Showing papers in "Journal of Physics: Conference Series in 2022"

Road Aerial Object Detection Based on Improved YOLOv5

Abstract: In this paper, the target detection technology based on deep learning is applied to the process of object detection in highway aerial photography. By detecting road objects such as vehicles or crosswalk, it lays the foundation for digitalization and informationization of roads. Firstly, the unmanned aerial vehicle is used to collect road images. Then based on YOLOv5 network, aiming at the problem of small detection target, attention mechanism is introduced to weigh different channels of feature graph, and SoftPool is introduced in SPP module to improve pooling operation and retain more detailed feature information. The experimental results show that the improved YOLOv5 network can identify and locate objects on aerial roads more accurately and effectively.

An Explicit Note on the Existence Theorem of Optimal Control Problem

Abstract: This short note presents an explicit step-by-step proof of the existence theorem of an optimal control problem applied to a deterministic model for a vector-borne disease.

Research on Safety Helmet Detection Algorithm of Power Workers Based on Improved YOLOv5

Abstract: The traditional helmet detection algorithm in power industry has low precision and poor robustness. In response to this problem, the helmet detection algorithm based on improved YOLOv5 (You only look once) is put forward in this paper. Firstly, the YOLOv5 network structure is improved. By increasing the size of the feature map, one scale is added to the original three scales, and the added 160*160 feature map can be used for the detection of small targets; Secondly, the K-means is used for re-clustering the helmet data set to get more suitable priori anchor boxes. The experimental results illustrate that the average accuracy of the improved YOLOv5 algorithm is increased by 2.9% and reaching 95% compared with the initial model, and the accuracy of helmet recognition is increased by 2.4% and reaching 94.6%. This algorithm reduces the rates of missing detection and misdetection of small target detection in original network, and has strong practicability and advanced nature. It can satisfy the requirements of real-time detection and has a certain role in promoting the safety of power industry.

Polyhydroxybutyrate/polylactic acid blends: An alternative feedstock for 3D printed bone scaffold model

Abstract: In this research, we developed 3D printing filaments from polyhydroxybutyrate (PHB)/poly(lactic acid) (PLA) blends to further its use in a fused filament fabrication (FFF) 3D printing technique as an alternative feedstock for manufacturing bone scaffold model. The filaments were fabricated with blending ratios of PHB/PLA at 100/0, 90/10, 70/30, 50/50, 30/70, 10/90, and 0/100 %wt. using an extrusion process. Furthermore, 10 phr of polypropylene glycol (PPG) was added as a processing aid to enhance the processability. The results of MFR showed that the suitable temperature for 3D printing of all blended filaments is 190 °C. The changes in thermal properties indicate the partial compatibility between PHB and PLA in the blends. PLA plays a vital role in improving the mechanical properties of PHB. 3D printing filament from PHB/PLA blends has been successfully developed.

Model-free control of the dynamic lift of a wind turbine blade section: experimental results

Abstract: This work addresses the problem of developing control algorithms for the control of the aerodynamic lift of wind turbine blades using air injection, taking into account disturbances caused by turbulent perturbations. For this, a test bench is used where the lift of a 2D blade section in a wind tunnel can be controlled by a set of micro-jets close to the trailing edge. Through a continuous, local identification of the lift variations a model-free control that does not need any prior knowledge of the system is proposed. It allows the control of the flow of the micro-jets and stabilizes the lift around a tracking reference. The ability of the proposed control algorithm to track the lift reference when subjected to external perturbations, i.e., gusts, is discussed. In particular, this work demonstrates that the lift can be set to particular values using the proposed control strategy, and can be re-stabilized to pre-gust lift conditions. Experimental results illustrate globally the feasibility of such a control.

Comparative analysis of multiple machine learning algorithms for epileptic seizure prediction

Abstract: Epilepsy is a common neurological disease that affects more than 2 percent of the population globally. An imbalance in brain electrical activities causes unpredictable seizures, which eventually leads to epilepsy. Neurostimulators have the power to intervene in advance and avoid the occurrence of seizures. Its efficiency can be increased with the help of heuristics like advanced seizure prediction. Early identification of preictal state will help easy activation of neurostimulator on time. This research concentrates on the performance analysis of various machine learning algorithms on recorded EEG data. Through this study, we aim to find the best model, which can be used to create an ensemble model for better learning. This involves modeling and simulation of classical machine learning technique like Logistic regression, Naive Bayes model, K nearest neighbors Random Forest, and deep learning techniques like an Artificial neural network, Convolutional neural networks, Long short term memory, and Autoencoders. In this analysis, Random Forest and Long Short-Term Memory performed well among all models in terms of sensitivity and specificity.

Fake News Detection from Online media using Machine learning Classifiers

Abstract: With the advancement in technology, the consumption of news has shifted from Print media to social media. The convenience and accessibility are major factors that have contributed to this shift in consumption of the news. However, this change has bought upon a new challenge in the form of “Fake news” being spread with not much supervision available on the net. In this paper, this challenge has been addressed through a Machine learning concept. The algorithms such as K-Nearest Neighbor, Support Vector Machine, Decision Tree, Naïve Bayes and Logistic regression Classifiers to identify the fake news from real ones in a given dataset and also have increased the efficiency of these algorithms by pre-processing the data to handle the imbalanced data more appropriately. Additionally, comparison of the working of these classifiers is presented along with the results. The model proposed has achieved an accuracy of 89.98% for KNN, 90.46% for Logistic Regression, 86.89% for Naïve Bayes, 73.33% for Decision Tree and 89.33% for SVM in our experiment.

Object detection based on Yolov4-Tiny and Improved Bidirectional feature pyramid network

Abstract: In the field of small object detection, Yolov4-Tiny is inadequate in feature extraction and does not make best of multi-scale features. In this paper, an improved BiFPN framework is proposed based on Yolov4-Tiny to increase object detection precision. Moreover, the Yolov4-Tiny is taken as the backbone network and introduce spatial pyramid pooling (SPP) to connect and merge multi-scale regions. Finally, our method can achieve 79.53% map on Pascal VOC dataset, which is 2.12% higher than the original Yolov4-Tiny model.

Safety Helmet Wearing Detection Based on YOLOv5 of Attention Mechanism

Abstract: Aiming at problems of low accuracy and strong detection interference of the existing safety helmet wearing detection algorithms, an object detection algorithm by adding the squeeze-and-excitation block based on the YOLOv5 algorithm is proposed in this paper. The proposed method can not only obtain the weight of picture channel, but also accurately separate the foreground and background of the picture. Keeping all parameters unchanged, the proposed method and the YOLOv5 algorithm are applied to detect the safety helmet data set in the experiment. The result shows that the YOLOv5 algorithm with the squeeze-and-excitation block has an average detection accuracy of 94.5% for safety helmets and an average detection accuracy of 92.7% for human heads. The mAP value detected by the proposed method is 2% ∼2.5% higher than using YOLOv5 algorithm directly.

Design and motion analysis of multi-motion mode pipeline robot

Abstract: Aiming to improve the adaptability level of single motion form pipeline robots, a wheeled pipeline robot with multiple motion modes is designed. The overall scheme of the pipeline robot is given. The self-adaptive diameter-changing mechanism, wheel displacement mechanism, and turning mechanism are designed. The motion and mechanical models of pipeline robots during travelling are established; on this basis, the robot‘s main body structure is optimised.

Research on Effective Path Planning Algorithm Based on Improved A* Algorithm

Abstract: In recent years, with the rapid development of technology, mobile robots are being applied in all aspects of production and life. The A* algorithm can be used for global path planning of mobile robots. Aimed at the problem of the traditional A* algorithm planning path having many turning points and do not satisfying the global optimality, an A* algorithm based on the adaptive neighborhood search and steering cost has been proposed. Based on the information of the surrounding obstacles, the algorithm adaptively selects the appropriate neighborhood to search for the optimal child node. By establishing the steering cost model of the mobile robot, the steering cost is joined into the evaluation function of the A* algorithm. Based on the adaptive neighborhood search, the A* algorithm with steering cost breaks through the constraint of eight neighborhood search nodes, shortens the path length, and uses the steering cost to search for the global optimal path with fewer turning points. The simulation results show that compared to the traditional A* algorithm, the total path cost of the A* algorithm based on adaptive neighborhood search and steering cost is reduced by 19.3%, and the number of turning points is reduced by 44.4%.

Enhancement of Security of Diffie-Hellman Key Exchange Protocol using RSA Cryptography.

Abstract: Cryptography is related and referred to as the secured transmission of messages amongst the sender and the intended receiver by ensuring confidentiality, integrity, and authentication. Diffie – Hellman (DH) key exchange protocol is a well-known algorithm that would generate a shared secret key among the sender and the intended receiver, and the basis of cryptosystems for using public and private key for encryption and decryption process. But it is severely affected by the Man in the Middle (MITM) attack that would intercept and manipulate thus eavesdropping the shared secret key. This paper proposes a model of integrating the public-key RSA cryptography system with the DH key exchange to prevent the MITM attack. The performance of the proposed work has been compared to the DH Key Exchange algorithm as well as RSA Cryptosystem to conclude for effectiveness of the proposed model.

Surrogate modelling of wind fields from point-wise atmospheric turbulence measurements

Abstract: We present an advanced model for the generation of synthetic wind fields that can be understood as an extension of the well-known Mann model. In contrast to such Gaussian random field models which control second-order statistics (i.e., velocity correlation tensors or spectra), we demonstrate that our extended model incorporates the effects of higherorder statistics as well. In particular, the empirically observed phenomenon of small-scale intermittency, a key feature of atmospheric turbulent flows, can be reproduced with high accuracy and at considerably low computational cost. Our method is based on a recently developed multipoint statistical description of a turbulent velocity field [J. Friedrich et al., J. Phys. Complex. 2 045006 (2021)] and consists of a superposition of multivariate Gaussian statistics with fluctuating covariances. Furthermore, we explicitly show how such superstatistical Mann fields can be constraint on a certain number of point-wise measurement data. We give an outlook on the relevance of such surrogate wind fields in the context of fatigue loads on wind turbines.

Research on Automatic Driving Target Detection Based on YOLOv5s

Abstract: Aiming at the problems of low detection accuracy, small targets and inaccurate recognition of automatic driving in natural scenes such as different lighting, different weather and different perspectives. The yolov5s network, migration learning method and varifocal loss function are used, and the hyper-parameter adjustment and network fine-tuning are completed. Experimental results show that the proposed method achieves mean Average Precision of 46.68%, and the detection speed can reach 7ms/frame. Automatic driving target detection based on YOLOv5s model network has fast training results, easy deployment, strong robustness, and high recognition accuracy in different weather conditions. It can not only meet the accuracy of automatic driving target detection, but also meet the real-time detection.

Multistate Active Combined Power and Message/Data Rate Adaptive Decentralized Congestion Control Mechanisms for Vehicular Ad Hoc Networks

Abstract: Vehicular ad hoc networks (VANETs) have emerged in time to reduce on-road fatalities and provide efficient information exchange for entertainment-related applications to users in a well-organized manner. VANETs are the most instrumental elements in the Internet of Things (IoT). The objective lies in connecting every vehicle to every other vehicle to improve the user’s quality of life. This aim of continuous connectivity and information exchange leads to the generation of more information in the medium, which could congest the medium to a larger extent. Decentralized congestion control (DCC) techniques are specified to reduce medium congestion and provide various safety applications. This article presents two DCC mechanisms that adapt message rate and data rate combined with transmit power control mechanism. These mechanisms are developed under multi-state active design proposed by the standard. The proposed methods deliver better performance over other mechanisms in terms of power, channel load, and channel utilization using real-time-based scenarios by simulation in SUMO.

Energy storage battery SOC estimate based on improved BP neural network

Abstract: The SOC estimation of the battery is the most significant functions of batteries’ management system, and it is a quantitative evaluation of electric vehicle mileage. Due to complex battery dynamics and environmental conditions, the existing data-driven battery status estimation technology is not able to accurately estimate battery status. Aiming at this problem, the multi-implicit BP neural network model and the error elimination due to genetic algorithm are combined to appraise the battery’s state of charge. Firstly, a multi-hidden layer BP neural network is applied to learn about the nonlinear connection between the battery SOC and the measurable variables of lithium-ion batteries, for instance, current, voltage, and temperature. Secondly, the prediction error of the neural network type is denoised by the genetic method to smooth the prediction results. The method proposed in this paper captures long-term dependencies between measurable variables and battery state. Finally, the improvement effect of the method proposed in this paper is verified by comparison with the traditional neural network method.

An Open-Source Frequency-Domain Model for Floating Wind Turbine Design Optimization

Abstract: A new frequency-domain dynamics model has been developed that uses open-source components to efficiently represent a complete floating wind turbine system. The model, called RAFT (Response Amplitudes of Floating Turbines), incorporates quasi-static mooring reactions, strip-theory and potential-flow hydrodynamics, blade-element-momentum aerodynamics, and linear turbine control. The formulation is compatible with a wide variety of support structure configurations and no manual or time-domain preprocessing steps are required, making RAFT very practical in design and optimization workflows. The model is applied to three reference floating wind turbine designs and its predictions are compared with results from time-domain OpenFAST simulations. There is good agreement in mean offsets as well the statistics and spectra of the dynamic response, verifying RAFT’s general suitability for floating wind analysis. Follow-on work will include verification of potential-flow and turbine-control features and application to optimization problems.

Comparison of various machine learning approaches uses in heart ailments prediction

Abstract: Heart disease has been the leading cause of a huge number of deaths in recent years. As a result, an accurate and feasible system is required to diagnose this disease early to provide better treatment. Advances in machine learning have the potential to enhance healthcare access. Given the importance of a crucial organ like the heart, medical professionals and physicians have made it a priority to forecast heart failure-related events in clinical practice, nevertheless, forecasting heart failure-related events in clinical practice has generally failed to achieve high accuracy. The objective here is to demonstrate how machine learning may be used to solve the problem. By analyzing hundreds of healthcare data and other semantics, machine learning algorithms can analyze related cases with diseases and health conditions. Here a demonstration of how to load the data, generate predictions through different models from patient data is shown. The metrics are then compared for a better understanding of their function and what impact can be inferred from them.

Field-Induced Quantum Spin Nematic Liquid Phase in the S=1 Antiferromagnetic Heisenberg Chain with Additional Interactions

Abstract: The magnetization process of the S = 1 antiferromagnetic chain with the single-ion anisotropy D and the biquadratic interaction is investigated using the numerical diagonalization. Both interactions stabilize the 2-magnon Tomonaga-Luttinger liquid (TLL) phase in the magnetization process. Based on several excitation gaps calculated by the numerical diagonalization, some phase diagrams of the magnetization process are presented. These phase diagrams reveal that the spin nematic dominant TLL phase appears at higher magnetizations for sufficiently large negative D.

Survey of Image Augmentation Based on Generative Adversarial Network

Abstract: On the visual side of computer science, image data is very important in the training of neural network models. Sufficient training data can alleviate the over-fitting problem of the model during training and help the model obtain the optimal solution. However, in many computer vision assignments, it is not easy and costly to obtain sufficient training samples. Therefore, image augmentation has become a commonly used method to increase training samples. Generative Adversarial Network (GAN) is a generative method of machine learning that can generate realistic images and provide a new solution for image augmentation. This article first introduces image augmentation and its commonly used four types of methods. Secondly, the basic principles of GAN and its direct and integrated methods in image augmentation are introduced, and the typical methods used to calculate whether the images from the networks meets the requirements; then the research status of GAN in image augmentation is analyzed. Finally, the problems and development trends of GAN model in image augmentation are summarized and prospected.

Network intrusion detection: a comparative study of four classifiers using the NSL-KDD and KDD’99 datasets

Abstract: Abstract As most of the population acquires access to the internet, protecting online identity from threats of confidentiality, integrity, and accessibility becomes an increasingly important problem to tackle. By definition, a network intrusion detection system (IDS) helps pinpoint and identify anomalous network traffic to bring forward and classify suspicious activity. It is a fundamental part of network security and provides the first line of defense against a potential attack by alerting an administrator or appropriate personnel of possible malicious network activity. Several academic publications propose various artificial intelligence (AI) methods for an accurate network intrusion detection system (IDS). This paper outlines and compares four AI methods to train two benchmark datasets- the KDD’99 and the NSL-KDD. Apart from model selection, data preprocessing plays a vital role in contributing to accurate solutions, and thus, we propose a simple yet effective data preprocessing method. We also evaluate and compare the accuracy and performance of four popular models- decision tree (DT), multi-layer perceptron (MLP), random forest (RF), and a stacked autoencoder (SAE) model. Of the four methods, the random forest classifier showed the most consistent and accurate results.

Real Time Face Mask Detection using Google Cloud ML and Flutter

Abstract: Coronavirus (Covid-19) pandemic has impacted the whole world and has forced health emergencies internationally. The contact of this pandemic has been fallen over almost all the development sectors. A lot of precautionary measures have been taken to control the Covid-19 spread, where wearing a face mask is an essential precaution. Wearing a face mask correctly has been essential in controlling the Covid-19 transmission. Moreover, this research aims to detect the face mask with fine-grained wearing states: face with the correct mask and face without mask. Our work has two challenging tasks due to two main reasons firstly the presence of augmented data set available in the online market and the training of large datasets. This paper represents a mobile application for face mask detection. The fully automated Machine Learning Cloud service known as Google Cloud ML API is used for training the model in TensorFlow file format. This paper highlights the efficiency of the ML model. Additionally, this paper examines the advancement of the cloud technology used for machine learning over the traditional coding methods.

HIKE: High Intensity Kaon Experiments at the CERN SPS

Abstract: The availability of high intensity kaon beams at the CERN SPS North Area gives rise to unique possibilities for sensitive tests of the Standard Model in the quark flavor sector. Precise measurements of the branching ratios for the flavor-changing neutral current decays K→πνν− can provide unique constraints on CKM unitarity and, potentially, evidence for new physics. Building on the success of the NA62 experiment, plans are taking shape at CERN for a comprehensive program that will include experimental phases to measure the branching ratio for K+→π+νν− to ∼5% and to KL→π0νν− to ∼20% precision. These planned experiments would also carry out lepton flavor universality tests, lepton number and flavor conservation tests, and perform other precision measurements in the kaon sector, as well as searches for exotic particles in kaon decays. We overview the physics goals, detector requirements, and project status for the next generation of kaon physics experiments at CERN.

Effects of self-induced gravity waves on finite wind-farm operations using a large-eddy simulation framework

Abstract: In the present study, we use large-eddy simulation (LES) to investigate how a capping inversion in combination with a stable free atmosphere influences the flow development and energy extraction in a large finite wind farm with a staggered and aligned layout. In the conventionally neutral boundary layer (CNBL), we find that gravity waves induce an unfavourable pressure gradient in the induction region of the farm which contributes to the upstream blockage, decreasing the available energy for first-row turbines. However, a favourable pressure gradient establishes through the farm in such conditions, which redistributes the energy and enhances wake recovery. These results are compared with a farm operating in the neutral boundary layer (NBL). Here, we find that only hydrodynamic effects induced by the turbines drag play a role, which cause minor pressure perturbations across the domain. For the selected atmospheric conditions, the power losses generated by the upstream blockage are balanced by the enhanced wake recovery promoted by the favourable pressure gradient throughout the farm. Consequently, the staggered farm efficiency in the CNBL is 8.8% higher than in the NBL. We note that this difference in efficiency is slightly enhanced by the 0.5? difference in wind direction at the location of the first-row turbines between the CNBL and NBL cases, which is caused by the presence of flow blockage. Since both simulations are forced with an equal turbulent velocity profile, the variation in performance is solely caused by the different vertical temperature profiles in the main domain. Finally, the staggered layout leads to a slightly stronger flow blockage than the aligned one when both farms operate in the CNBL.

Design of Intelligent Garbage Classification System Based on Internet of Things Technology

Abstract: With the increasingly prominent problem of environmental pollution, it is extremely urgent to carry out garbage classification. This paper designs an intelligent garbage classification system based on Internet of Things technology, The system is mainly composed of relay driving circuit, infrared induction, metal detection and humidity detection modules. Single chip microcomputer and multi-channel sensors are used to collect and process related data to realize metal garbage recovery, dry garbage and wet garbage classification and delivery, and the collected related data are displayed on the display screen through serial communication. The experimental results show that the system has the characteristics of simple structure, stable performance and convenient operation, which provides a feasible solution for the current garbage classification and treatment.

A Comparative Study on Part-of-Speech Taggers’ Performance on Examination Questions Classification According to Bloom’s Taxonomy

Abstract: Examination questions classification according to Bloom’s Taxonomy uses Natural Language Processing (NLP) approach, a series of text processing approach that generally can divided into the keywords identification stage and then the identified keywords classification to Bloom’s Taxonomy levels stage. Since this NLP approach is a pipeline processes, the keywords identification stage’s performance in term of accuracy is affecting the subsequent stage - the identified keywords classification and subsequently limits the final accuracy performance of the questions classification. The keywords identification stage’s performance is mainly depending on the effectiveness of Part-Of-Speech (POS) tagging. Thus, this paper aims to identify the best performing POS tagger in keywords identification stage and enhance the tagger’s performance with rule-based approach to achieve high accuracy performance and benefit the subsequent keyword classification and then the questions classification accuracy. The Perceptron tagger and the Stanford POS tagger are selected to be evaluated their performance in identifying the keywords of the randomly selected 200 examination questions from STEM subjects. This paper has observed the Stanford POS tagger is the best performing tagger in POS tagging with accuracy of 80.5%. Some rules are applied to the POS tagging to improve the accuracy further to 91.5%.

Searching for New Physics with DarkLight at the ARIEL Electron-Linac

Abstract: The search for a dark photon holds considerable interest in the physics community. Such a force carrier would begin to illuminate the dark sector. Many experiments have searched for such a particle, but so far it has proven elusive. In recent years the concept of a low mass dark photon has gained popularity in the physics community. Of particular recent interest is the 8Be and 4He anomaly, which could be explained by a new fifth force carrier with a mass of 17 MeV/c 2. The proposed Darklight experiment would search for this potential low mass force carrier at ARIEL in the 10-20 MeV/c 2 e+e− invariant mass range. This proceeding will focus on the experimental design and physics case of the Darklight experiment.

Optimization of a broad-side coupling modal filter by evolutionary strategy algorithm with setting the ranges of the optimization parameters

Abstract: The paper considers the optimization of a modal filter with broad-side coupling by evolutionary strategy algorithm with limitations (i.e., setting the ranges of the optimization parameters). An improved block diagram of the algorithm with limitations is presented. The sequential optimization that uses the original and improved algorithms with different numbers of computations is performed. The performance of the improved algorithm is demonstrated. The main advantages and disadvantages of this algorithm are shown.

Design of a mobile application for the vaccination registry of children under 5 years

Abstract: Information and Communication Technologies are allowing to improve the development of certain processes optimizing the management of information for users, as in the case of the vaccination control process to digitize the use of the vaccination card, being very important for the registration of the administration of vaccines to the child during his five years. This information should be easily accessible and available to parents and health personnel during each stage of vaccination. In the present work we developed a mobile application called Children’s Vaccine, with the purpose of registering and storing vaccine data (type, date and place of vaccination), as well as relevant information for the management of vaccines, the prototype developed was implemented in the health Center: Juan Pablo II Maternal and Child Center located in the District of Villa el Salvador, in Lima, Peru, in order to improve the quality of the service of attention. We present the results as performance tests to measure the influence of the application, this evaluation was carried out through an observation sheet with the comparison of records before and after the use of the application called “pre-test” and “post-test”. This application is applicable and scalable for various health centers.

Innovation of simple binoculars as a teaching aid for optical practicum at school

Abstract: Learning aids are tools used by teachers in learning and prevent verbalization in students. The purpose of this research is to make simple earth telescopes as visual aids for optical practicum in schools. The method used is the experimental method, namely analyzing, designing, making and testing the props made. The subjects of the research were 17 students of physics education FKIP University of Mataram. The binoculars test saw the object of letters and writing with the font Times New Roman 350 at a distance of 50 meters and 100 meters. The results showed that the object was clearly visible with the satisfaction level of using the teaching aids a score of 3.71 the criteria were satisfied and suitable for use. The use of simple earth telescopes is suitable for use in optical practicums in schools. Teachers are expected to be able to create and develop teaching aids to improve students’ understanding of concepts.