Applied sciences

Bulletin of the Polish Academy of Sciences Technical Sciences

Content

Bulletin of the Polish Academy of Sciences Technical Sciences | Early Access

Download PDF Download RIS Download Bibtex

Abstract

The interpretation of breast magnetic resonance imaging (MRI) in the healthcare field depends on the good knowledge and experience of radiologists. Recent developments in artificial intelligence (AI) have shown advances in the field of radiology. However, the desired levels have not been reached in the field of radiology yet. In this study, a novel model structure is proposed to characterize the diagnostic performance of AI technology for individual breast dynamic contrast material–enhanced (DCE) MRI sequences. In the proposed model structure, Inception-v3, EfficientNet-B3 and DenseNet-201 models were used as hybrids together with the Yolo-v3 algorithm to detect breast and cancer regions. In the proposed model, DCE-MRI sequences (T2, ADC, Diffusion, Non-Contrast Fat Non-Suppressed T1, Non-Contrast Fat Suppressed T1, Contrast Fat Suppressed T1, and Subtraction T1) were evaluated separately and validation was made, thus providing a unique perspective. According to the validation results, the model structure with the best performance was determined as Yolo-v3 + DenseNet-201. With this model structure, 92.41 accuracy, 0.5936 loss, 92.44% sensitivity, and 92.44% specificity rates were obtained. In addition, it was determined that the results obtained without using contrast material in the best model were 91.53% accuracy, 0.9646 loss, 92.19% sensitivity, and 92.19% specificity. Therefore, it is predicted that the need for contrast material use can bereduced with the help of this model structure.
Go to article

Authors and Affiliations

Volkan KAYA
İsmail AKGÜL
Erdal KARAVAŞ
Sonay AYDIN
Ahmet BARAN
Download PDF Download RIS Download Bibtex

Abstract

Optimization of industrial processes such as manufacturing or processing of specific materials is a point of interest for many researchers, and its application can lead not only to speeding up the processes in question, but also to reducing the energy cost incurred during them. This article presents a novel approach to optimizing the spindle motion of a computer numeric control (CNC) machine. The proposed solution is to use deep learning with reinforcement to map the performance of the Reference Points Realization Optimization (RPRO) algorithm used in industry. A detailed study was conducted to see how well the proposed method performs the targeted task. In addition, the influence of a number of different factors and hyperparameters of the learning process on the performance of the trained agent was investigated. The proposed solution achieved very good results, not only satisfactorily replicating the performance of the benchmark algorithm, but also, speeding up the machining process and providing significantly higher accuracy.
Go to article

Authors and Affiliations

Dawid Kalandyk
Bogdan Kwiatkowski
ORCID: ORCID
Damian Mazur
ORCID: ORCID
Download PDF Download RIS Download Bibtex

Abstract

The performance of long-wave infrared (LWIR) x = 0.22 HgCdTe avalanche photodiodes (APDs) was presented. The dark current-voltage characteristics at temperatures 200 K, 230 K, and 300 K were measured and numerically simulated. Theoretical modeling was performed by the numerical Apsys platform (Crosslight). The effects of the tunneling currents and impact ionization in HgCdTe APDs were calculated. Dark currents exhibit peculiar features which were observed experimentally. The proper agreement between the theoretical and experimental characteristics allowed to determine the material parameters of the absorber was reached. The effect of the multiplication layer profile on the detector characteristics was observed but was found to be insignificant.
Go to article

Authors and Affiliations

Tetiana Manyk
ORCID: ORCID
Jan Sobieski
ORCID: ORCID
Kacper Matuszelański
Jarosław Rutkowski
ORCID: ORCID
Piotr Martyniuk
ORCID: ORCID
Download PDF Download RIS Download Bibtex

Abstract

Microstructure, mechanical and corrosion properties of as-cast pure zinc and its binary and ternary alloys with magnesium (Mg), and copper (Cu) additions were investigated. Analysis of microstructure conducted by scanning electron microscopy revealed that alloying additives contributed to decreasing average grain size compared to pure zinc. Corrosion rate was calculated based on immersion and potentiodynamic tests and its value was lower for materials with Cu content. Moreover, it was shown that the intermetallic phase, formed as a result of Mg addition, constitutes a specific place for corrosion. It was observed that a different type of strengthening was obtained depending on the additive used. The presence of the second phase with Mg improved the tensile strength of the Zn-based materials, while Cu dissolved in the solution had a positive effect on their elongation.
Go to article

Authors and Affiliations

Magdalena Gieleciak
Karolina Janus
ORCID: ORCID
Łukasz Maj
Paweł Petrzak
Magdalena Bieda
Anna Jastrzębska
Download PDF Download RIS Download Bibtex

Abstract

Federated Learning is an upcoming concept used widely in distributed machine learning. Federated learning (FL) allows a large number of users to learn a single machine learning model together while the training data is stored on individual user devices. Nonetheless, federated learning lessens threats to data privacy. Based on iterative model averaging, our study suggests a feasible technique for the federated learning of deep networks with improved security and privacy. We also undertake a thorough empirical evaluation while taking various FL frameworks and averaging algorithms into consideration. Secure Multi Party Computation, Secure Aggregation, and Differential Privacy are implemented to improve the security and privacy in a federated learning environment. In spite of advancements, concerns over privacy remain in FL, as the weights or parameters of a trained model may reveal private information about the data used for training. Our work demonstrates that FL can be prone to label-flipping attack and a novel method to prevent label-flipping attack has been proposed. We compare standard federated model aggregation and optimization methods, FedAvg and FedProx using benchmark data sets. Experiments are implemented in two different FL frameworks - Flower and PySyft and the results are analysed. Our experiments confirm that classification accuracy increases in FL framework over a centralized model and the model performance is better after adding all the security and privacy algorithms. Our work has proved that deep learning models perform well in FL and also is secure.
Go to article

Authors and Affiliations

R Anusuya
D Karthika Renuka
Download PDF Download RIS Download Bibtex

Abstract

The present work investigated the effect of modifying an epoxy resin using two different modifiers. The mechanical and thermal properties were evaluated as a function of modifier type and content. The structure and morphology were also analyzed and related to the measured properties. Polyurethane (PUR) was used as a liquid modifier, while Cloisite Na+ and Nanomer I.28E are solid nanoparticles. Impact strength (IS) of hybrid nanocomposites based on 3.5 wt% PUR and 2 wt% Cloisite or 3.5wt% PUR and 1wt% Nanomer was maximally increased by 55% and 30% respectively compared to the virgin epoxy matrix, exceeding that of the two epoxy/nanoparticle binaries but not that of the epoxy/PUR binary. Furthermore, a maximum increase in IS of approximately 20% compared to the pristine matrix was obtained with the hybrid epoxy nanocomposite containing 0.5 wt% Cloisite and 1 wt% Nanomer, including a synergistic effect, due most likely to specific interactions between the nanoparticles and the epoxy matrix. The addition of polyurethane and nanoclays increased significantly the thermal stability of epoxy composites. However, DSC results showed that the addition of flexible polyurethane chains decreased the glass transition temperatures, while the softening point and the service temperature range of epoxy nanocomposites containing nanofillers were increased. FTIR analysis confirmed the occurrence of interaction between the epoxy matrix and added modifiers. All SEM micrographs showed significant roughness of the fracture surfaces with the formation of elongated platelets, explaining the increase in mechanical properties of the epoxy matrix.
Go to article

Authors and Affiliations

Anita Białkowska
Patryk Suroń
Wojciech Kucharczyk
Barbora Hanulikova
Mohamed Bakar
ORCID: ORCID
Download PDF Download RIS Download Bibtex

Abstract

This study aims to evaluate the effectiveness of machine learning (ML) models in predicting concrete damage using electromechanical impedance (EMI) data. From numerous experimental evidence, the damaged mortar sample with surface-mounted piezoelectric (PZT) material connected to the EMI response was assessed. This work involved the different ML models to identify the accurate model for concrete damage detection using EMI data. Each model has been evaluated with evaluation metrics with the prediction/true class and each class is classified into three levels for testing and trained data. Experimental findings indicate that as damage to the structure increases, the responsiveness of PZT decreases. Therefore, examined the ability of ML models trained on existing experimental data to predict concrete damage using the EMI data. The current work successfully identified the approximately close ML models for predicting damage detection in mortar samples. The proposed ML models not only streamline the identification of key input parameters with models but also offer cost-saving benefits by reducing the need for multiple trials in experiments. Lastly, the results demonstrate the capability of the model to produce precise predictions.
Go to article

Authors and Affiliations

Asraar Anjum
Meftah Hrairi
Abdul Aabid
ORCID: ORCID
Norfazrina Yatim
Maisarah Ali
Download PDF Download RIS Download Bibtex

Abstract

The selection of a reference model (RM) for a Model-Reference Adaptive Control is one of the most important aspects of the synthesis process of the adaptive control system. In this paper, the four different implementations of RM are developed and investigated in an adaptive PMSM drive with variable moment of inertia. Adaptation mechanisms are based on the Widrow-Hoff rule (W-H) and the Adaptation Procedure for Optimization Algorithms (APOA). Inadequate order or inaccurate approximation of RM for the W-H rule may provide poor behavior and oscillations. The results prove that APOA is robust against an improper selection of RM and provides high-performance PMSM drive operation.
Go to article

Authors and Affiliations

Rafał Szczepański
Tomasz Tarczewski
Lech Grzesiak
Download PDF Download RIS Download Bibtex

Abstract

To improve the dynamic adaptability and flexibility of process route during manufacturing, a dynamic optimization method of multi-process route based on improved ant colony algorithm driven by digital twin is proposed. Firstly, on the basis of part manufacturing features analysis, the machining methods of each process are selected, and the fuzzy precedence constraint relationship between machining metas and processes is constructed by intuitionistic fuzzy information. Then, the multi-objective optimization function driven by digital twin is established with the optimization objectives of least manufacturing cost and lowest carbon emission, also the ranking of processing methods is optimized by an improved adaptive ant colony algorithm to seek the optimal processing sequence. Finally, the transmission shaft of some equipment is taken as an engineering example forverification analysis, which shows that this method can obtain a process route that gets closer to practical production.
Go to article

Authors and Affiliations

Zhaoming Chen
Jinsong Zou
Wei Wang
ORCID: ORCID
Download PDF Download RIS Download Bibtex

Abstract

This article examines in depth the most recent thermal testing techniques for lithium-ion batteries (LIBs). Temperature estimation circuits can be divided into six divisions based on modeling and calculation methods, including electrochemical computational modeling, equivalent electric circuit modeling (EECM), machine learning (ML), digital analysis, direct impedance measurement, and magnetic nanoparticles as a base. Complexity, accuracy, and computational cost-based EECM circuits are feasible. The accuracy, usability, and adaptability of diagrams produced using ML have the potential to be very high. However, both cannot anticipate low-cost integrated BMS live due to their high computational costs. An appropriate solution might be a hybrid strategy that combines EECM and ML.
Go to article

Authors and Affiliations

Ahmed Abd El Baset Abd El Halim
Ehab Hassan Eid Bayoumi
Walid El-Khattam
Amr Mohamed Ibrahim
Download PDF Download RIS Download Bibtex

Abstract

Direction-splitting implicit solvers employ the regular structure of the computational domain augmented with the splitting of the partial differential operator to deliver linear computational cost solvers for time-dependent simulations. The finite difference community originallye mployed this method to deliver fast solvers for PDE-based formulations. Later, this method was generalized into so-called variationals plitting. The tensor product structure of basis functions over regular computational meshes allows us to employ the Kronecker product structureo f the matrix and obtain linear computational cost factorization for finite element method simulations. These solvers are traditionally usedf or fast simulations over the structures preserving the tensor product regularity. Their applications are limited to regular problems and regularm odel parameters. This paper presents a generalization of the method to deal with non-regular material data in the variational splitting method. Namely, we can vary the material data with test functions to obtain a linear computational cost solver over a tensor product grid with nonregularm aterial data. Furthermore, as described by the Maxwell equations, we show how to incorporate this method into finite element methods imulations of non-stationary electromagnetic wave propagation over the human head with material data based on the three-dimensional MRI scan.
Go to article

Authors and Affiliations

Marcin Łoś
ORCID: ORCID
Maciej Woźniak
ORCID: ORCID
Maciej Paszynski
ORCID: ORCID
Download PDF Download RIS Download Bibtex

Abstract

The study concentrates on two different Genetic Programming approaches for determining Passenger Car Equivalent(PCE) values and observing the impact on capacity estimation at urban unsignalized intersections. Considering heterogeneoustraffic conditions, a new PCE value is introduced to encompass sustainable modes of public transit vehicles, specifically Slow-moving three-wheelers (SM3W), commonly known as E-Rickshaws. Since PCE value is considered an important parameter forcapacity calculations, the present study considered 14 unsignalized intersections located in Ranchi city of India. An automaticplate recognition system is employed to have the count of vehicular traffic. The methodologies include Age layered populationstructure genetic programming (ALPSGP), and the Offspring selection genetic programming (OSGP) approach that incorporatesstatic and dynamic variables. Based on the significance test and ranking of the Genetic programming (GP) models, the OSGPmodel is recommended as the most appropriate model for heterogeneous traffic. Sensitivity analysis reported that laggingheadway (����) is the most contributing factor in PCE estimation. The PCE value of SM3W is found to be 0.81 and that could beincorporated as a new classification of vehicles in Indo-HCM. It is observed that evaluated capacity based on OSGP’s PCEvalues performed admirably in both normal and congested traffic situations.
Go to article

Authors and Affiliations

Aarohi Kumar Munshi
Ashish Kumar Patnaik
Download PDF Download RIS Download Bibtex

Abstract

This paper addresses the problem of designing secure control for networked multi-agent systems (MASs) under Denial-of-Service (DoS) attacks. We propose a constructive design method based on the interaction topology. The MAS with a non-attack communication topology, modeled by quasi-Abelian Cayley graphs subject to DoS attacks, can be represented as a switched system. Using switching theory, we provide easily applicable sufficient conditions for the networked MAS to remain asymptotically stable despite DoS attacks. Our results are applicable to both continuoustime and discrete-time systems, as well as to discrete-time systems with variable steps or systems that combine discrete and continuous times.
Go to article

Authors and Affiliations

Ewa Girejko
1
Agnieszka Malinowska
1

  1. Bialystok University of Technology,Wiejska 45, 15-351 Białystok, Poland
Download PDF Download RIS Download Bibtex

Abstract

A companion robot is capable of performing a variety of activities and thus supporting the elderly and people withdisabilities. It should be able to overcome obstacles on its own, respond to what is happening around it in real-time, andcommunicate with its surroundings. It is particularly important to pay attention to these issues, as a companion robot is likely tobecome a participant in traffic. The aim of the research is to develop a mathematical model that takes into account the use of twonavigation solutions in the companion robot. Thanks to this, it will be possible to use the obtained mathematical relationships tocompare various types of navigation and make a rational choice, enabling the implementation of the assumed activities in aspecific external environment. What is new in this article is the analysis of several navigation methods and the presentation ofresearch carried out in real time using an actual robot.
Go to article

Authors and Affiliations

Karolina Krzykowska-Piotrowska
Emilia Grabka
Ewa Dudek
Adam Rosiński
ORCID: ORCID
Kamil Maciuk
Download PDF Download RIS Download Bibtex

Abstract

Directionality of light and modelling effects impact lighting quality in interiors. The modelling effects depend on luminaires’ photometric characteristics and their layout but also on interior size and reflectance. The objective of this research was to evaluate lighting design limitations and impact of interior and luminaires’ characteristics on the modelling effects, as well as elaborate a prediction method of the modelling effects in interior lighting. The General Index of Modelling was used for the analysis of the modelling effects in interiors. The implementation of the research objectives was based on the simulation and statistical analysis. 432 situations, varied interior size and reflectance, lighting class, luminaire downward luminous intensity distribution and layout were considered. The results show that achieving the required range of the General Index of Modelling in interior lighting is substantially limited. Luminaires’ layout impacts the General Index of Modelling the most. The elaborated multiple linear regression models can have a practical use for interior lighting design and analysis in terms of obtaining therequired range of the General Index of Modelling.
Go to article

Authors and Affiliations

Piotr Pracki
ORCID: ORCID
Paulina Komorzycka

This page uses 'cookies'. Learn more