The study proposed an approach to cells classification in a blood smear image, consisting of two stages: object segmentation and application of machine learning methods to determine the cell type. To apply this approa...
The study proposed an approach to cells classification in a blood smear image, consisting of two stages: object segmentation and application of machine learning methods to determine the cell type. To apply this approach, a dataset consisting of cell type labels and the corresponding cell area was created. Object segmentation was performed using image processing methods: noise removal, image binarization, contouring, and area determination using the Gaussian formula. The obtained data set was pre-processed: duplicate data and outliers were removed, and the data was divided into testing and training sets. Training for Gaussian Naive Bayes Classifier, Decision Tree Classifier, Random Forest, Support Vector Classifier models were performed. It was found that the accuracy of all classifiers is 97-98%. In addition, metrics were calculated to evaluate the effectiveness of the models. The Decision Tree Classifier method was chosen to apply the approach to cell classification on new images. All image processing, data collection and processing, model training and evaluation were performed using the Python programming language and relevant libraries.
The purpose of this paper is to show how to identify and handle suspension system vibrations using two different control strategies: Sliding Mode control (SMC) and a Proportional Integral Derivative (PID) controller. ...
详细信息
To directly investigate the dynamic nanoscale phenomenon on the surface being processed in wet conditions such as precision polishing, and cleaning in semiconductor industrial, an optical method for visualization and ...
To directly investigate the dynamic nanoscale phenomenon on the surface being processed in wet conditions such as precision polishing, and cleaning in semiconductor industrial, an optical method for visualization and observation of each sub-100 nm sized particle that is moving on an interface such as a silica glass surface by applying an evanescent field have been proposing. Subsequently, we developed an experimental apparatus equipped with an optical microscopy system for verifying the moving particle observation method in a laboratory scale. This article introduces some experimentally direct observation results of duplicated wet processes.
With the rapid development of computer technology and the Internet of Things (IoT), a large amount of data containing both healthy and faulty states can be collected from existing devices in factories. However, the us...
With the rapid development of computer technology and the Internet of Things (IoT), a large amount of data containing both healthy and faulty states can be collected from existing devices in factories. However, the usable faulty state data is scarce. Transfer learning makes it possible to train efficient and reliable neural networks with a small number of samples. Domain adaptation, as a branch of transfer learning, can effectively realize knowledge transfer between small sample data. Starting from three subcategories of deep learning-based domain adaptation methods, this article summarizes the latest research achievements and provides insights into future research directions.
For RGB semantic segmentation, a two-branch network was proposed to effectively utilize both local detail information and global contextual information within an RGB image. This architecture combines a shallow spatial...
For RGB semantic segmentation, a two-branch network was proposed to effectively utilize both local detail information and global contextual information within an RGB image. This architecture combines a shallow spatial path with a deeper context path, resulting in high performance and FPS. Research on RGB-dept. segmentation has shown the performance gain that the dept. map could provide complementary information to the RGB model. However, the advantage of fusing RGB and dept. map within a two-branch network framework is unclear due to the distinct characteristics of these modalities. To address this, we present a novel fusion RGB-dept. architecture that takes into account the attributes of local context, global context, RGB, and dept. map. Through the bidirectional image dept. fusion technique, we effectively leverage each of the modalities, achieving a performance of 81.23 mIoU. This marks a gain of 1.27% when compared to the RGB-only model and 0.45% when contrasted with the element-wise feature addition fusion baseline.
To a great extent, electric machine (EM) lifetime is affected by electrical failures that compromise the integrity of the windings electrical insulation system (EIS). During EMs operation, the EIS suffered various str...
详细信息
ISBN:
(数字)9798331529277
ISBN:
(纸本)9798331529307
To a great extent, electric machine (EM) lifetime is affected by electrical failures that compromise the integrity of the windings electrical insulation system (EIS). During EMs operation, the EIS suffered various stresses, such as, thermal, electrical, mechanical and ambient (TEAM), which lead the deterioration and decrease the dielectric property of EIS. However, under rated operating conditions, the entire degradation process can take several years before insulation breakdown occurs. Thus, accelerated lifetime tests (ALT) are conventionally performed on various high stress level. Furthermore, the suitable specimen topology should be chosen according to the ALT task objectives, which is also crucial to the resulted lifetime values. In this paper, ALTs at variable temperature profile are performed using two types of specimens (i.e., coils and motorettes). Then, the collected lifetime data are analyzed by statistical tools and the underlying stress mechanism is discussed. Finally, the thermal-mechanical stress is pointed out as the primary factor contributing to lifetime prediction mismatch.
This paper presents investigations of a transformerless half bridge sub-module (HBSM) modular multilevel converter based distribution static synchronous compensator (MMC D-STATCOM) performance in reactive power compen...
This paper presents investigations of a transformerless half bridge sub-module (HBSM) modular multilevel converter based distribution static synchronous compensator (MMC D-STATCOM) performance in reactive power compensation considering the application of the modified ADRC controllers in the inner current loop and in the outer DC voltage loop using Synchronous Reference Frame control. Simulation results with the imposition of severe variations in reactive loads were obtained on a 5 km medium voltage feeder and will demonstrate that the modified ADRC controller enables superior performance for flexible operation with fast and efficient response of the MMC D-STATCOM in reactive power compensation.
Currently, the quadcopter Unmanned Aerial Vehicles (UAVs) are playing a significant role in combating the COVID-19 pandemic crisis, which induced the researchers to design robust control techniques. In this paper, a f...
详细信息
This paper aims at demonstrating how and that model predictive control (MPC) strategies can be used to determine optimal intervention policies against the COVID-19 pandemic. Especially for the time after a first wave ...
详细信息
This paper aims at demonstrating how and that model predictive control (MPC) strategies can be used to determine optimal intervention policies against the COVID-19 pandemic. Especially for the time after a first wave of infection and before a vaccine can be safely distributed to a sufficient extent, the intervention experience from the first outbreak can be utilized to guide the policy decision in this period. The MPC problem in this paper takes the pandemic in different regions of a country and its neighboring countries into account, while policies such as wearing masks or social distancing are selected as inputs to be optimized. This optimized policy balances the risk of a second outbreak and socio-economic costs, while considering that the measure should not be too severe to be rejected by the population. Effectiveness of this policy compared to standard intervention policies is compared through numerical simulations.
暂无评论