next day
all days

View: session overviewtalk overview

09:00-10:00 Session 1: Keynote Lecture
Semiconductor Devices and Passive Components Characterization for High Frequency Applications
10:10-11:50 Session 2A: Electrical
Differential Protection Analisys For Detection Of Ground Faults On Stator Of Generators

ABSTRACT. The analysis of the restricted earth fault differential protection based on the implementation of a low impedance scheme that allows the protection of 3.5 kVA generator windings using the overcurrent relay in differential structure is presented. The configuration and testing of the 7SJ85 overcurrent protection relay is mainly focused on the built-in ANSI 87N, REF protection function. The configuration, parameterization and operation of the elements that make up the protection system are performed using OMICRON’s CMC 356 and the 7SJ85 relay with their respective application software Test Universe and Digsi 5. The values used for the tests are based on the short-circuit analysis applying the IEC 60909 standard and the Complete method (equivalent to Thevenin), which were calculated and simulated in MatLab and DigSilent software respectively.

Modeling of VSC-based BESS and PV Plants as Distributed Resources for Steady-State Studies

ABSTRACT. This paper addresses the modeling of VSC-based distributed energy resources such as battery energy storage systems (BESS) and photovoltaic (PV) power plants for steady-state power flow studies of electrical networks. These components are modeled in detail following their basic principles of operation for practical results. The modeling and solution framework is based on the well-known Newton-Raphson method. And so, the main hallmark of this formulation is that the most essential internal variables of the BESS and PV units are readily available upon convergence. This formulation is applied to a 33-bus electrical network including four PV plants and two BESS to analyze its steady-state performance over a 24-hr multi-period operation with varying conditions of both solar irradiance of PV plants and set point powers of BESS.

Modelling and Simulation of Brushless DC Motor Considering Magnetic Saturation

ABSTRACT. In this paper, the modelling and simulation of a brushless DC motor including magnetic saturation is presented. Hypotheses related to the behaviour of the dispersed magnetic flux, the flux in the air gap and the reluctance of the magnetic material are discussed and adopted. The simulation of the model is carried out in five stages and it includes different alternatives for permanent magnets materials. The results obtained are compared with those available in the Simulink block.

Implementation of an IEC61850 Virtual Relay Network in a Protection Laboratory

ABSTRACT. Electrical protections fulfill an elementary function within electrical power systems, it is a field that is constantly changing according to the needs of each of the equipment. The document presents the implementation of a virtual relay that will have a configuration equal to that made in the physical relay. In this case study, a subtransmission line will be analyzed in which there will be a main protection (distance relay) and a backup protection (overcurrent relay) and the necessary adjustments will be made for each of them. Both GOOSE and MMS messaging are configured through an IEC 61850 station with the help of the IEC 61850 System Configurator® software. As a last point, obtaining the SCL files necessary to duplicate the physical relay in a virtual relay using the IEDScout software will be analyzed. Horizontal communication in GOOSE and MMS between the virtual relays and the server was validated.

Determination of Optimal Steady-State Equilibrium Points in Electrical Systems Constrained by Static Thermal Limits, using the Interior Point Method

ABSTRACT. This paper presents a digital program that automatically determines the static thermal limits of power system transmission lines, according to the procedure recommended by IEEE Std. 738, from conventional power flow data and some additional meteorological variables. The static model of the transmission thermal limits, and a concise description of the heat balance equations and their adaptation to this approach are presented. Subsequently, the limits are included as operating constraints of the transmission lines within an optimal power flow study to obtain a steady state operating point, using a non-linear interior point optimization method, which due to its characteristics and the inclusion of the logarithmic barrier, can provide a better quality in the local optimum found.

10:10-11:50 Session 2B: Data Science
Data analysis architecture using Techniques of Machine Learning for the prediction of the quality of blood donations against the hepatitis C virus

ABSTRACT. Nowadays the WHO (World health Organization) has difficulties improving the access to safe blood. The WHO have published that the problem with blood donations is that of the millions of blood donations that they receive one in four donations made from low-income countries do not test all the donated blood. This is a big problematic because a hospital cannot ensure a patient if the blood, he/she is receiving is safe. As a solution to this problematic, we have proposed the use a method based on CRISP-DM, where as a first procedure we apply a preparation to the data, then we prepared the dataset by cleaning the null variables, transforming the dataset by applying Hot Encoding, analysis the data with PCA (Principal Component Analysis) and using the 85% of variance, and using oversampling for the class that we have chosen. Once the dataset has been preprocessed we proceed to use the techniques of machine learning to help evaluate if a donor’s blood is qualified or not for its use. We have applied a variety of machine learning techniques such as: RandomForest, KNN (K-Nearest-Neighbor), SVM (Support Vector Machine), and a neural network ANN (Artificial Neural Network). As a final step, we interpreted the results and got to a conclusion that the classifier that had the highest precision is the Random Forest classifier. For this this research we found a public dataset gathered by the university of Germany. This investigation has the objective to help improve the detection of hepatitis C in low-income countries and hopes to help improve the access to safe blood for patients who need them. In addition, we can apply this data analysis method for future investigations from which we encourage that tests be made with other techniques or models to analyze data.

Design and development of an intent-based intelligent network using machine learning for QoS provisioning

ABSTRACT. The demand for bandwidth is currently a challenge for the use of the Internet that companies require, it is common to see daily the exorbitant amount of data that circulate through the network as files, calls, video calls, online shopping or subscription streaming services, this generates a bottleneck in network traffic therefore this makes maintenance and management difficult requiring time and human effort. To solve this problem it is proposed to use the architecture of differentiated services to prioritize a type of traffic and obtain QoS, in addition the automation of activities will be used, that is, with artificial intelligence (AI) a neural network is developed to identify patterns and obtain predictions, machine learning (ML) that will predict when there will be events that alter the resources in the network. To demonstrate the effectiveness of the method, a proprietary dataset generated with data from the developed infrastructure is used, therefore the methods are evaluated with the quality metric MAE. At the end an Intention Based Network (IBN) will have been implemented, therefore this research intends to leave a base so that the proposed system can be improved or other methods can be developed to automate data centers.

Methodology for Weapon Detection in Social Media Profiles using an Adaptation of YOLO-V5 and Natural Language Processing Techniques

ABSTRACT. Weapon identification has been a hot topic in the area of Object Recognition in recent years. However, its application has been virtually explored in social media. This work focuses on the detection of weapons in profiles that explicitly advocate their procession, both graphically and textually. This is a challenge, since access to a dataset is difficult; and once the samples are obtained, the dimensions and attributes of the images can vary significantly. In addition, the possession of a weapon does not imply that any offense or crime is being committed. To tackle this challenges, this manuscript presents a regularized adaptation of a Fast-Convolutional Neural Network (F-CNN) based on YOLO-V5, to merge and improve the results of the algorithm, along with a textual fingerprinting technique, to first corroborate if the intent of the post contains red flags of crime and violence. The results demonstrate that regularized adaptive models, mainly using Data Image Augmentation techniques, along with text classification, can provide better performance on unstructured data, such as those found in social media.

Exhaustive Search Applied to Time Series Forecasting Methods Using Parallel Processing

ABSTRACT. This paper presents an algorithm that performs an exhaustive search of the parameters available in the time series forecasting methods of moving averages (MA), weighted moving averages (WMA), exponential smoothing (ES), least squares (LS), and Holt-Winters (HW). In order to select the optimal parameter and, based on reducing the mean absolute percentage error (MAPE), present the best semiannual forecast of electricity demand. Through the use of parallel processing based on threads, the implementation of the algorithm is carried out, since, by means of a multi-processor architecture, the process involved in the exhaustive search is distributed and divided into small tasks in order to optimize the execution time.

An analysis method for predicting breast cancer using data science processes and machine learning

ABSTRACT. In two decades, the number of people with breast cancer has almost doubled: in 2000, about 10 million patients had the disease; by 2020, it had reached 19 million. It is estimated that one in five people today will develop some form of cancer in their lifetime. Studies suggest that the number of people diagnosed with cancer will increase in the coming years, being approximately 50% higher in 2040 than in 2020. This article provides an analysis method to predict or diagnose breast cancer using data science processes and machine learning. The analysis method is structured into three phases. The first one is a data preparation phase, the second one is a predictive analysis phase, and the last one is an evaluation metric. Therefore, the predictions are experimented with machine learning techniques, which are: KNN, gradient boosting classifier, and random forest, for which evaluation metrics are presented with the next quality measures: accuracy, precision, recall, and f1 score. The dataset selected for this phase of analysis is Wisconsin breast cancer. These data analysis techniques can be extended to other learning techniques and can also be used in future scientific work such as disease prediction or medicine in general.

10:10-11:50 Session 2C: Electronics
Fault Diagnosis in Nonlinear Systems based on Real Time Generalized Unknown Inputs Observer Scheme (GUIOS)

ABSTRACT. This work presents a design method for a Generalized Unknown Input Observer Scheme (GUIOS), applied to a Two-Tank laboratory prototype, modeled as a Takagi- Sugeno system, under a LMI approach. The proposed scheme was validated in simulation using Matlab® and with real time experimentation.

iPaddle: An Intelligent Swimming Training System.

ABSTRACT. This paper presents an implementation of an intel- ligent swimming rowing system which consists of a swimming paddle attached to a motor connected by an inextensible cable. This system consists of a combination of two elements to create a device that simulates an aquatic environment out of water The first element consists of: an inertial system measurement unit, a communication device and a battery; all placed inside the swimming paddle. The second element is composed of an odometric arrangement consisting of an effort returning motor that coils the cable to compute by odometry the distance from motor to the paddle. In combination with the acceleration and rotational speed provided by the IMU, the motor apply resistant to simulate an aquatic environment. This device has proved is uttility because as being out of the water, the coach has a better control to analyze and correct the swimming technique of the athletes. Moreover, position, speed, acceleration and effort done by swimmer can be analyzed later offline.

Real-time Optimal Harmonics Estimation: An Internet of Things Approach

ABSTRACT. In this work, the application of an optimal dynamic estimation algorithm for the harmonic content in a waveform of a current signal in an electrical system is presented. The harmonics estimation is a task that requires a great computational effort, however, in this work, the real-time implementation of a harmonic estimation algorithm in a low-cost device (NI myRIO®) has been demonstrated. Additionally, the transmission of the obtained data through a Wi-Fi network to a cloud server is presented, making the information obtained by the estimator available in a cloud service accessible from any mobile device. Experimental results in the identification of the harmonic content produced by the operation of a nonlinear load in an electrical network, demonstrate the effectiveness of the optimal dynamic harmonic estimator, its implementation in real-time and data transmission.

A Fault Diagnosis Comparative Approach for a Quadrotor UAV.

ABSTRACT. This work presents a comparative between a Fault Diagnosis and Isolation (FDI) approach, and a Fault Detection and Estimation scheme (FDE) applied to a quadrotor UAV gyroscopic sensor. The system is modeled using a convex Takagi-Sugeno strategy and a descriptor representation for the vehicle rotational dynamics. The proposed schemes are based upon convex observer design, using a linear matrix inequality (LMI) approach. Simulations on the nonlinear model are presented in order to validate the schemes.

Time delay detector in seismic signals using real time multipoint radiofrequency links

ABSTRACT. Seismic waves are composed of a superposition of waves that differ in their frequencies, speed of propagation, and amplitude. In particular, primary seismic waves (P waves) travel faster than secondary seismic waves (S waves) which are more destructive. This property allows P waves to be detected first before the presence of S waves, which gives the opportunity to activate an alarm, if considered so. This project consists of the proposal of a system to detect the P waves of an earthquake in different geographical points. This is done by means of seismometers based on Lehman pendulums, and establishing radio communication links between the different nodes. The signals from the seismographs are continuously Fourier transformed and cross-correlated to discern between the presence of a seismic signal or noise. Experimentally, a real-time multipoint communication system was developed within the UHF band, made up of detectors at two distant points that transmit data from seismic sensors through a transceiver to a third central node, which, apart from having its own detector, performs signal processing and information display through a graphical interface.

10:10-11:50 Session 2D: Advanced Signal Processing Techniques for Condition Monitoring of Electric Machines and Systems
Broken Rotor Bar Detection in Induction Motors using Digital Taylor-Fourier Transform

ABSTRACT. Bar damage is one of the most frequent faults in induction machines. The bar can be partially or fully broken, and the damage can appear in more than one bar. This type of damage may cause adverse effects as temperature rise, mechanical stress, frequency variation, increase in electricity consumption, and increase in motor vibrations, among others. Therefore, to schedule maintenance operations and accelerate repair processes, an opportune detection and classification of faults are imperative. This goes in concordance with the philosophy of electrical systems in the world, which consists of guiding them towards the concept of intelligent systems based on algorithms to track the dynamics of the systems accurately. This paper focuses on a motor current signature analysis through the Digital Taylor-Fourier transform, aiming to apply the digital Taylor-Fourier filters in the spurious frequencies, with the final purpose to reconstruct the filtered signal to obtain its frequency spectrum and, through statistical methods, identify in a precise way the type of bar damage. The proposed methodology is conducted in Matlab and evaluated for a group of data corresponding to a motor with one broken bar under 3/4 and 1/2 load conditions.

Selective Signal Extraction based on OMP algorithm and DCT and DST Dictionaries

ABSTRACT. Filtering and signal extraction are essential applications in some areas such as medical, electrical, pattern recognition, and other disciplines. For example, in regions of automatic detection, it is necessary to isolate specific signals to study them or accomplish the training process to reach a high detection rate. This paper proposes a new technique of signal extraction based on sparse representation and the Orthogonal Matching Pursuit algorithm, using discrete sine transform and discrete cosine transform to generate the dictionaries. Then, the signal is represented in the base of the given dictionary and limited to the desired bandwidth, extracting the desired information with almost no changes in phase and amplitude, reaching almost vertical slopes in lateral bands.

Analysis of the Optimal Decomposition Level Based on Discrete Wavelet Transform for Detection of Power Quality Disturbances

ABSTRACT. The improvement in Power Quality has been a concern for both the public and private sectors in recent years. One of the main problems within this topic is the appearance of anomalies or disturbances in the power supply, which represent sudden changes in the waveforms of the signals and cause severe damage to the utility grid. This paper presents a comparative study of different resolution levels for the analysis of eight types of Single Power Quality Disturbances using Multiresolution Analysis. A set of disturbances was generated in MATLAB through their mathematical models and the extracted wavelet-based features were normalized by Z-score. The results show that the use of nine resolution levels leads to an optimal decomposition, since it allows obtaining a greater amount of information, without compromising the computational performance, which would facilitate a future classification process.

Detection and Classification of Faults in Transmission Lines Compensated with TCSC in the Time Domain

ABSTRACT. Thyristor controlled series compensation (TCSC) is widely used in long transmission lines to mainly improve power transfer capability. However, TCSC produces complicated impedance that negatively affects distance protection operation. This paper presents a mathematical model in the time domain to achieve the detection of the fault in a compensated line instead of an analysis in the frequency domain. In addition, a PNN classifier is used to classify the different types of faults, with different values in the resistance of the fault and different inception angle. The proposed algorithm is capable to detect the moment in which the failure occurs in the signals of the voltage lines. The PNN was trained using 317 and tested using 79 results under different conditions obtaining 72 true positives and 7 false positives in the classification, therefore the relative error was as 0.0886, the classifier has a precision of 0.9113.

Using Partial Least Square for Identification of Powdery Mildew in Cucurbits Plants

ABSTRACT. This study establishes a methodology to identify some defined powdery mildew infection levels in cucurbits plants based on the Partial Least Square Regression (PLSR) coefficients extracted from their spectral signatures. Here, the PLRS decreases the amount of data to be processed. Thus, some conditions of the leaves could be classified by using the PLSR coefficients and Support Vector Machine (SVM). For this case, different parameters in the plants and the growing season are the response concerning variables. To classify four stages of plant: i) healthy leaves, ii) leaves in a stage of germination of the fungus, iii) leaves with first symptoms and iv) diseased leaves. The powdery mildew levels were identified with an accuracy of 92% and a kappa value of 0.81. This analysis propose a feature extraction that could be applied to other plants analysis.

10:10-11:50 Session 2E: Biomedical Applications
Recurrence Quantification Analysis in Accelerometric Signals of the Rectus Femoris during Quiet Stance

ABSTRACT. Postural control involves the activity of three physiological systems, visual, vestibular, and proprioceptive. Galvanic vestibular stimulation (GVS) is a noninvasive method that activates the vestibular system and influences its related physiological pathways. The aim of this study was to determine the effect of GVS (perturbation) on the human balance system. We used accelerometric sensors to record postural oscillations before and during bilateral bipolar GVS. Accelerometry was recorded from 9 healthy subjects that stood for 60 seconds in two conditions: without and with GVS. Stimulation current ranged 0.1 – 1 mA. The accelerometric sensors were located bilaterally over the rectus femoris, a postural muscle. Accelerometric signals were analyzed with parametric statistics and with recurrence quantification analysis (QRA). Statistical analysis showed no significant differences between experimental conditions. However, qualitative and quantitative RQA showed a quasi-periodic-chaotic system behavior. RQA proved to be a valuable tool to describe postural performance according to sensory input.

Skin color detection by digital image processing to compensate deviations in a non-invasive blood glucose estimation

ABSTRACT. Diabetes is a chronic disease characterized by abnormal levels of glucose concentration in the blood. To reduce the risk of medical complications associated with inadequate control of diabetes, continuous monitoring of blood glucose levels is necessary. Near-infrared spectroscopy (NIRS) is a non-invasive technique based on optical methods, so it is a more comfortable, painless, and prick-free method than conventional measuring and reduces the risk of infection in the patient. However, NIRS, like other optical methods, presents an error in the estimation due to the differences between the physical and functional parameters of the skin and tissues of each subject and its interaction with light. This work focuses on the automatic determination of skin tone, one of the factors that interfere with glucose measurement. Deviations due to skin tone parameter in NIRS-based optical blood glucose measurement can be compensated for through image processing.

Regions detection in cervical cytology to help the diagnosis of cervical cancer

ABSTRACT. Abstract—Cervical cancer is an important cause of mortality in women worldwide, it is the neoplasm with the greatest potential proven secondary prevention. This disease is completely preventable and curable. Many computer vision tools are being exploited in the biomedical field to discover patterns and analyze the morphology of cervical cancer among women. The main objective of this work is to show the detection of cells in a cervical cytology because the cytological screening analysis is based on the experience of the cytotechnologist. However, in this document the Gaussian Laplacian is used for the extraction of stains within a cytological image, converts the image to binary and applies a threshold range, which generates stains with properties that remain approximately constant.

Methodology for Characterization and Calibration of Linear Infusion Electromedical Devices

ABSTRACT. This paper presents an application of standard NEN-EN-IEC 60601-2-24 to characterize an electro-medical infusion device for future insulin dispensing. The proposed device is composed of a linear actuator attached to a reservoir and a cannula. The characterization methodology includes the use of a scale with four decimals of precision to weight each drop that can be dispensed from the device. Accurate device characterization allows precise product dispensing.

Support tool for presumptive diagnosis of Glaucoma using fundus image processing and artificial intelligence implementation

ABSTRACT. Blindness is a global health problem and glaucoma is one of the diseases that is considered of vital importance to treat since it is a neurodegenerative disease that causes irreversible blindness that still has no cure, however, it can be treated if detected early; Most people begin to feel symptoms when this disease is already in an advanced stage, therefore in this work we have developed a tool to support the medical diagnosis through digital image processing for it has been consulted in several databases for the study and classification of retinal images among these images we have healthy eyes, suspected glaucoma and diagnosed with glaucoma. The region of interest we worked with was the optic disc since this is where the blood vessels are interconnected and it is an important area for analysis.

12:10-13:50 Session 3A: Electrical
Analytical Formula of the Mutual Impedance of a Partial Core Transformer with Conductive Tapes

ABSTRACT. The present work shows a first approximation of the analytical model of the mutual impedance of a transformer with a partial cylindrical core considering the effective tensor values of magnetic permeability and electrical conductivity. This first model is based on the use of constant values to represent the effective relative permeability of the core material of this transformer. The analytical expressions obtained allow modeling the mutual impedance values for the following cases: filamentary conductor, conductors with a rectangular cross-section and for a two-coil device.

Fast steady-state solution of electric systems under harmonic distortion conditions based on sparse matrix LU decomposition

ABSTRACT. This paper presents the application of companion-circuit analysis (CCA) to the periodic steady-state solution of small, medium and large-scale electric systems under conditions of harmonic distortion in the time-domain (TD), whose model is obtained by numerical integration using the trapezoidal rule (TR) method, i.e. CCA-TR. This method allows obtaining a model based on a matrix relationship that consists mainly of a symmetric and sparse conductance matrix in the case of medium and/or large-scale electric systems. The solution of the CCA-TR method is achieved through a factorization process based on the sparse matrix LU decomposition. The combined CCA-TR-LU method is applied to the analysis of the modified IEEE test systems under harmonic distortion conditions. In particular, the performance of the sparse CCA-TR-LU method is given in terms of CPU time and accuracy. The reported results are validated against the response obtained with the the PSCAD/EMTDC simulator.

Design of shell-type single-fase distribution transformers using genetics algorithms

ABSTRACT. In the present work, the optimal design of distribution transformers using genetic algorithms is proposed. A program was developed using this method to obtain a design with the lowest evaluated cost for distribution transformers, with a methodology that determines for their construction all the parameters of the magnetic and electrical elements. The optimization is carried out with the heuristic variant of genetic algorithms and shows how the reduction of the evaluated cost is guaranteed, satisfyingly with the design restrictions (percentage of impedance, total losses, core losses, efficiency, percentage of excitation current) according to the standards. The case being analyzed is a single-phase distribution transformer of 25 kV A, 13600 Y/7620-240/120 V. For the implementation of the design and its corresponding optimization applying genetic algorithms, the free software Python 3.6 was used.

Comparative study of DC short-circuit in MMC-HVDC system considering FCL

ABSTRACT. One of the main challenges of high voltage direct current (HVDC) systems is the detection and isolation of DC short circuit faults, which present a pronounced dc current rise rate; these faults must be detected and isolated within a few milliseconds of their occurrence to prevent the collapse of the HVDC system. This is crucial to give the reliability that one of these systems must have. This article analyzes and describes three types of short-circuit faults located at different points in the HVDC system to describe the behavior of the short circuit phenomenon, as well as the influence of the FCL in short circuit conditions. The case studies are evaluated in the PSCAD/EMTDC electromagnetic transient analysis tool.

Bio-inspired Optimization Strategy for Identify Coherent Nonlinear Generators Response

ABSTRACT. Coherency identification is one of the key steps to carry out different control system strategies to avoid a partial or complete blackout of a power system. However, the oscillatory trends, and the non-linear dynamic behaviour of the system measurements in the first seconds of the transient period often mislead the appropriate knowledge of the actual coherent groups, making wide-area coherency monitoring a challenging task. Inspired by swarm behavior in nature, we propose a modified Particle Swarm Optimization (PSO) approach based on centroids codification to identify coherency within a short observation window. A user defined inter-group function is proposed and compare its results against conventional intra-group functions to examine the quality of the compacting and separating features of the final clusters. We demonstrate the effectiveness of our technique to identify coherent generator groups while exhibiting robustness to clustering the measurement that experiment a higly nonlinear dynamic changes into the first seconds on the 39-bus New England system response.

12:10-13:50 Session 3B: Vision and Language / Data Science / Computer
Linear control of the non-linear model of the convey-crane

ABSTRACT. By using the Euler-Lagrange method, the non-linear mathematical model of the convey-crane is obtained. Using a tangential linearization, around the equilibrium point, the linear mathematical model is obtained in state space and transfer function forms. Using classical methods, a single-input singleoutput linear controller is designed for the output y = x + Lθ (translational projection of the pendulum position). Finally, the behavior of the nonlinear system, controlled through the linear controller, is simulated and satisfactory results are obtained.

A data analytics method based on data science and machine learning for bank risk prediction in credit applications for financial institutions

ABSTRACT. Nowadays, banks grant credits so that customers can acquire a good or service, start or improve a business, among other benefits. The problems that may arise are over-indebtedness and low saving possibilities on the part of customers, so the tendency is the risk of default. Financial institutions require tools for default risk analysis and problem prediction. Therefore, in this research, a data analysis method based on data science and machine learning is proposed for bank risk prediction in credit applications for financial institutions. For the analysis process and for the prediction of a credit, predictive analysis methods are used: Genetic Algorithms (GA), Random Forest (RF), K-Nearest-Neighbor (KNN), Support Vector Machines (SVM) and Neural Network (NN). Quality metrics such as Accuracy, Precision, Recall and F1 Score are used to evaluate the results. A public dataset called Statlog is used. This work opens the door for data analysis in different banking services. The main objective of this research is to help financial companies to optimize their processes.

Sampling algorithms for Unsupervised Prototype Selection

ABSTRACT. The $K$ Nearest Neighbor classifier (K-NN) is one of the most widely used classification algorithms. It is simple, non-parametric, and does not have a training phase, but it requires storing all observations (training set). The class for an unknown sample $x_q$ is labeled using a similarity/distance function d to compute the $K$ most similar elements (neighbors). The latter makes that K-NN prediction process unfeasible for large datasets. Two main approaches are studied to optimize the prediction process: a pre-computed index to speed up the neighbors' computation and by generating of a subset of representative elements called prototypes. This work presents an analysis of sampling algorithms to perform unsupervised prototype selection. Each proposed algorithm's performance is measured by measuring the capacity of copying the original data set class distribution and the performance for a classification task.

Towards the Development of an Acne-Scar Risk Assessment Tool Using Deep Learning

ABSTRACT. Early estimation of the risk of having acne-induced scars is crucial in acne sufferers to ensure appropriate treatment and prevention. This paper explores the feasibility of using Convolutional neural networks (CNN) to estimate the risk of developing acne-induced scars based only on image analysis as a complementary tool for diagnosis. A database of acne sufferers whose acne-induced scar risks have been evaluated by specialized dermatologists applying a four item-Acne-Scar Risk Assessment Tool (4-ASRAT) was used. The training dataset includes images of patients with a low, moderate, and high risk of suffering from acne scarring. The dataset was used to train a CNN model based on the VGG16 architecture for the binary and triple classification problem. Poor performance was achieved for the threefold classification problem, while the best model for the binary classification problem achieved an accuracy value of 93.15% and a loss of 19.45% with 0.931 AUC. Although these initial results are promising, much work is still required to improve the performance of the model.

A low-cost stereo vision system for eye-to-hand calibration

ABSTRACT. In automated systems, there are tasks such as object pick and place. In this task, a vision system detects the coordinates of the work object. The robot uses these coordinates, link lengths, and joint values to implement inverse kinematics to move the robot to a point. However, the vision system obtains work object coordinates in its reference system, but the robot needs to move to a point in its reference system.Changing the camera coordinate system to the robot coordinate system is necessary. We propose a crucial task methodology to compute this rigid transformation. This document proposes a method where a stereo vision system obtains the 3D coordinates of the center of a sphere in the robot end-effector. This way, we have two sets of points with different reference systems. We can find the transformation between robot and vision system references finding the rigid transformation that reduces the Euclidean distance between the two sets of points. Because the real length of the link has an error derived in the manufacturing and assembly process is necessary to perform a robot calibration. The error obtained from adjusting both sets of points in the first experiment was 3.3136 mm, and after geometrical parameters compensation, this error was reduced to 1.9927 mm. It means a reduction of 39.86%.

12:10-13:50 Session 3C: Power Converters Applied to Renewable Energy Systems
Multi-Variable Fuzzy+PID Control of a Buck-Boost Four Port Converter for Renewable Energy System

ABSTRACT. This work presents the design and control of a Buck-Boost Four Port DC-DC Converter (BBFPC) using a mixed PID + Fuzzy Controller for power flow control. The proposed system was designed considering the solar irradiation profile in Cumbay\'a-Ecuador. The system manages two solar panel arrays and a battery energy storage system (BESS). The system feeds an isolated load. Pulse width modulation plus phase shift modulation are used to control the converter. Results and proofs of concept are presented through simulation.

Small-Signal Models of Transmission Lines Oriented to the Impedance-Based Method

ABSTRACT. This research work proposes an exact procedure to compute the impedance models of networks with power electron- ics penetration and the full frequency-dependent transmission lines (TL). It is described how to interface large-signal average models of converters in DQ-domain with TL models in the abc-domain to compute the exactly corresponding steady-state solution with the balance method. Subsequently, it is developed a form to use the relationships of voltage-current on the sending side of the TL to compute the impedance in DQ-domain using its exact hyperbolic solution and the Park transform. This alternative does not rely neither on rational fitting or repetitive time domain simulations. The effectiveness and accuracy of the proposal are validated through simulations in PSCAD-EMTDC.

Design Assessment of a High-Reduction Ratio, Doubly Interleaved Buck Converter

ABSTRACT. Abstract—In recent years, Electric Vehicles (EVs) have become a necessity more than a luxury, justified by the urgency of transitioning to a sustainable movability, being DC-DC and DC-AC converters the heart of this technology. Interleaved Buck DC-DC Converters are commonly used as energy buffers in this application, being high-power density circuits a critical aim in this type of converters. This paper presents a preliminary design assessment of an interleaved buck DC-DC converter with high reduction ratio, that uses a double magnetic together with a coupling capacitor to increase the reduction conversion ratio and power density. A brief survey around the steady-state principle of operation of this DC-DC circuit is presented together with a description of the initial setup towards the building of a 4 kW prototype.

Grid-Feeding Power Converter in an AC Microgrid with a Photovoltaic Array and a Battery Pack

ABSTRACT. In this paper, a grid-feeding power converter topology is proposed so that a photovoltaic array and the mains, if necessary, can charge a battery pack. The proposal presented also considers extracting the maximum power of the photovoltaic array. The grid-feeding power converter topology is composed by a boost DC-DC converter and a bidirectional DC-AC converter connected to the mains, the control strategies for each converter are also presented. The control scheme for the boost DC-DC converter includes a maximum power point tracking algorithm. The control scheme for the bidirectional DC-AC converter has two operating modes, current-control mode and voltage-control mode. To validate the performance of the grid-feeding converter simulation results are presented. The grid-feeding converter control strategy successfully carries out the power management between the mains, the PV array and the battery.

Grid-Forming Type Converter Decoupling and Resynchronization for an AC Microgrid

ABSTRACT. The Microgrid (MG) operation modes are Grid-Tied Mode (GTM) and Islanded Mode (ISM). However, the transitions from GTM to ISM (GTM-ISM) and from ISM to GTM (ISM-GTM) are currently studied as operation modes. To correctly perform GTM-ISM and ISM-GTM, in this paper an Islanding Detection Algorithm (IDA) and Grid-Synchronization Algorithm (GSA) are developed. On the one hand, IDA is developed based on the DQ transform and the voltage and frequency limits defined by IEEE 1547 Standard. On the other hand, GSA is developed employing three Phase-Locked Loops (PLL) and the DQ transform. Both IDA and GSA operate together as a secondary control level to provide synchronizing signals to the primary control level. The results show a correct IDA and GSA performance, which allowed the GFTC to be capable of carrying out both GTM-ISM and ISM-GTM, correctly. This is possible because of the cascade control schemes outstandingly perform tasks such as: to extract mains power, to charge the battery, to regulate the DC bus voltage, and to synchronize and to regulate the GFTC output voltage.

12:10-13:50 Session 3D: Electronics
A Dynamic Sliding Mode Controller Approach for Open-Loop Unstable Systems

ABSTRACT. This document presents a Dynamic Sliding Mode Controller for open-loop unstable processes with time delay combining the sliding mode concepts and internal mode structures. The proposal aims to improve performance concerning sliding mode control, both in tracking and regulation, and to reduce the chattering effect on the control signal. An unstable system. An empirical method approximates an unstable third-order system by a First Order Plus Dead Time (FOPDT) model. Several set points and disturbances are used for testing the proposed approach.

Noise Shaping SAR ADCs: Fundamentals, challenges, trends and possibilities

ABSTRACT. The aim of this contribution is to serve as an introduction and overview for readers who want to initiate into the study of Noise Shaping type Successive Approximation Converters (SAR), reviewing a general description of the fundamentals of Noise Shaping SAR (NS-SAR). This paper addresses the difficulties (i.e. open problems), analyzes the fundamental challenges and summarizes the latest developments, including the main problems, which are primarily in the loop filter and its passive/active implementations, mismatch present in the DAC elements, among others. Some possible solutions for the dynamic correction of CDAC mismatch are also presented. Finally, some trends and future challenges are discussed.

Analysis of Predictive Control Strategies for High-Speed Operation of PMSM Drive

ABSTRACT. This work presents an analysis of predictive control strategies for a permanent magnet synchronous machine and designed initially at the nominal operating regimen. The motivation of this work is based on applications where high-speed requirement and limited supply voltage are needed such as powered by battery-pack. Three control strategies are analyzed and compared based on the following characteristics: Model-Based Predictive Control, avoided weight factor, and fixed frequency operation. The main contribution of this work yields in the behavior verification of the mentioned control strategies and the availability determination to work along to the field-weakening approach. Moreover, a combination of two strategies is presented generating a new contribution; this combination reduces the operating time interval while reduces the computational burden. The effectiveness of the analysis is experimentally validated.

12:10-13:50 Session 3E: Advanced Signal Processing Techniques for Condition Monitoring of Electric Machines and Systems
Real-time Harmonic Identification for unknown Frequency and Amplitude parameters of a Distorted AC signal

ABSTRACT. This paper proposes a novel method for the real-time estimation of the unknown parameters of time-varying (TV) frequency that can occur in distorted AC signals, as well as the amplitude of the harmonic content, through a sliding modes-based observer, which is a robust and fast converging estimator. The methodology can even be used for determining the signal harmonics of a non-periodical signal, where the amplitudes of the harmonics are time-varying. A numerical performance evaluation of the estimation scheme is developed to determine the referred parameters of a signal containing four harmonics. The real-time estimation methodology could be used for power quality monitoring, harmonic compensation in fast dynamic nonlinear loads, or frequency regulation in electrical

DSOGI-PLL with FLL to Mitigate Transient Oscillations in the Frequency and Phase Angle Measurement

ABSTRACT. One of the most used approaches to synchronize power electronic converters to electrical power systems is the phase-locked loop (PLL). PLLs with adaptive characteristics such as the PLL based on a double second order generalized integrator (DSOGI-PLL) are widely used in the synchronization of three-phase power converters, due to their adaptive filtering characteristics and the improvement that is obtained by adding a sequence component calculator before the loop filter. However, the accuracy of the PLL measurements, and therefore the calculation of the phase angle, is affected during transients, so that the frequency, phase angle, and voltage calculated by the PLL are a function of the disturbance type and the architecture of the PLL. This article proposes an improved DSOGI-PLL with a frequency-lock loop based on a double generalized second order integrator (DSOGI-FLL) which is more stable and shows a faster response than the conventional DSOGI-PLL. The main idea is to combine the strengths of these two algorithms into one architecture that will be called DSOGI-PLL with FLL. The performance of the proposed PLL is verified and validated under the processor in the loop (PIL) approach.

Incipient Damage Detection in a Truss-Type Bridge using vibration responses and MUSIC Technique

ABSTRACT. Nowadays, with the new developments in both the signal processing and the signal acquisition technology fields, it has been possible to design and develop novel sensing technologies that can detect and process low-amplitude vibration signals, allowing quantifying the generated changes to determine its mechanical condition. In this sense, one method that has proved to be effective for detecting subtle changes in the signals is the multiple signal classification method (MUSIC), since it considers that a vibration signal is formed by complex components embedded in a noisy signal. Since real-life signals fulfill the aforementioned concept, this paper explores the efficacy of the MUSIC algorithm for detecting early-stage damages in a 9-bay truss type steel bridge. The obtained results show that the proposal is effective for detecting early-stage damages of corrosion.

Raspberry-based low-resolution thermal image system using a smoothing filter-based Kalman

ABSTRACT. Since the emergence of global epidemics such as SARS-CoV-2, AH1N1, SARS and MERS, a wide range of systems for measuring temperature have been developed based on computer vision to reduce and prevent the virus contagious. By implementing a Raspberry-based low-resolution embedded system based and a FLIR® Lepton® sensor human body temperature is measured and improved by four different algorithms implemented. Firstly, three traditional time-series processes solving such as, Simple Mean (SM), Simple Moving Average (SMA), and Multi Lineal Regression (MLR), and secondly, and online filter-based Kalman predictor were implemented to increase the signal to noise ratio of the acquired temperature magnitude. Results of average prediction for different benchmarks demonstrate the best performance of Kalman Filter upon traditional processes. In addition, this algorithm achieves to smooth output temperature with fewer samples (~10% of samples) in comparison MLR and SMA. Finally, Raspberry-based low-resolution thermal image system is a feasible tool as a high-speed temperature estimator, by implementation of algorithms codified in Python programming language.

Modeling and parameter identification of an equivalent electrical circuit for corrosion-sensitive lead-acid battery

ABSTRACT. This paper presents the modelling of the corrosion phenomenon in a lead-acid battery via an equivalent electrical circuit (EEC). The parameter identification is carried out by a recursive least squares (RLS) algorithm with adaptive forgetting factor. In order to achieve the analysis and validation of the model, this was programmed in MATLAB to solve it numerically and to be able to simulate different current profiles. The validation of the model was proved by using voltage and current measurements in the PS-260 battery terminals, what was submitted to a charge/discharge profile. Modelling internal faults by EEC techniques is of interest to develop fault diagnosis schemes in a non-invasive way online for battery banks.

12:10-13:50 Session 3F: Renewable Energy
Economic dispatch of microgrids considering CO2 emissions reduction and BESS units integration

ABSTRACT. The integration of energy storage systems offers an important alternative to improve the reliability of the electricity grids when a large amount of renewable energy penetration is presented, as occurs in microgrids. This technology exhibits flexibility in its operation and can be used as power support when intermittent occurs. However, the storage energy is finite and must be used in the best way to improve the impact over the operation of electrical grid. Thus, this article presents an evaluation of a model to define a strategy for economic dispatch of microgrids considering penalties for the use of generation based on fossil fuels, as well as incentives to mitigate the emission of polluting gases. A demand response program is used as a bonus scheme to the consumers for making a decrease in their consumption during contingency periods. Renewable generation is considered within the day-ahead dispatch under the consideration that it is possible to determine these generation profiles. The dispatch strategy is evaluated in a system with the integration of renewable energy sources, battery energy storage systems under different wind and irradiance profiles.

Potential Impacts of the Reform to the Electricity Industry Law in the Energy Dispatch of the Mexican Wholesale Electricity Market

ABSTRACT. This paper shows the potential impacts that the modifications made to the Electricity Industry Law (LIE) would have on the economic dispatch, depending on alternative ways to interpret those changes. Using a simple dispatch model, an attempt is made to capture the essential characteristics of the generation dispatch carried out by the Market Operator in Mexico to show the impacts that the adoption of the changes established in the modification to the LIE would have on the dispatch mechanism.

Virtual Inertia Control applied to a Multi-Area Electrical Power System with Low Inertia and High Penetration of Renewable Energy Sources.

ABSTRACT. The integration of renewable energy sources in modern electrical power systems is a challenge for the operation, control, and protection of an electrical power system. This is mainly because renewable energy sources, when interconnected to the network through electronic devices, do not provide inertia to the system which in turn may cause frequency stability issues and cause system crash. To solve these problems, this work proposes to emulate the behavior of a synchronous machine and provide frequency support through virtual inertia control. In addition, the derivative method connected to an energy storage source is analyzed to provide frequency support to a multi-area control network with a high integration of renewable energy sources and to analyze the dynamic behavior of the system with load variations.

Optimal charging schedules for PEVs with voltage profile support in distribution grids

ABSTRACT. This paper presents a centralized approach for optimal charging schedules when several plug-in electric vehicles (PEV) are connected to the power distribution grid. Thus, the the proposed approach considers the constraints on PEV chargers, the limits on the state of charge (SoC), and the voltage levels of the grid to minimize the energy cost used to charge PEVs while providing a voltage support service to the grid. So, to incorporate the voltage levels, an approximated linear model for low voltage distribution systems based on the load profiles are employed. To confirm the effectiveness and performance of the proposed centralized approach for optimal charging schedules, it is tested by using the IEEE 33-bus radial network with a dynamic tariff system (ToU)

Theoretical and experimental investigation of a Faraday disc generator

ABSTRACT. This work deals the case of a conductive-solid-material in a hollow-cylinder shape that rotates at different speeds in a constant magnetic field, where a potential difference is measured at the inner and outer radius’ disc. This Faraday’s disc experiment is considered to be modeled as the ideal case of a conductive fluid flowing inside a cylindrical space whose viscosity is finite, and a constant magnetic field provided by permanent magnets cut across the rotating conductive material. Similar to the Faraday disc[1], which it can also be modeled as the ideal case of a flow of finite viscosity, and conductor fluid flowing in a cylindrical space in the presence of a magnetic field, so it is approached using a case reported in the literature [2]. It is important to understand the phenomena of electromagnetic induction for small-scale direct current generators, understanding and modeling the analytical and experimental cases that represent the phenomenon is of great importance since the use of generators on this scale are used to collect the kinetic energy of waste, (caused for example by human movement) which would allow to replace to a certain extent the use of batteries that are currently used in applications where certain body variables such as pulse, temperature, etc., are monitored. Thus in this work, the most basic analytical (taken from ideal case from [2]) and experimental development with their results and comparison are provided.

14:10-15:10 Session 4: Keynote Lecture
A family of contrast-pattern based classifiers for class-imbalance problems