Article
Engineering, Chemical
Mohamed Abohelwa, Bernd Benker, Mehran Javadi, Annett Wollmann, Alfred P. Weber
Summary: This study validates the limitations of the deflector wheel classifier in the separation process through CFD simulations and experimental results. The imbalance in the balance conditions leads to an increase in the cut size and a reduction in the separation sharpness.
Article
Engineering, Chemical
Fabian Krull, Julia Mathy, Paul Breuninger, Sergiy Antonyuk
Summary: The study focuses on the influence of surface roughness on particle-wall collisions. Experimental results show that an increase in surface roughness leads to a decrease in coefficient of restitution in air, but an increase in water.
Article
Geochemistry & Geophysics
Martin Weers, Leonard Hansen, Daniel Schulz, Bernd Benker, Annett Wollmann, Carsten Kykal, Harald Kruggel-Emden, Alfred P. Weber
Summary: The separation characteristics of deflector wheel classifiers are not fully understood and existing models fail to accurately predict them. This study developed a model that critically examined common assumptions, such as ideal airflow, neglecting particle interactions, and particle sphericity. By studying airflow, particle shape, particle-particle interactions, and interactions between particles and the deflector wheel, the separation behavior was successfully estimated.
Article
Engineering, Electrical & Electronic
Dawei Li, Yida Li, Qian Xie, Yuxiang Wu, Zhenghao Yu, Jun Wang
Summary: This article discusses the issue of aero-engine blade surface defect detection in large images and proposes a vision-based framework for defect detection in a coarse-to-fine manner. This method aims to improve the accuracy and efficiency of defect detection effectively.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Review
Engineering, Biomedical
Fei Lin, Hang Sun, Lu Han, Jing Li, Nan Bao, Hong Li, Jing Chen, Shi Zhou, Tao Yu
Summary: An intelligent BI-RADS grading prediction method was proposed in this study, which extracted features and employed a two-layer classifier integration for fine grading prediction, achieving high AUC values on both the testing set and DDSM dataset, significantly outperforming doctors' diagnosis.
INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY
(2022)
Article
Engineering, Chemical
J. B. Starrett, K. P. Galvin
Summary: This study used a REFLUXTM Classifier with water as the fluidising medium to classify feed suspensions based on particle size. It improved on a previous method called Split Fluidisation to achieve precise separations, focusing on controlling the separation size and ensuring efficient separations at high throughputs. The partition curves were sharp, with minimal ultrafine entrainment and oversize particles misplaced. The study also identified the fluidisation velocity as the key control variable for controlling the separation size.
MINERALS ENGINEERING
(2023)
Article
Remote Sensing
Yixuan Song, Fei Song, Lei Jin, Tao Lei, Gang Liu, Ping Jiang, Zhenming Peng
Summary: A new attention cut classification network (ACCN) is proposed to improve ship classification accuracy by combining randomly cut images, attentionally cut images, and raw images. This network achieves a balance between detailed features and global features, and outperforms previous classification models in terms of accuracy.
REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
S. Thirumaladevi, K. Veera Swamy, M. Sailaja
Summary: In this study, a method using transfer learning and fine-tuning strategy is proposed to address the problem of scene classification in high-resolution remote sensing imagery. By replacing the layers in the classifier stage and using a classifier ensemble with a voting strategy, the accuracy of the classification is significantly improved.
Article
Engineering, Chemical
Sonia Acheli, Martin Weers, Annett Wollmann, Alfred P. Weber
Summary: The separation process in a deflector wheel classifier has been extensively researched in the past, but its dynamic behavior has received less attention. It can take several minutes to achieve a steady state. A hold-up is formed between the powder feed and the discharge via fine and coarse material flow. This study investigates the dynamics of a deflector wheel classifier during start-up and material change. The characteristic times are mainly influenced by the feed loading and have less dependence on the rotor speed. For high revolution rates, the characteristic times of both processes are comparable.
CHEMIE INGENIEUR TECHNIK
(2023)
Article
Chemistry, Multidisciplinary
Nabeela Kausar, Abdul Hameed, Mohsin Sattar, Ramiza Ashraf, Ali Shariq Imran, Muhammad Zain ul Abidin, Ammara Ali
Summary: The widespread disease of skin cancer is challenging for dermatologists to diagnose due to the similarity of classes, leading to an accuracy of 62% to 80%. However, utilizing machine learning for classification has shown promise in improving accuracy. Our deep learning-based ensemble models have demonstrated higher accuracy in multiclass skin cancer classification compared to individual deep learners and dermatologists' diagnosis.
APPLIED SCIENCES-BASEL
(2021)
Article
Engineering, Chemical
Huandi Yang, Zhanpeng Sun, Chunyu Liu, Zhiyuan Wang, Yang Yao, Guang Yang
Summary: Particle classifiers with three rotary cages have a significant advantage in powder handling capacity. The flow field inside the classifiers was investigated using the Q criterion to study vortex formation and distribution. The structure of the guide cone was optimized, leading to improved classifying performance. The optimized guide cone enhanced the separation degree of fine and coarse particles, resulting in finer silica powder with a median size of 2.5 μm and higher Newton efficiency of approximately 71.5%.
ADVANCED POWDER TECHNOLOGY
(2023)
Article
Geochemistry & Geophysics
Michael Betz, Hermann Nirschl, Marco Gleiss
Summary: Centrifugal air classifiers are commonly used in various industries for particle classification. This paper presents a new solver based on the MP-PIC method, which allows for detailed investigation of flow processes and the fish-hook effect in the classifier. Experimental data validation shows that numerical simulations with particle-particle interaction consideration provide more accurate results.
Article
Environmental Sciences
Wenjie Lv, Qi Wei, Yujie Ji, Bing Liu, Hongpeng Ma, Yuan Huang, Haitao Huang, Hualin Wang, Pengbo Fu
Summary: Ore resources in the mining process generate a large amount of unmanageable tailings, which can cause serious pollution when released into the environment, but are also valuable resources. The conventional cyclone separation for fine particles has low recovery and utilization rates and requires performance optimization. This study proposes a new volute feed structure to enhance the classification and recovery of fine mineral particles. Through numerical simulation and experimental research, the effects of various structural and operating parameters on flow field distribution, particle motion, and classification performance are systematically examined. The results reveal that the new feed structure can effectively improve flow field stability and particle classification efficiency, leading to an increase in the classification efficiency of fine particles by 10-18%.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2023)
Article
Environmental Sciences
Kun Yang, Haiyan Li, William Perrie, Randall Kenneth Scharien, Jin Wu, Menghao Zhang, Fan Xu
Summary: A new method based on feature selection from Gaofen-3 polarimetric SAR observations was proposed for sea ice classification. The method classified sea ice into four categories: open water, new ice, young ice, and first-year ice. The study re-examined 70 parameters and used the separability index to select optimal features. The classification accuracy of open water, new ice, young ice, and first-year ice reached 95%, 96%, 98%, and 85%, respectively.
Article
Engineering, Chemical
Chen Chen, Yong Zhu, Mingxia Chen, Wenfeng Shangguan
Summary: A novel approach based on the Lagrangian particle tracking method was proposed to investigate particle collision mechanism in a wire-plate electrostatic precipitator. The results showed that electric force controlled particle trajectory direction, while Brownian force only caused particle fluctuation. Binary collision models with determination of dominant force regions were developed, demonstrating the validity of the proposed method in particle collision investigation.
CHEMICAL ENGINEERING RESEARCH & DESIGN
(2021)