Search results for: laminated segmented rotor flux switching permanent magnet machine

Authors: Oluwaseun Simon Adekanle, M'hammed Guisser, Elhassane Abdelmounim, Mohamed Aboulfatah

Abstract:

In this paper, modeling and control of a grid connected 660KW Doubly-Fed Induction Generator wind turbine is presented. Stator flux orientation is used to realize active-reactive power decoupling to enable independent control of active and reactive power. The recursive Integral Backstepping technique is used to control generator speed to its optimum value and to obtain unity power factor. The controller is combined with High Gain Observer to estimate the mechanical torque of the machine. The most important advantage of this combination of High Gain Observer and the Integral Backstepping controller is the annulation of static error that may occur due to incertitude between the actual value of a parameter and its estimated value by the controller. Simulation results under Matlab/Simulink show the robustness of this control technique in presence of parameter variation.

Keywords: doubly-fed induction generator, field orientation control, high gain observer, integral backstepping control

Procedia PDF Downloads 351

Authors: Luis Fonseca, Filipe Cabral Pinto, Susana Sargento

Abstract:

The increase of the world population, especially in large urban centers, has resulted in new challenges particularly with the control and optimization of public safety. Thus, in the present work, a solution is proposed for the prediction of criminal occurrences in a city based on historical data of incidents and demographic information. The entire research and implementation will be presented start with the data collection from its original source, the treatment and transformations applied to them, choice and the evaluation and implementation of the Machine Learning model up to the application layer. Classification models will be implemented to predict criminal risk for a given time interval and location. Machine Learning algorithms such as Random Forest, Neural Networks, K-Nearest Neighbors and Logistic Regression will be used to predict occurrences, and their performance will be compared according to the data processing and transformation used. The results show that the use of Machine Learning techniques helps to anticipate criminal occurrences, which contributed to the reinforcement of public security. Finally, the models were implemented on a platform that will provide an API to enable other entities to make requests for predictions in real-time. An application will also be presented where it is possible to show criminal predictions visually.

Keywords: crime prediction, machine learning, public safety, smart city

Procedia PDF Downloads 104

Authors: Ryan Messina, Mehedi Hasan

Abstract:

This research is to examine the impacts of using data to generate performance requirements for automation in visual inspections using machine vision. These situations are intended for design and how projects can smooth the transfer of tacit knowledge to using an algorithm. We have proposed a framework when specifying machine vision systems. This framework utilizes varying levels of automation as contingency planning to reduce data processing complexity. Using data assists in extracting tacit knowledge from those who can perform the manual tasks to assist design the system; this means that real data from the system is always referenced and minimizes errors between participating parties. We propose using three indicators to know if the project has a high risk of failing to meet requirements related to accuracy and reliability. All systems tested achieved a better integration into operations after applying the framework.

Keywords: automation, contingency planning, continuous engineering, control theory, machine vision, system requirements, system thinking

Procedia PDF Downloads 197

Authors: Shael Brown, Reza Farivar

Abstract:

Persistence diagrams capture valuable topological features of datasets that other methods cannot uncover. Still, their adoption in data pipelines has been limited due to the lack of publicly available tools in R (and python) for analyzing groups of them with machine learning and statistical inference. In an easy-to-use and scalable R package called TDApplied, we implement several applied analysis methods tailored to groups of persistence diagrams. The two main contributions of our package are comprehensiveness (most functions do not have implementations elsewhere) and speed (shown through benchmarking against other R packages). We demonstrate applications of the tools on simulated data to illustrate how easily practical analyses of any dataset can be enhanced with topological information.

Keywords: machine learning, persistence diagrams, R, statistical inference

Procedia PDF Downloads 78

Authors: Maitham Al-Safwani

Abstract:

Motor vibration issues were analyzed by several studies. It is generally accepted that vibration issues result from poor equipment installation. We had a water injection pump tested in the factory and exceeded the pump the vibration limit. Once the pump was brought to the site, its half-size shim plates were replaced with full-size shims plates that drastically reduced the vibration. In this study, vibration data was recorded for several similar motors run at the same and different speeds. The vibration values were recorded -for two and a half hours- and the vibration readings were analyzed to determine when the readings became consistent. This was as well supported by recording the audio noises produced by some machines seeking a relationship between changes in machine noises and machine abnormalities, such as vibration.

Keywords: vibration, noise, installation, machine

Procedia PDF Downloads 180

Authors: Arzu K. Kamberli, Tolga Ulusoy

Abstract:

Technological developments and the resulting global communication have made the 21st century when large capitals are moved from one end to the other via a button. As a result, the flow of capital inflows has accelerated, and capital inflow has brought with it crisis-related infectiousness. Considering the irrational human behavior, the financial crisis in the world under the influence of the whole world has turned into the basic problem of the countries and increased the interest of the researchers in the reasons of the crisis and the period in which they lived. Therefore, the complex nature of the financial crises and its linearly unexplained structure have also been included in the new discipline, econophysics. As it is known, although financial crises have prediction mechanisms, there is no definite information. In this context, in this study, using the concept of electric field from the electrostatic part of physics, an early econophysical approach for global financial crises was studied. The aim is to define a model that can take place before the financial crises, identify financial fragility at an earlier stage and help public and private sector members, policy makers and economists with an econophysical approach. 2001 Turkey crisis has been assessed with data from Turkish Central Bank which is covered between 1992 to 2007, and for 2001 Argentina crisis, data was taken from IMF and the Central Bank of Argentina from 1997 to 2007. As an econophysical method, an analogy is used between the Gauss's law used in the calculation of the electric field and the forecasting of the financial crisis. The concept of Φ (Financial Flux) has been adopted for the pre-warning of the crisis by taking advantage of this analogy, which is based on currency movements and money mobility. For the first time used in this study Φ (Financial Flux) calculations obtained by the formula were analyzed by Matlab software, and in this context, in 2001 Turkey and Argentina Crisis for Φ (Financial Flux) crisis of values has been confirmed to give pre-warning.

Keywords: econophysics, financial crisis, Gauss's Law, physics

Procedia PDF Downloads 149

Authors: Tae-Uk Yun, Pullip Chung, Kee-Young Kwahk, Hyunchul Ahn

Abstract:

Fake news is defined as the news articles that are intentionally and verifiably false, and could mislead readers. Spread of fake news may provoke anxiety, chaos, fear, or irrational decisions of the public. Thus, detecting fake news and preventing its spread has become very important issue in our society. However, due to the huge amount of fake news produced every day, it is almost impossible to identify it by a human. Under this context, researchers have tried to develop automated fake news detection using machine learning techniques over the past years. But, there have been no prior studies proposed an automated fake news detection method for Korean news to our best knowledge. In this study, we aim to detect Korean fake news using text mining and machine learning techniques. Our proposed method consists of two steps. In the first step, the news contents to be analyzed is convert to quantified values using various text mining techniques (topic modeling, TF-IDF, and so on). After that, in step 2, classifiers are trained using the values produced in step 1. As the classifiers, machine learning techniques such as logistic regression, backpropagation network, support vector machine, and deep neural network can be applied. To validate the effectiveness of the proposed method, we collected about 200 short Korean news from Seoul National University’s FactCheck. which provides with detailed analysis reports from 20 media outlets and links to source documents for each case. Using this dataset, we will identify which text features are important as well as which classifiers are effective in detecting Korean fake news.

Keywords: fake news detection, Korean news, machine learning, text mining

Procedia PDF Downloads 269

Authors: Aishi Kundu, Elhan Raza

Abstract:

"Necessity is the mother of invention" - Rapid increase in the global human population has directed the agricultural domain toward machine learning. The basic need of human beings is considered to be food which can be satisfied through farming. Farming is one of the major revenue generators for the Indian economy. Agriculture is not only considered a source of employment but also fulfils humans’ basic needs. So, agriculture is considered to be the source of employment and a pillar of the economy in developing countries like India. This paper provides a brief review of the progress made in implementing Machine Learning in the agricultural sector. Accurate predictions are necessary at the right time to boost production and to aid the timely and systematic distribution of agricultural commodities to make their availability in the market faster and more effective. This paper includes a thorough analysis of various machine learning algorithms applied in different aspects of agriculture (crop management, soil management, water management, yield tracking, livestock management, etc.).Due to climate changes, crop production is affected. Machine learning can analyse the changing patterns and come up with a suitable approach to minimize loss and maximize yield. Machine Learning algorithms/ models (regression, support vector machines, bayesian models, artificial neural networks, decision trees, etc.) are used in smart agriculture to analyze and predict specific outcomes which can be vital in increasing the productivity of the Agricultural Food Industry. It is to demonstrate vividly agricultural works under machine learning to sensor data. Machine Learning is the ongoing technology benefitting farmers to improve gains in agriculture and minimize losses. This paper discusses how the irrigation and farming management systems evolve in real-time efficiently. Artificial Intelligence (AI) enabled programs to emerge with rich apprehension for the support of farmers with an immense examination of data.

Keywords: machine Learning, artificial intelligence, crop management, precision farming, smart farming, pre-harvesting, harvesting, post-harvesting

Procedia PDF Downloads 99

Authors: Hy Dang, Bo Mei

Abstract:

The development of machine learning methods and techniques has opened doors for investigation in many areas such as medicines, economics, finance, etc. One active research area involving machine learning is stock market prediction. This research paper tries to consider multiple techniques and methods for stock movement prediction using historical price or price factors. The paper explores the effectiveness of some deep learning frameworks for forecasting stock. Moreover, an architecture (TimeStock) is proposed which takes the representation of time into account apart from the price information itself. Our model achieves a promising result that shows a potential approach for the stock movement prediction problem.

Keywords: classification, machine learning, time representation, stock prediction

Procedia PDF Downloads 140

Authors: Delphine Vendryes, Dushyanth Sekhar, Baojia Tong, Matthew Theisen, Chester Curme

Abstract:

Much of the data that inform the decisions of governments, corporations and individuals are harvested from unstructured documents. Data extraction is defined here as a process that turns non-machine-readable information into a machine-readable format that can be stored, for instance, in a database. In financial services, introducing more automation in data extraction pipelines is a major challenge. Information sought by financial data consumers is often buried within vast bodies of unstructured documents, which have historically required thorough manual extraction. Automated solutions provide faster access to non-machine-readable datasets, in a context where untimely information quickly becomes irrelevant. Data quality standards cannot be compromised, so automation requires high data integrity. This multifaceted task is broken down into smaller steps: ingestion, table parsing (detection and structure recognition), text analysis (entity detection and disambiguation), schema-based record extraction, user feedback incorporation. Selected intermediary steps are phrased as machine learning problems. Solutions leveraging cutting-edge approaches from the fields of computer vision (e.g. table detection) and natural language processing (e.g. entity detection and disambiguation) are proposed.

Keywords: computer vision, entity recognition, finance, information retrieval, machine learning, natural language processing

Procedia PDF Downloads 103

Authors: Jay L. Fu

Abstract:

Patients with Alzheimer's disease progressively lose their memory and thinking skills and, eventually, the ability to carry out simple daily tasks. The disease is irreversible, but early detection and treatment can slow down the disease progression. In this research, publicly available MRI data and demographic data from 373 MRI imaging sessions were utilized to build models to predict dementia. Various machine learning models, including logistic regression, k-nearest neighbor, support vector machine, random forest, and neural network, were developed. Data were divided into training and testing sets, where training sets were used to build the predictive model, and testing sets were used to assess the accuracy of prediction. Key risk factors were identified, and various models were compared to come forward with the best prediction model. Among these models, the random forest model appeared to be the best model with an accuracy of 90.34%. MMSE, nWBV, and gender were the three most important contributing factors to the detection of Alzheimer’s. Among all the models used, the percent in which at least 4 of the 5 models shared the same diagnosis for a testing input was 90.42%. These machine learning models allow early detection of Alzheimer’s with good accuracy, which ultimately leads to early treatment of these patients.

Keywords: Alzheimer's disease, clinical diagnosis, magnetic resonance imaging, machine learning prediction

Procedia PDF Downloads 138

Authors: Wei-Song Hung

Abstract:

We utilized pressure, vacuum, and evaporation-assisted self-assembly techniques through which graphene oxide (GO) was deposited on modified polyacrylonitrile (mPAN). The fabricated composite GO/mPAN membranes were applied to dehydrate 1-butanol mixtures by pervaporation. Varying driving forces in the self-assembly techniques induced different GO assembly layer microstructures. XRD results indicated that the GO layer d-spacing varied from 8.3 Å to 11.5 Å. The self-assembly technique with evaporation resulted in a heterogeneous GO layer with loop structures; this layer was shown to be hydrophobic, in contrast to the hydrophilic layer formed from the other two techniques. From the pressure-assisted technique, the composite membrane exhibited exceptional pervaporation performance at 30 C: concentration of water at the permeate side = 99.6 wt% and permeation flux = 2.54 kg m-2 h-1. Moreover, the membrane sustained its operating stability at a high temperature of 70 C: a high water concentration of 99.5 wt% was maintained, and a permeation flux as high as 4.34 kg m-2 h-1 was attained. This excellent separation performance stemmed from the dense, highly ordered laminate structure of GO.

Keywords: graphene oxide, self-assembly, alcohol dehydration, polyacrylonitrile (mPAN)

Procedia PDF Downloads 290

Authors: Mais Khader, Xingjie Wei

Abstract:

This research sheds light on female entrepreneurs by providing new insights on the survival predictions of companies led by females in the UK. This study aims to build a predictive machine learning model of the survival of female-led & co-led small & medium enterprises (SMEs) in the UK over the period 2000-2020. The predictive model built utilised a combination of financial and non-financial features related to both companies and their directors to predict SMEs' survival. These features were studied in terms of their contribution to the resultant predictive model. Five machine learning models are used in the modelling: Decision tree, AdaBoost, Naïve Bayes, Logistic regression and SVM. The AdaBoost model had the highest performance of the five models, with an accuracy of 73% and an AUC of 80%. The results show high feature importance in predicting companies' survival for company size, management experience, financial performance, industry, region, and females' percentage in management.

Keywords: company survival, entrepreneurship, females, machine learning, SMEs

Procedia PDF Downloads 91

Authors: S. Ozeri, D. Shmilovitz

Abstract:

Micro‐Turbine Generators (MTG) are small size power plants that consist of a high speed, gas turbine driving an electrical generator. MTGs may be fueled by either natural gas or kerosene and may also use sustainable and recycled green fuels such as biomass, landfill or digester gas. The typical ratings of MTGs start from 20 kW up to 200 kW. The primary use of MTGs is for backup for sensitive load sites such as hospitals, and they are also considered a feasible power source for Distributed Generation (DG) providing on-site generation in proximity to remote loads. The MTGs have the compressor, the turbine, and the electrical generator mounted on a single shaft. For this reason, the electrical energy is generated at high frequency and is incompatible with the power grid. Therefore, MTGs must contain, in addition, a power conditioning unit to generate an AC voltage at the grid frequency. Presently, this power conditioning unit consists of a rectifier followed by a DC/AC inverter, both rated at the full MTG’s power. The losses of the power conditioning unit account to some 3-5%. Moreover, the full-power processing stage is a bulky and costly piece of equipment that also lowers the overall system reliability. In this study, we propose a new type of power conditioning stage in which only a small fraction of the power is processed. A low power converter is used only to program the rotor current (i.e. the excitation current which is substantially lower). Thus, the MTG's output voltage is shaped to the desired amplitude and frequency by proper programming of the excitation current. The control is realized by causing the rotor current to track the electrical frequency (which is related to the shaft frequency) with a difference that is exactly equal to the line frequency. Since the phasor of the rotation speed and the phasor of the rotor magnetic field are multiplied, the spectrum of the MTG generator voltage contains the sum and the difference components. The desired difference component is at the line frequency (50/60 Hz), whereas the unwanted sum component is at about twice the electrical frequency of the stator. The unwanted high frequency component can be filtered out by a low-pass filter leaving only the low-frequency output. This approach allows elimination of the large power conditioning unit incorporated in conventional MTGs. Instead, a much smaller and cheaper fractional power stage can be used. The proposed technology is also applicable to other high rotation generator sets such as aircraft power units.

Keywords: gas turbine, inverter, power multiplier, distributed generation

Procedia PDF Downloads 234

Authors: Carlos A. Acosta, Amar Bhalla, Ruyan Guo

Abstract:

Scraped surface heat exchangers (SSHE) use a rotor shaft assembly with scraping blades to homogenize viscous fluids during the heat transfer process. Obtaining in-situ measurements is difficult because the rotor and scraping blades spin continuously inside the mixing chamber, obstructing the instrumentation pathway. Computational fluid dynamics simulations provide useful insight into the flow behavior around the scraper blades for a variety of fluids and blade geometries. However, numerical solutions often focus on the fluid dynamics and heat transfer phenomena of rotating flow, ignoring the glass-transition temperature and freezing point depression. This research studies the multi-phase fluid dynamics and freezing point depression inside the SSHE with non-isothermal conditions in a time dependent process using an aqueous solution that contains 13.5 wt.% high fructose corn syrup and CO₂. The computational results were validated with in-situ pressure, temperature, and optical spectroscopy measurements. Results from the numerical model show good quantitatively agreement with experimental values.

Keywords: computational fluid dynamics, freezing point depression, phase-transition temperature, multi-phase flow

Procedia PDF Downloads 141

Authors: P. S. Jagadeesh Kumar, Tracy Lin Huan, S. Meenakshi Sundaram

Abstract:

Alzheimer’s disease is one of the prevalent kind of ailment, expected for impudent reconciliation or an effectual therapy is to be accredited hitherto. Probable detonation of patients in the upcoming years, and consequently an enormous deal of apprehension in early discovery of the disorder, this will conceivably chaperon to enhanced healing outcomes. Complex impetuosity of the brain is an observant symbolic of the disease and a unique recognition of genetic sign of the disease. Machine learning alongside deep learning and decision tree reinforces the aptitude to absorb characteristics from multi-dimensional data’s and thus simplifies automatic classification of Alzheimer’s disease. Susceptible testing was prophesied and realized in training the prospect of Alzheimer’s disease classification built on machine learning advances. It was shrewd that the decision trees trained with deep neural network fashioned the excellent results parallel to related pattern classification.

Keywords: Alzheimer's diagnosis, decision trees, deep neural network, machine learning, pattern classification

Procedia PDF Downloads 291

Authors: Daisy Biswas, Samar Kanti Bose, M. Mozaffar Hossain

Abstract:

The development of value added composite products from bamboo with the application of gluing technology can play a vital role in economic development and also in forest resource conservation of any country. In this study, the gluability of Bambusa balcooa and Bambusa vulgaris, two locally grown bamboo species of Bangladesh was assessed. As the culm wall thickness of bamboos decreases from bottom to top, a culm portion of up to 5.4 m and 3.6 m were used from the base of B. balcooa and B. vulgaris, respectively, to get rectangular strips of uniform thickness. The color of the B. vulgaris strips was yellowish brown and that of B. balcooa was reddish brown. The strips were treated in borax-boric, bleaching and carbonization for extending the service life of the laminates. The preservative treatments changed the color of the strips. Borax–boric acid treated strips were reddish brown. When bleached with hydrogen peroxide, the color of the strips turned into whitish yellow. Carbonization produced dark brownish strips having coffee flavor. Chemical constituents for untreated and treated strips were determined. B. vulgaris was more acidic than B. balcooa. Then the treated strips were used to develop three-layered bamboo laminated panel. Urea formaldehyde (UF) and polyvinyl acetate (PVA) were used as binder. The shear strength and abrasive resistance of the panel were evaluated. It was found that the shear strength of the UF-panel was higher than the PVA-panel for all treatments. Between the species, gluability of B. vulgaris was better and in some cases better than hardwood species. The abrasive resistance of B. balcooa is slightly higher than B. vulgaris; however, the latter was preferred as it showed well gluability. The panels could be used as structural panel, floor tiles, flat pack furniture component, and wall panel etc. However, further research on durability and creep behavior of the product in service condition is warranted.

Keywords: Bambusa balcooa, Bambusa vulgaris, polyvinyl acetate, urea formaldehyde

Procedia PDF Downloads 258

Authors: You-Lin Wu, Yi-Hsing Sung, Shih-Hung Lin, Jing-Jenn Lin

Abstract:

Performance improvement in optoelectronic devices such as solar cells and photodetectors has been reported when a polymer/ZnO nanorods stack is used. Resistance switching of polymer/ZnO nanocrystals (or nanorods) hybrid has also gained a lot of research interests recently. It has been reported that high- and low-resistance states of a metal/insulator/metal (MIM) structure diode with a polystyrene (PS) and ZnO hybrid as the insulator layer can be switched by applied bias after a high-voltage forming process, while the same device structure merely with a PS layer does not show any forming behavior. In this work, we investigated the current-voltage (I-V) characteristics of an MIM device with a PS/ZnO nanorods stack deposited on fluorine-doped tin oxide (FTO) glass substrate. The ZnO nanorods were grown by a hydrothermal method using a mixture of zinc nitrate, hexamethylenetetramine, and DI water. Following that, a PS layer was deposited by spin coating. Finally, the device with a structure of Ti/ PS/ZnO nanorods/FTO was completed by e-gun evaporated Ti layer on top of the PS layer. Semiconductor parameters analyzer Agilent 4156C was then used to measure the I-V characteristics of the device by applying linear ramp sweep voltage with sweep sequence of 0V → 4V → 0V → 3V → 0V → 2V → 0V → 1V → 0V in both positive and negative directions. It is interesting to find that the I-V characteristics are bias dependent and hysteretic, indicating that the device Ti/PS/ZnO nanorods/FTO structure has ferroelectricity. Our results also show that the maximum hysteresis loop height of the I-V characteristics as well as the voltage at which the maximum hysteresis loop height of each scan occurs increase with increasing maximum sweep voltage. It should be noticed that, although ferroelectricity has been found in ZnO at its melting temperature (1975℃) and in Li- or Co-doped ZnO, neither PS nor ZnO has ferroelectricity at room temperature. Using the same structure but with a PS or ZnO layer only as the insulator does not give and hysteretic I-V characteristics. It is believed that a charge polarization layer is induced near the PS/ZnO nanorods stack interface and thus causes the ferroelectricity in the device with Ti/PS/ZnO nanorods/FTO structure. Our results show that the PS/ZnO stack can find a potential application in a resistive switching memory device with MIM structure.

Keywords: ferroelectricity, hysteresis, polystyrene, resistance switching, ZnO nanorods

Procedia PDF Downloads 309

Authors: Asad Islam, Khalid Parvez

Abstract:

Compressor fans in modern aircraft engines are of considerate importance, as they provide majority of thrust required by the aircraft. Their challenging environment is frequently subjected to non-uniform inflow conditions. These conditions could be either due to the flight operating requirements such as take-off and landing, wake interference from aircraft fuselage or cross-flow wind conditions. So, in highly maneuverable flights regimes of fighter aircrafts affects the overall performance of an engine. Since the flow in compressor of an aircraft application is highly sensitive because of adverse pressure gradient due to different flow orientations of the aircraft. Therefore, it is prone to unstable operations. This paper presents the study that focuses on axial compressor response to inlet flow orientations for the range of angles as 0 to 15 degrees. For this purpose, NASA Rotor-37 was taken and CFD mesh was developed. The compressor characteristics map was generated for the design conditions of pressure ratio of 2.106 with the rotor operating at rotational velocity of 17188.7 rpm using CFD simulating environment of ANSYS-CFX®. The grid study was done to see the effects of mesh upon computational solution. Then, the mesh giving the best results, (when validated with the available experimental NASA’s results); was used for further distortion analysis. The flow in the inlet nozzle was given angle orientations ranging from 0 to 15 degrees. The CFD results are analyzed and discussed with respect to stall margin and flow separations due to induced distortions.

Keywords: axial compressor, distortions, angle, CFD, ANSYS-CFX®, bladegen®

Procedia PDF Downloads 447

Authors: Y. Lacroix, J.-F. Malbranque

Abstract:

We present a chronological evolution for naval telecommunication networks. We distinguish periods: with or without multiplexers, with switch systems, with federative systems, with medium switching, and with medium switching with wireless networks. This highlights the introduction of new layers and technology in the architecture. These architectures are presented using layer models of transmission, in a unified way, which enables us to integrate pre-existing models. A ship of a naval fleet has internal communications (i.e. applications' networks of the edge) and external communications (i.e. the use of the means of transmission between edges). We propose architectures, deduced from the layer model, which are the point of convergence between the networks on board and the HF, UHF radio, and satellite resources. This modelling allows to consider end-to-end naval communications, and in a more global way, that is from the user on board towards the user on shore, including transmission and networks on the shore side. The new architectures need take care of quality of services for end-to-end communications, the more remote control develops a lot and will do so in the future. Naval telecommunications will be more and more complex and will use more and more advanced technologies, it will thus be necessary to establish clear global communication schemes to grant consistency of the architectures. Our latest model has been implemented in a military naval situation, and serves as the basic architecture for the RIFAN2 network.

Keywords: equilibrium beach profile, eastern tombolo of Giens, potential function, erosion

Procedia PDF Downloads 287

Authors: Mahtab Davari, Charles Edward Okon, Somayeh Aghanavesi

Abstract:

Every institute of learning is usually interested in the performance of enrolled students. The level of these performances determines the approach an institute of study may adopt in rendering academic services. The focus of this paper is to evaluate students' academic performance in given courses of study using machine learning methods. This study evaluated various supervised machine learning classification algorithms such as Logistic Regression (LR), Support Vector Machine, Random Forest, Decision Tree, K-Nearest Neighbors, Linear Discriminant Analysis, and Quadratic Discriminant Analysis, using selected features to predict study performance. The accuracy, precision, recall, and F1 score obtained from a 5-Fold Cross-Validation were used to determine the best classification algorithm to predict students’ performances. SVM (using a linear kernel), LDA, and LR were identified as the best-performing machine learning methods. Also, using the LR model, this study identified students' educational habits such as reading and paying attention in class as strong determinants for a student to have an above-average performance. Other important features include the academic history of the student and work. Demographic factors such as age, gender, high school graduation, etc., had no significant effect on a student's performance.

Keywords: student performance, supervised machine learning, classification, cross-validation, prediction

Procedia PDF Downloads 118

Authors: Navab Karimi, Tohid Alizadeh

Abstract:

An intelligent vision-based system was designed to measure the quality and purity of raisins. A machine vision setup was utilized to capture the images of bulk raisins in ranges of 5-50% mixed pure-impure berries. The textural features of bulk raisins were extracted using Grey-level Histograms, Co-occurrence Matrix, and Local Binary Pattern (a total of 108 features). Genetic Algorithm and neural network regression were used for selecting and ranking the best features (21 features). As a result, the GLCM features set was found to have the highest accuracy (92.4%) among the other sets. Followingly, multiple feature combinations of the previous stage were fed into the second regression (linear regression) to increase accuracy, wherein a combination of 16 features was found to be the optimum. Finally, a Support Vector Machine (SVM) classifier was used to differentiate the mixtures, producing the best efficiency and accuracy of 96.2% and 97.35%, respectively.

Keywords: sun-dried organic raisin, genetic algorithm, feature extraction, ann regression, linear regression, support vector machine, south azerbaijan.

Procedia PDF Downloads 68

Authors: Mohsen Torabi, Kaili Zhang

Abstract:

This work considers forced convection in a channel partially filled with porous media from local thermal non-equilibrium (LTNE) point of view. The channel is heated with constant heat flux from the lower side and is isolated on the top side. The wall heat flux is considered to be divided between the solid and fluid phases based on their temperature gradients and effective thermal conductivities. The general forms of the velocity and temperature fields are analytically obtained. To obtain the constant parameters for temperature equations, a numerical solution is considered. Using different thermophysical parameters, both velocity and temperature fields are comprehensively illustrated. Discussions regarding bifurcation phenomenon are provided. Since this geometry has not been considered yet, the present analysis is a useful addition to the literature on thermal performance of porous systems from LTNE perspective.

Keywords: local thermal non-equilibrium, forced convection, thermal bifurcation, porous-fluid interface, combined analytical-numerical solution

Procedia PDF Downloads 358

Authors: Mehrdad Ebrahimi, Oliver Schmitz, Axel Schmidt, Peter Czermak

Abstract:

“Produced water” (PW) is any fossil water that is brought to the surface along with crude oil or natural gas. By far, PW is the largest waste stream by volume associated with oil and gas production operations. Due to the increasing volume of waste all over the world in the current decade, the outcome and effect of discharging PW on the environment has lately become a significant issue of environmental concerns. Therefore, there is a need for new technologies for PW treatment due to increase focus on water conservation and environmental regulation. The use of membrane processes for treatment of PW has several advantages over many of the traditional separation techniques. In oilfield produced water treatment with ceramic membranes, process efficiency is characterized by the specific permeate flux and by the oil separation performance. Apart from the membrane properties, the permeate flux during filtration of oily wastewaters is known to be strongly dependent on the constituents of the feed solution, as well as on process conditions, e.g. trans-membrane pressure (TMP) and cross-flow velocity (CFV). The research project presented in these report describes the application of different ceramic membrane filtration technologies for the efficient treatment of oil-field produced water and different model oily solutions.

Keywords: ceramic membrane, membrane fouling, oil rejection, produced water treatment

Procedia PDF Downloads 174

Authors: Siwar Fadhel, Imen Bahri, Man Zhang

Abstract:

The switched reluctance machine (SRM) has undeniable qualities in terms of low cost and mechanical robustness. However, its highly nonlinear character and its uncertain parameters justify the development of complicated controls. In this paper, authors present the design of a robust H-infinity current controller for an 8/6 SRM with taking into account the nonlinearity of the SRM and with rejection of disturbances. The electromagnetic torque is indirectly regulated through the current controller. To show the performances of this control, a robustness analysis is performed by comparing the H-infinity and PI controller simulation results. This comparison demonstrates better performances for the presented controller. The effectiveness and robustness of the presented controller are also demonstrated by experimental tests.

Keywords: current regulation, experimentation, robust H-infinity control, switched reluctance machine

Procedia PDF Downloads 307

Authors: Mohamed H. Sorour, Hayam F. Shaalan, Heba A. Hani, Mahmoud A. El-Toukhy

Abstract:

This paper reports the results of nanofiltration (NF) polymeric membrane for the recovery of divalent ions (calcium and magnesium) from Red Seawater. Pilot plant experiments have been carried out using Alfa-Laval (NF 2517/48) membrane module. System was operated in both total recirculation mode (permeate and brine) and brine recirculation mode under hydraulic pressure of 15 bar. Impacts of some chelating agents on both flux and rejection have been also investigated. Results indicated that pure water permeability ranges from 17 to 85.5 L/m²h at 2-15 bar. Comparison with seawater permeability under the same operating pressure values reveals lower values of 8.9-31 L/m²h manifesting the effect of the osmotic pressure of seawater. Overall total dissolved solids (TDS) reduction was almost constant without incorporation of chelating agents. On the contrary of expectations, the use of chelating agents N-(2-hydroxyethyl) ethylene diamine-N,N´,N´-triacetic acid (HEDTA) and ethylene glycol bis (2-aminoethyl ether)-N,N,N´,N´-tetraacetic acid (EGTA) showed flux decline of about 3-15%. Analysis of rejection data of total recirculation mode showed reasonable rejection values of 35%, 59% and 90% for Ca, Mg and SO₄, respectively. Operating under brine recirculation mode only showed a decrease of rejection to 33%, 56% and 86% for Ca, Mg and SO₄, respectively. The use of chelating agents has no substantial effect on NF membrane performance except for increasing the total Ca rejection to 48 and 65% for EGTA and HEDTA, respectively. Results, in general, confirmed the powerful separation of NF technology for softening and recovery of divalent ions from seawater. It is anticipated that increasing operating pressure beyond the limits of our investigations would improve the rejection and flux values. A trade-off should be considered between operating cost (due to higher pressure and marginal benefits as manifested by expected improved performance). The experimental results fit well with the formulated rejection empirical correlations and the published ones.

Keywords: nanofiltration, seawater, recovery, calcium, magnesium

Procedia PDF Downloads 160

Authors: Shuen-Tai Wang, Fang-An Kuo, Chau-Yi Chou, Yu-Bin Fang

Abstract:

2016 has become the year of the Artificial Intelligence explosion. AI technologies are getting more and more matured that most world well-known tech giants are making large investment to increase the capabilities in AI. Machine learning is the science of getting computers to act without being explicitly programmed, and deep learning is a subset of machine learning that uses deep neural network to train a machine to learn  features directly from data. Deep learning realizes many machine learning applications which expand the field of AI. At the present time, deep learning frameworks have been widely deployed on servers for deep learning applications in both academia and industry. In training deep neural networks, there are many standard processes or algorithms, but the performance of different frameworks might be different. In this paper we evaluate the running performance of two state-of-the-art distributed deep learning frameworks that are running training calculation in parallel over multi GPU and multi nodes in our cloud environment. We evaluate the training performance of the frameworks with ResNet-50 convolutional neural network, and we analyze what factors that result in the performance among both distributed frameworks as well. Through the experimental analysis, we identify the overheads which could be further optimized. The main contribution is that the evaluation results provide further optimization directions in both performance tuning and algorithmic design.

Keywords: artificial intelligence, machine learning, deep learning, convolutional neural networks

Procedia PDF Downloads 200

Authors: Shola Haruna Adeosun, F. Adebiyi Ajoke, Odedeji Adeoye

Abstract:

Video billboard advertising by networks and brand switching was conceived out of inquisition at the huge billboard advertising expenditures made by the three major GSM network operators in Nigeria. The study was anchored on Lagos State Metropolis with a current census population over 1,000,000. From this population, a purposive sample of 400 was adopted, and the questionnaire designed for the survey was carefully allocated to members of this ample in the five geographical zones of the city so that each rung of the society was well represented. The data obtained were analyzed using tables and simple percentages. The results obtained showed that subscribers of these networks were hardly influenced by the video billboard advertisements. They overwhelmingly showed that rather than the slogans of the GSM networks carried on the video billboards, it was the incentives to subscribers as well as the promotional strategies of these organizations that moved them to switch from one network to another. These switching lasted only as long as the incentives and promotions were in effect. The results of the study also seemed to rekindle the age-old debate on media effects, by the unyielding schools of the theory of ‘all-powerful media’, ‘the limited effects media’, ‘the controlled effects media’ and ‘the negotiated media influence’.

Keywords: evaluation, impact, video billboard, marketing, services

Procedia PDF Downloads 245

Authors: Karthik Mittal

Abstract:

This study performs an unsupervised machine learning analysis to find clusters of related SNPs which highlight biological pathways that are important for the biological mechanisms of breast cancer. Studying genetic variations in isolation is illogical because these genetic variations are known to modulate protein production and function; the downstream effects of these modifications on biological outcomes are highly interconnected. After extracting the SNPs and their effect on different types of breast cancer using the MRBase library, two unsupervised machine learning clustering algorithms were implemented on the genetic variants: a k-means clustering algorithm and a hierarchical clustering algorithm; furthermore, principal component analysis was executed to visually represent the data. These algorithms specifically used the SNP’s beta value on the three different types of breast cancer tested in this project (estrogen-receptor positive breast cancer, estrogen-receptor negative breast cancer, and breast cancer in general) to perform this clustering. Two significant genetic pathways validated the clustering produced by this project: the MAPK signaling pathway and the connection between the BRCA2 gene and the ESR1 gene. This study provides the first proof of concept showing the importance of unsupervised machine learning in interpreting GWAS summary statistics.

Keywords: breast cancer, computational biology, unsupervised machine learning, k-means, PCA

Procedia PDF Downloads 137

Authors: Peter C. St. John, Michael F. Crowley, Yannick J. Bomble

Abstract:

Production of chemicals from engineered organisms in a batch culture typically involves a trade-off between productivity, yield, and titer. However, strategies for strain design typically involve designing mutations to achieve the highest yield possible while maintaining growth viability. Such approaches tend to follow the principle of designing static networks with minimum metabolic functionality to achieve desired yields. While these methods are computationally tractable, optimum productivity is likely achieved by a dynamic strategy, in which intracellular fluxes change their distribution over time. One can use multi-stage fermentations to increase either productivity or yield. Such strategies would range from simple manipulations (aerobic growth phase, anaerobic production phase), to more complex genetic toggle switches. Additionally, some computational methods can also be developed to aid in optimizing two-stage fermentation systems. One can assume an initial control strategy (i.e., a single reaction target) in maximizing productivity - but it is unclear how close this productivity would come to a global optimum. The calculation of maximum theoretical yield in metabolic engineering can help guide strain and pathway selection for static strain design efforts. Here, we present a method for the calculation of a maximum theoretical productivity of a batch culture system. This method follows the traditional assumptions of dynamic flux balance analysis: that internal metabolite fluxes are governed by a pseudo-steady state and external metabolite fluxes are represented by dynamic system including Michealis-Menten or hill-type regulation. The productivity optimization is achieved via dynamic programming, and accounts explicitly for an arbitrary number of fermentation stages and flux variable changes. We have applied our method to succinate production in two common microbial hosts: E. coli and A. succinogenes. The method can be further extended to calculate the complete productivity versus yield Pareto surface. Our results demonstrate that nearly optimal yields and productivities can indeed be achieved with only two discrete flux stages.

Keywords: A. succinogenes, E. coli, metabolic engineering, metabolite fluxes, multi-stage fermentations, succinate

Procedia PDF Downloads 207