Archives of Computational Methods in Engineering最新文献

This comprehensive survey addresses the notable yet relatively uncharted territory of machine learning (ML) applications within the realm of earthquake engineering. While previous reviews have touched on ML’s involvement, this work strives to fill a gap by providing an extensive analysis of the extent to which ML has permeated earthquake engineering. It delves into how ML is facilitating and propelling research endeavors while aiding decision-makers in mitigating the repercussions of seismic hazards on civil structures. Earthquake engineering, an interdisciplinary field, encompasses the assessment of seismic hazards, characterization of site-specific effects, analysis of structural responses, evaluation of seismic risk and vulnerability, and examination of seismic protection measures. ML algorithms find application in a multitude of scenarios within each of these subfields, contributing to advancements in earthquake engineering research and practice.

In image processing, multi-level thresholding is a sophisticated technique used to delineate regions of interest in images by identifying intensity levels that differentiate different structures or objects. Multi-range intensity partitioning captures the complexity and variability of an image. The aim of metaheuristic algorithms is to find threshold values that maximize intra-class differences and minimize inter-class differences. Various approaches and algorithms are reviewed and their advantages, limitations, and challenges are discussed in this paper. In addition, the review identifies future research areas such as handling complex images and inhomogeneous data, determining thresholding levels automatically, and addressing algorithm interpretation. The comprehensive review provides insights for future advancements in multilevel thresholding techniques that can be used by researchers in the field of image processing.

Regression is one of the most important supervised learning methods in data mining that is used to predict and discover knowledge in data mining science. After reviewing the studies conducted in the field of regression, it has been found that the tendency to use this method is increasing day by day among researchers. This study reviews 500 articles from about 230 reputable journals under one framework over the twenty-first century and also discusses the status and use of regression in data mining research. The systematic framework presented in this study includes the following steps: 1—Examining the position of regression in research conducted in data mining and determining the trend of different journals to conduct research in the field of regression in different years 2—Examining different study areas in the field of regression and determining the trend to conduct research in various areas of study in different years 3—Examining the algorithms used in the field of regression and determining the most widely used and trend to use algorithms by researchers in different years 4—Examining the keywords used in regression research in data mining and determining the strongest and most attractive rules obtained from the relationships of these keywords with each other using the Apriori algorithm.

Mpox, a zoonotic viral disease, poses a significant threat to human health, characterized by its potential for human-to-human transmission and its manifestation in severe flu-like symptoms and distinctive skin lesions. This paper offers a comprehensive exploration of Mpox detection and classification, beginning with an introduction to the subject and a description of the research objectives and scope. A thorough examination of the historical context and epidemiology of Mpox sets the stage for a detailed discussion of the fundamental background concepts, encompassing medical imaging, various types of medical imaging techniques, machine learning (ML) applications, convolutional neural networks (CNNs), and available architectural families. The study highlights essential model evaluation metrics to provide a robust framework for assessing the efficacy of different approaches. Methodologically, the paper outlines the systematic approach employed in the literature review and study selection process. With an emphasis on benchmark datasets, the research delves into the diverse AI-based methodologies, encompassing both ML and deep learning (DL) approaches, utilized in Mpox detection. The paper meticulously describes the challenges inherent in these methodologies and concludes with a thoughtful exploration of future prospects in the field. The main purpose is to provide a comprehensive overview of the current landscape and pave the way for advancements that can significantly impact the diagnosis and management of Mpox outbreaks.

Nowadays, nature-inspired artificial intelligent metaheuristic optimization algorithms (MHOAs) have gained many attentions from researchers all over the world due to their capabilities in solving various decision-making problems. These algorithms are inspired and modelled based on the searching behaviour of animals in real life. This review paper provides in-depth discussions on various challenges and breakthroughs in numerous state-of-the-art nature-inspired artificial intelligence (AI) algorithms in solving multi-objective optimization engineering problems with emphasis on the mathematical modelling and algorithm developments. From conventional analysis such as speeds and accuracies to relatively advanced benchmarks such as complexities and convergence patterns, the comparison criteria of population-based and nature-inspired search mechanisms have evolved in the effort to further enhance the overall performance and reachability of these heuristic algorithms. This paper provides a platform for young readers and new researches who are about to indulge in the realm of various AI optimization techniques. Comprehensive analysis and discussions are presented on various state-of-the-art methods, with possible fields of applications proposed. Suitability of search mechanisms to specific optimization problem categories has also been investigated and presented, with combined or hybrid methods under scrutiny.

This article systematically identifies and comparatively analyzes state-of-the-art supply chain (SC) forecasting strategies and technologies within a specific timeframe, encompassing a comprehensive review of 152 papers spanning from 1969 to 2023. A novel framework has been proposed incorporating Big Data Analytics in SC Management (problem identification, data sources, exploratory data analysis, machine-learning model training, hyperparameter tuning, performance evaluation, and optimization), forecasting effects on human workforce, inventory, and overall SC. Initially, the need to collect data according to SC strategy and how to collect them has been discussed. The article discusses the need for different types of forecasting according to the period or SC objective. The SC KPIs and the error-measurement systems have been recommended to optimize the top-performing model. The adverse effects of phantom inventory on forecasting and the dependence of managerial decisions on the SC KPIs for determining model performance parameters and improving operations management, transparency, and planning efficiency have been illustrated. The cyclic connection within the framework introduces preprocessing optimization based on the post-process KPIs, optimizing the overall control process (inventory management, workforce determination, cost, production and capacity planning). The contribution of this research lies in the standard SC process framework proposal, recommended forecasting data analysis, forecasting effects on SC performance, machine learning algorithms optimization followed, and in shedding light on future research.

Data mining is the most widely used method for discovering knowledge. There are numerous data mining tasks, with classification being the most frequently encountered task in various application domains such as fraud detection, disease diagnosis, text classification, and so on. Many classification techniques, such as Bayesian classifiers, decision trees, genetic algorithms, neural networks (NNs), and so on, are available to help researchers solve problems in a variety of domains. However, NNs are the most frequently used classification approach because they are effective at solving classification problems that cannot be divided into linear and non-linear categories, have high classification accuracy on large datasets, and require minimal processing effort. Despite having good classification performances, NNs have a pitfall associated with them which hinders their applicability in some real-world applications. NNs are black boxes in nature, which means they cannot make transparent decisions that humans can interpret. Because of this limitation, NNs are unsuitable for many applications that require transparency in decision-making as well as high accuracy, such as audit mining or medical diagnosis. The well-known solution to this inherent disadvantage of NNs is to extract explainable decision rules from them. The extracted rules provide a detailed understanding of how NNs work in a human-readable format. Rule extraction is a well-established technique with a plethora of literature on the subject. However, there are very few papers whose primary goal is to survey the existing literature. As a result, the goal of this work is to provide a detailed analysis of the existing literature and to create a framework for existing and new researchers to conduct research in this field. The paper examines the state-of art from the perspective of designing framework of the algorithms, evaluation criteria, and applications.

Finite element analysis (FEA) is an extensively exercised numerical procedure to address numerous problems in several engineering fields. However, the accuracy of conventional FEA solutions is significantly affected in specific circumstances where the problem demands near-incompressibility or incompressibility of domain or analysis of thin structural geometries. Over time, several advanced FE models are developed to improve the quality of solutions in stated situations. However, the extensive comparative aspects of these methods are spared limited attention. In the present paper, a comprehensive review and comparison of the selected FE models have been presented. The detailed implementation procedure, along with the relative efficacy of the methods, has been derived for selective reduced integration (SRI), enhanced assumed strain (EAS), assumed natural strain (ANS), and a specific class of hybrid stress elements alongside the conventional FE formulation. The quality of results is assessed by evaluating the relative error norms in displacement and stress on well-established benchmark numerical examples. Furthermore, the paper investigates the methods for several parameters that include the method’s best-suited environment, robustness, and efficiency. The findings in the paper provide an elaborate understanding of the optimal choice of the method in locking-dominated problems.

This review article provides an overview of recent research on deep learning (DL) methods for identifying and classifying lung nodules in medical images, with a focus on X-ray and CT scans. It encompasses a thorough analysis of studies published in reputed/peer-reviewed journals and international conferences. The review explores various aspects, including the development and implementation of DL models, the use of data augmentation techniques to enhance model performance and the application of transfer learning to adapt existing models to new datasets. The findings highlight the effectiveness of DL techniques in improving accuracy and efficiency in lung nodule detection and classification. Furthermore, these methodologies can be employed to cultivate automated systems that have the potential to aid radiologists in the processes of diagnosis and treatment planning. This review underscores the importance of continued research and development into the present state of DL research about detecting and classifying lung nodules.

Weather is influenced by various factors such as temperature, pressure, air movement, moisture/water vapor, and the Earth’s rotating motion. Accurate weather forecasting at a high geographical resolution is a complex and computationally expensive task. This study employs a nowcasting approach using meteorological radar images. Building upon the principles of unsupervised representation in deep learning, we delve into the emerging field of next-frame prediction in computer vision. This research focuses on predicting future images based on prior image data, with applications ranging from robot decision-making to autonomous driving. We present the latest advancements in next-frame prediction networks, categorizing them into two approaches: Machine Learners and deep learners. We discuss the merits and limitations of each approach, comparing them based on various parameters. Finally, we outline potential directions for future research in this field, aiming to make weather forecasting more precise and accessible.