of multiple algorithms for generating sphere clumps as approximations for irregularly-shaped particles in granular systems. Investigated algorithms both include four previously used techniques and two new ones developed in this study. They are often built on common concepts such as distance transforms, filling, packing techniques, and particle medial surface. The two new algorithms herein proposed arrange individual spheres in a clump shape using either a greedy volume coverage or a clustering approach based on a k-means machine learning technique. The performances of the various algorithms are evaluated in terms of both the number of spheres generated per clump and the volume error. The evaluation is conducted on diverse superquadric shapes, serving as ground-truth references, as well as on real rock pieces. Among considered clump generators, results show that existing algorithms may output dispersed results depending on user parameters that are difficult to calibrate, while both proposed algorithms generate realistic sphere clumps, with the volume coverage one being more convenient than the k-means based approach. As a matter of fact, the volume coverage technique is found to be the most effective approach among the studied algorithms in terms of sphere generation and volume precision.

In recent years, finite element analysis has advanced significantly in the clinical study of meniscus diseases. As a numerical simulation technique, finite element analysis provides accurate biomechanical information for diagnosing and treating orthopedic conditions. Compared to traditional methods, finite element analysis is more efficient, convenient, and economical, generating precise data to validate models, guide designs, and optimize clinical protocols. However, there is currently a lack of reviews investigating finite element analysis’s application in meniscal studies. This review addresses this gap by examining current research and practices. It begins by discussing the biomechanical value of finite element analysis in meniscal anatomy and diseases. To thoroughly evaluate the application of finite element analysis in meniscus tear injuries, congenital meniscus abnormalities, and the development of artificial meniscus implants, we explore various research directions from a medical perspective: bionic design, treatment strategy comparison, modeling optimization, prognostic prediction, damage process simulation, damage state analysis, and specific movement investigation. The findings indicate that while finite element analysis shows substantial promise in meniscal research and treatment, challenges remain in establishing standardized experimental protocols and achieving clinical translation. Finally, the paper explored potential directions that may advance the application of finite element analysis in the medical field.

Data’s increasing complexity and uncertainty across domains continue to drive the demand for more robust, efficient, and accurate computational methods, including machine learning algorithms for pattern recognition and classification problems. Kohonen’s Learning Vector Quantization algorithms have been integral to classification algorithms for decades. However, variants such as the Generalized Matrix Learning Vector Quantization have emerged as highly promising and capable computational models for analyzing complex patterns in high-dimensional and noisy datasets with increased performance in recent literature. As a result, this systematic literature review attempts to comprehensively examine recent studies on Generalized Matrix Learning Vector Quantization algorithms, focusing on algorithmic enhancements and variations, inherent features like feature relevance and metric learning, application domains, and performance. Using the Denyer and Tranfield 5-stage systematic literature review method, 61 studies published between 2015 and 2024 are selected for analysis from Scopus, Web of Science, IEEE, and Sprinter. The findings reveal significant advancements and applications of the Generalized Matrix Learning Vector Quantization across healthcare, bioinformatics, and agriculture. The analyzed empirical studies highlight the algorithm’s adaptability to various classification problems and enhanced performance. While the cross-disciplinary potential for Generalized Matrix Learning Vector Quantization is well documented, the review identifies gaps in the literature, particularly in the manufacturing domain. Given the rapid advances in manufacturing and the voluminous amounts of data generated, Generalized Matrix Learning Vector Quantization holds great potential in advancing intelligent decision-making across the domain, such as in the selection and management of manufacturing processes.

Water Distribution Networks (WDNs), as critical urban infrastructures, face heightened vulnerability to damage and failure due to aging systems and external factors such as environmental changes, operational demands, and urban development pressures. Accurate predictive integrity assessment for pipeline systems is crucial for implementing proactive maintenance strategies that prevent catastrophic failures and ensure service reliability. In recent decades, the application of Machine Learning (ML) has emerged as a promising technique for processing and extracting complex interactions between influencing factors and failure trends within WDN systems. This article systematically reviews application scenarios, critical factors influencing WDN integrity, and the modeling and analysis of ML-based predictive models for WDNs. The review analyzes pertinent literature from the past two decades, up to 2024, using the PRISMA procedure and the snowballing method. The findings highlight the superior capabilities of specific ML models, such as tree-based algorithms, artificial neural networks, support vector machines, and other recent deep learning methods in predicting network failures and enhancing system health diagnostics. In addition, key challenges identified include: (i) insufficient standardization in variable selection, model selection and evaluation; (ii) limited data availability due to inconsistent historical failure records; (iii) a lack of systematic feature engineering pipelines for data preprocessing; and (iv) constraints in real-world generalization across finer temporal scales and different geographical regions. Furthermore, the main future research recommendations include developing a standardized framework for variable selection and model architectures, improving multi-source data fusion and collection techniques, enhancing feature engineering methodologies, and conducting systematic evaluations across diverse operational environments.

The model reduction technique (MRT) is an integral part of the finite element model updating (FEMU) approach to address the issue of incompleteness in measurement. It basically condenses the size of a finite element (FE) model to fit with the available responses at limited degrees of freedom. The developments in MRTs and substructure coupling for structural health monitoring (SHM) applications have been enormous. The MRTs are partly discussed in the review articles on FEMU. However, no article is dedicated explicitly to MRTs in SHM applications. Thus, a review article on MRTs will likely augment the state-of-the-art developments of MRTs in FEMU for SHM applications. This review article synthesises the growing literature on different variants of MRTs in time and frequency domains. In doing so, the fundamentals of MRT, salient modifications on the basic MRTs to ease the computational efforts and understanding of its implementation and related developments are presented first. Further, the developments of various substructure coupling techniques used to reduce the order of large FE models are presented. The authors’ recently proposed improved MRTs are also briefly presented. Finally, the prospects and challenges in MRT and substructuring techniques are critically discussed. The review, in general, reveals that the developments in MRTs are gaining importance due to their excellent capability of handling incomplete measurements, indicating the relevance of reviewing the subject from time to time to update the latest developments.

Since its inception in 2009 to being highlighted as the method of the year in 2013, single cell sequencing technology has shown tremendous potential to study various omics profiles or data at an unprecedented resolution. The advances in single cell technology have led to the development of multi-omics techniques which can profile more than one modality from a single cell simultaneously. Thus, providing a significant measure of information which can be utilized to study the cell state and functions eventually the disease and health. The multi-omics profiling has led to a significant increase in production of single cell data. The single cell data is complex due to the heterogeneous nature, thus offers various challenges to deal with such largely complex data. Several computational methods have been proposed to get insights from the single cell multi-omics data. A comprehensive review describing the methods would be great step towards the growth of the field of single cell analysis. Here we provide an in-depth survey of the deep learning computational methods for single cell applications. We provide a brief history of sequencing technologies with a timeline depicting the evolution of various profiling techniques developed over the time. We identify various deep learning techniques that have been employed for single cell applications. This paper presents in-depth survey of deep learning based methods for various downstream applications such as imputation, batch effect (BE) removal, single cell integration and more. We identify various challenges and issues associated with each application which are critical to be addressed. This review will serve as a source of knowledge for new researchers aspiring to begin their research journey in building computational methods to overcome various challenges faced by the field.

This study comprehensively reviews recent advances in the digital reconstruction of polycrystalline materials. Digital reconstruction serves as both a representative volume element for multiscale modelling and a source of quantitative data for microstructure characterisation. Three main types of digital reconstruction in polycrystalline materials exist: (i) experimental reconstruction, which links processing-structure-properties-performance by reconstructing actual polycrystalline microstructures using destructive or non-destructive methods; (ii) physics-based models, which replicate evolutionary processes to establish processing-structure linkages, including cellular automata, Monte Carlo, vertex/front tracking, level set, machine learning, and phase field methods; and (iii) geometry-based models, which create ensembles of statistically equivalent polycrystalline microstructures for structure-properties-performance linkages, using simplistic morphology, Voronoi tessellation, ellipsoid packing, texture synthesis, high-order, reduced-order, and machine learning methods. This work reviews the key features, procedures, advantages, and limitations of these methods, with a particular focus on their application in constructing processing-structure-properties-performance linkages. Finally, it summarises the conclusions, challenges, and future directions for digital reconstruction in polycrystalline materials within the framework of computational materials engineering.