Archives of Computational Methods in Engineering最新文献

With the rise of the Internet of Things (IoT), applications have become more competent and smart, and connected devices have given rise to the exploitation of all aspects of a modern city. In today’s era, the problem of parking is also increasing due to the increase in the number of vehicles. Motorists waste time and fuel searching for parking, which may be far from their intended destination. Historically, parking in a congested urban environment has been challenging, frequently depending on manual techniques. Several parking facilities have implemented computerized systems and monitoring technology such as CCTV cameras for tracking car movements. However, these existing systems remain primarily inefficient. This growing challenge emphasizes the pressing demand for enhanced vision and IoT-based solutions to manage parking in urban environments, minimizing time and energy expenditure while improving overall convenience. In the past decade, several research efforts have been conducted to create an intelligent system for detecting and classifying parking spaces, turning into an attractive research domain. To build such a system, researchers have employed various machine learning (ML), deep learning (DL), and IoT. These techniques have been explored to enhance the effectiveness and utility of smart parking. This review paper provides an extensive, comparative, and systematic examination of parking space detection and classification methods. The study provides a detailed discussion of the publicly available datasets used for the performance evaluation of existing ML, DL, and vision techniques integrated with IoT. The review identifies the gaps in existing parking space detection and classification techniques, which further require investigation to improve the effectiveness and capability of smart parking.

Radiology also termed as the medical imaging is the medical specialty that involves the creation of images of the body parts for the purpose of diagnostics or treatment. The procedures involved therefore helps the medical professionals in diagnosing the diseases and injuries. The medical image analysis of the brain is considered as the major area of interest because of its complexity and significance and the automation of the same can be done using various tools and techniques. There are variety of image processing techniques used for the brain image analysis, to name a few are the Deep Learning, Machine Learning, hybrid models etc. There are variety of reasons such as the shape, dimension, textures and other related features due to which the analysis of the brain tumors tends to become complicated. Henceforth, this review will give a comprehensive review of the brain tumor image analysis, with the inclusion of the topics such as the fundamentals of brain tumors, brain imaging, actions involved in brain image analysis, models utilized, characteristics of brain tumor images, metrics for model evaluation and datasets of brain tumor and medical images that are available.

The symptoms of ovarian cancer are nonspecific, and current screening methods lack sufficient accuracy for early diagnosis. This often leads to detection at a later, more advanced stage of the disease. Medical imaging provides morphological and functional data to help characterize ovarian tumors, but more research is needed to develop reliable early screening tools. This review examines recent machine learning techniques applied to imaging data for improving ovarian cancer detection and diagnosis. A literature search was conducted on PubMed, IEEE, and ACM databases for studies from 2010 to 2023 utilizing machine learning with ultrasound, magnetic resonance imaging, computed tomography, or other imaging data and clinical records to detect ovarian cancer. Key information extracted included imaging modality and clinical recordings, machine learning methods, classification tasks, performance metrics, and datasets. This work identified 81 relevant studies. Artificial intelligence approaches included traditional methods like support vector machines, random forest and logistic regression, and deep learning models like convolutional neural networks, vision transformers, and graph neural networks. Most studies focused on the binary classification of benign vs. malignant adnexal masses. The range of reported diagnostic accuracy across different modalities is 75–99%. Deep learning generally outperformed traditional machine learning models. Consequently, machine learning, especially deep learning, shows promising performance in detecting ovarian cancer from medical images. However, the heterogeneity of imaging protocols, data labeling biases, model interpretability, and validation on multi-center datasets is challenging. Future work should focus on robust and generalizable solutions that can be deployed as clinical tools for improving ovarian cancer outcomes.

The coaxial compound helicopter has enhanced the maximum flight speed, while retaining the hover and low-speed maneuvering capabilities of conventional helicopters. As a novel configuration, its coaxial rigid rotors and tail-mounted propeller introduce unique aerodynamic and flight dynamic characteristics. This paper reviews both the aerodynamic and flight dynamic characteristics of coaxial compound helicopters. This review begins with an introduction to the unique features of coaxial compound helicopters, such as lift offset and control redundancy. In the aerodynamic section, various numerical simulation methods used in aerodynamic research are summarized, containing Computational Fluid Dynamics trimming methods. The aerodynamic characteristics of coaxial compound helicopters are reviewed in terms of aerodynamic interference, aerodynamic loads, and aerodynamic noise. Aerodynamic interference between the rotors is a major cause of unsteady aerodynamic loads, which in turn lead to aerodynamic noise. Subsequently, this paper introduces the models involved in flight dynamics modeling, detailing the rotor aerodynamic model, inflow model, blade motion model, and aerodynamic interference model. Based on this, the trim, stability, controllability, and flight control design of coaxial compound helicopters are reviewed. The differences in trim results between coaxial compound helicopters and conventional helicopters, as well as the control coupling effects of coaxial compound helicopters, are addressed. Finally, this paper summarizes the aerodynamic and flight dynamic characteristics of coaxial compound helicopters and provides some suggestions for future research.

Composites materials reinforced with natural fibers are currently gaining traction in many industries including automotive, aerospace, marine, packaging and construction due to their ecological consciousness and high strength to weight ratio. To enhance the overall performance and use of natural fibers composites (NFC) in different industries, it is crucial to understand their acoustic properties, moisture absorption, mechanical characteristics, manufacturing processes, tribological behavior and damage mechanics. Analyzing the performance of NFC is a complex process due to the heterogeneity and anisotropic nature of NFC coupled with their susceptibility to environmental factors that lead to a significant variability in their composites. Research on NFC performance typically depends on the time consuming and costly experiments with limited reproducibility and computationally intensive simulations. Machine learning (ML) techniques can efficiently uncover data patterns and offer high reproducibility. Additionally, advancements in NFC manufacturing and testing have produced vast amounts of data. The current review not only discusses the application of ML methods in enhancing NFC performance, but also identifies the challenges and opportunities associated with using ML in NFC research. By utilizing ML methods, NFC limitations can be overcome, leading to improved performance.

Autoencoders have become a fundamental technique in deep learning (DL), significantly enhancing representation learning across various domains, including image processing, anomaly detection, and generative modelling. This paper provides a comprehensive review of autoencoder architectures, from their inception and fundamental concepts to advanced implementations such as adversarial autoencoders, convolutional autoencoders, and variational autoencoders, examining their operational mechanisms, mathematical foundations, typical applications, and their role in generative modelling. The study contributes to the field by synthesizing existing knowledge, discussing recent advancements, new perspectives, and the practical implications of autoencoders in tackling modern machine learning (ML) challenges.

Heart disease (HD) is a leading cause of death globally, posing a significant healthcare burden. Early and correct diagnosis is crucial for effective management and improved patient outcomes. Machine learning (ML) has emerged as a promising tool for developing decision support systems to aid HD detection. This systematic review examined the current landscape of ML-based HD diagnostic systems, focusing on the utilized techniques, performance metrics, validation approaches, and publicly available datasets. The authors identified key research gaps, including data heterogeneity, class imbalance, lack of real-world validation, and limited integration of multi-modal data. Additionally, the authors discussed challenges related to model interpretability, ethical considerations, and the need for personalized medicine approaches. Finally, the authors explored promising future directions, such as the use of quantum machine learning and dynamic prediction systems for continuous monitoring. This comprehensive review presented valuable insights for researchers and healthcare professionals aiming to leverage the power of ML for improved HD diagnosis and patient care.

Early detection of infectious lung diseases is vital, and various researchers have created models to help with this. Different experts may have different opinions about how to classify a particular image in the dataset. The expertise, level of experience, or personal preferences of the experts might be the source of these differences. Automatic disease classification can help radiologists by reducing workload and improving patient care. Recent advancements in deep learning have helped the diagnosis and classification of lung diseases in medical imaging. As a result, there are several research in the literature utilising deep learning to identify lung diseases. A comprehensive review of the most recent DL and ML methods for lung disease diagnosis is given in this work. The selected studies are published from 2019 until 2024. Overall, total seventy-seven carefully chosen papers from various publications, including Nature, IEEE, Springer, Elsevier, and Wiley, are included in this study. Deep learning techniques for the detection of infectious lung diseases from medical images are presented in this paper. In addition to providing a taxonomy of the most advanced deep learning and machine learning-based lung disease detection systems, this comprehensive review also seeks to identify existing challenges, present the trends in the field’s current research, and provide projections about potential future directions.

Supply chains with diverse and conflicting objectives striving for optimal performance often land in NP (Nondeterministic Polynomial)-hard combinatorial optimization problems employing tools from classical and non-classical approaches. This paper aims to collect these studies on quantitative optimization models applied to supply chains and conduct a comprehensive review of the literature published during 2006–2023. A total of 283 research articles were collected from several relevant databases to present the taxonomy, trend and insights gained from the analysis of the data. The taxonomies presented are based on extended classification schemes such as modelling approach, objective functions, data sources, optimization tools and their hybridization, etc. Five research questions (RQs) are formed based on the required taxonomy to properly guide the review work. Statistical analysis has been carried out to comprehend any transitions observed during the review period. The review is concluded with key observations on the status of research, and future directions.

The development of new metaheuristic algorithms and their enhancements has seen significant growth, yet many of these algorithms share similar limitations. This is largely due to insufficient studies analyzing their structures and performance prior to proposing modifications. The Tunicate Swarm Algorithm (TSA), a recently developed nature-inspired algorithm, offers a simple structure, distinctive stabilizing features, and impressive efficiency. Inspired by the social behaviors of tunicates and their jet propulsion for movement and foraging, the TSA employs a dynamic weighting mechanism to simulate their influence during the search process. Its notable traits, including simplicity, adaptability, minimal parameters, and independence from derivatives, have contributed to its rapid adoption across various optimization problems. This review focuses on the foundational research underlying the TSA, exploring its development and effectiveness as highlighted in existing studies. It also examines enhancements to the algorithm’s behavior, particularly efforts to align search space geometry with practical optimization challenges. Finally, potential directions for future improvements and adaptations are proposed to further advance the TSA’s capabilities.