Electrical and Electronics Engineering Institute

University of the Philippines - Diliman

Thesis/Dissertation Presentations

The graduate programs of the Institute provide advanced training and specialization in a broad range of areas in electrical and electronics engineering and its allied fields to prepare students to solve complex technological problems and to contribute new knowledge to the field. As a requirement for graduation, our students present and defend their respective thesis/dissertation.

Below is a list of the defense schedule of our students for the Midyear Term AY 2019-2020.

Date and Time

Venue Name Title Abstract

10 Jul 2020 (9:00AM)

will be

streamed

 online

Aldrich Zeno

PhD Dissertation: Differentiating Power Swing from Fault Using Rate of Change of Impedance Angle

(recorded presentation)

Power swing causes swinging of impedances thus leading the distance relay to maloperate due to momentary entering of impedance into tripping region. This research used time domain simulations in Matlab®/Simulink and RTDS® to show that power swing produces a much slower impedance trajectory than faults. This led to: (i) Deriving generic transformation from time domain voltages and currents in terms of resistance, reactance, impedance angle and rate of change of impedance angle using analytical approach; (ii) Proposing a scheme which improved the detection of faults, stable power swing and unstable power swing by redefining the operating/restraining regions using impedance angle and rate of change of impedance angle in θ-dθ/dt plane; and (iii) Usage of electronic hardware circuits which verified the proposed scheme, and PMU which measured the impedance angles and rate of change of impedance angles from RTDS during normal operation, faults, stable and unstable swings. 

The scope of this research was extended to renewable energy source integration to analyse the effect of VRE penetration on the rate of change of impedance angle. Results showed that impedance angle and rate of change of impedance angle had no variations in maximum values whereas, large changes were observed in resistance and reactance for different levels of wind energy penetration.  With comprehensive mathematical expressions (analytical models), software simulations (Matlab® and RTDS®) and hardware setup (PMU measurements from RTDS signals and operational amplifier-based circuit), it was shown that faults and unstable swings produce large rate of change of impedance angles compared to normal operation and stable swings. This procedure can contribute in the development of an impedance angle-based relay. This relay will effectively identify and trip for faults and unstable swing while avoiding incorrect trip for normal operation and stable swings. 

23 Jul 2020 (3:00PM)

will be

streamed

online

Darvy P. Ong

MS Thesis: Student Risk Assessment: Identifying Contributing Factors and Predicting Success in Graduation of Undergraduate Students in the College of Engineering

(recorded presentation)

From the academic year 2009 to 2013, 60.27% of the freshmen from the College of Engineering at University of the Philippines Diliman graduated within the same college. Hence, there is a need to understand the cause for the College’s low graduation rates. Existing studies on student graduation rates typically use statistical analysis and machine learning methods to correlate a student’s profile and their chances of graduation. Building on the success of these methods for other institutions, we used Logistic Regression, Support Vector Machines and Neural Networks to evaluate the contributing factors that may affect student graduation chances. The results show that the main factors include enrolment in the preferred degree, Math scores in the college admission test, high school academic performance, proximity to the university, and economic background. While all three models are good at predicting graduation outcome, the Neural Networks yielded consistently high scores in classification accuracy (75.93% to 80.83%) and class separation (ROC-AUC score between 75.44% and 84.63%). Hence, the model can be used to perform student risk assessment and develop plans to increase the chance of graduation.

7 Aug 2020 (2:00PM)

will be

streamed

online

Juan Miguel M. Consolacion

MS Thesis: A Study on the Weighted Least Squares and Weighted Least Absolute Value Methods of Branch Current-based Distribution System State Estimation

(recorded presentation)

Power system state estimation is a data processing algorithm for converting redundant meter readings and other available information into an estimate of the static-state vector. Most distribution system state estimators in literature use WLS BCDSE because of the efficiency of WLS and the branch current-based approach. However, it is well established that WLS is not robust against bad data. It is important to consider the presence of bad data because they exist in any information system. The WLAV approach, on the other hand, has been shown to reject bad data but its computational inefficiency hinders it from being adopted in power systems. WLAV BCDSE, a novel approach to BCDSE, is proposed. The author hypothesizes that using BCDSE and PMUs may improve the efficiency of WLAV, allowing it to be feasible for real-world applications. 

The tests conducted show that WLAV BCDSE is not as robust as WLS BCDSE insofar as consistent convergence using different measurement types is concerned. Unlike WLS, WLAV was inconsistent when SCADA measurements were used, which may partly explain why WLAV BCDSE has not been explored in literature. When PMUs were used, WLAV BCDSE converged consistently, especially when voltage measurements were not present. Aside from convergence, WLS and WLAV BCDSE reacted similarly to different measurement types. In summary, current phasor measurements must be prioritized to maximize the efficiency and accuracy of WLS and WLAV BCDSE. Power flow measurements also help in improving the performance of the estimators when SCADA current measurements were used. Lastly, voltage measurements, both PMU and SCADA, degrade the efficiency of BCDSE while providing a minimal increase in accuracy, at best.

WLAV BCDSE was shown to be able to detect and reject bad data when appropriate measurement weights were used. Applying to WLAV the same assignment of weight as WLS severely limits the former's capability to properly handle bad data. When the measurements of a single device were replaced with bad data, WLAV was able to limit the maximum voltage total vector error to 0.02 Vpu, whereas the maximum error of WLS reached up to 0.5 Vpu. WLAV was also able to produce estimates with acceptable accuracy even when three out of eight of the available devices contained bad data.

28 Aug 2020 (2:00PM)

will be

streamed

online

Jean Marriz M. Manzano

MS Thesis: Investigation of Surface Roughness Effects on Piezoelectric Cantilever MEMS-Based Energy Harvesters

(recorded presentation)

Recent IOT devices include energy harvesters to make the system autonomous, self-sustaining, and maintenance-free. However, the fabrication steps that these harvesters go through may result in surface irregularities that may affect its harvesting efficiency. Recent models incorporating surface irregularities focus on other types of MEMS devices but not on energy harvesters. This thesis investigates the effect of surface roughness on the characteristics of piezoelectric MEMS energy harvesters by creating an analytical model which incorporates realistic roughness data from actual micro-fabricated samples. To verify the analytical model developed, 3D multiphysics simulations of cantilevers with surface roughness were also set-up. The surface roughness of the fabricated samples are measured using a high-resolution 3D laser microscope. The resonant frequencies of both the developed analytical model and multiphysics simulations show good agreement having around 1% error. Analysis shows that an approximately 1.5μm Sq roughness causes frequency deviation of about 5.53% or 308.31Hz from the ideal frequency of the beam with dimensions 5120 x 1280 x 100μm3. The center region of the wafer shows the lowest percent deviation of frequency among all other regions where parameters Ssk and Sku values were found to be closer to a normally distributed roughness data. This study gives insight into the importance of considering fabrication related irregularities such as surface roughness in the design of MEMS structures, particularly in energy harvesters. The results of this study can be utilized to adjust the energy harvester's design once critical fabrication steps have been characterized, thereby preventing excessive fabrication retrials and saving time, effort, and physical resources.

Below is a list of the defense schedule of our students for the Second Semester AY 2019-2020.

Date and Time Venue Name Title Abstract

22 Jan 2020 (2:00PM)

EEEI

ULYS3ES

3rd floor

Keziah B. Bartilad

MS Thesis: Design and Characterization of Antennas with Photovoltaic Cells as Antenna Radiating Elements

 

With the widespread use of solar cells and solar panel modules, it is advantageous to integrate antennas with them for added functionality and to maximize resources. In integrating, the solar cell can either be a radiating or a non-radiating antenna element. To avoid shading and power loss in the solar cell, and for more flexibility in antenna design, the former is chosen. In this paper, we design and characterize a patch antenna integrated with solar cells and solar module layers. The effect of the properties of the solar cell and solar module layer on the antenna performance is investigated by simulating the integrated layers and the patch antenna. The values for the typical properties of solar cells and module layers are used. In integrating layers of the solar cell, the silicon semiconductor, as the thickest layer, had the most effect on antenna performance. The rear contact has negligible effect while the front metal lattice improves antenna performance, particularly return loss and gain with increased area coverage. Overall, only slight differences in antenna performance were seen. In integrating solar module layers to the patch antenna, significant shift in resonant frequency, deterioration in return loss performance, and increase in gain were observed. The effects of the layers, recommendations on material properties, and the trade offs in antenna performance to consider are presented in this paper.

27 May 2020 (2:00PM)

will be

streamed

online

Alexis Czezar C. Torreno

MS Thesis: Power and Area Oriented Implementations of Lightweight Cryptographic Algorithms for FPGA based Wireless Sensor Networks

(recorded presentation)

 

Security is a concern in wireless sensor networks, which are inherently prone to third party attacks. As such, cryptography is used to make a secure mode of communication among nodes and/or between nodes and base stations. However, conventional algorithms are resource-hungry and therefore not fit for small devices, hence the creation of a new category of cryptography known as Lightweight Cryptographic Algorithms. These algorithms are still continuously being improved to fit in the decreasing sizes of small scale devices like FPGA-based wireless sensor networks. Different optimizations have been used to improve area and power efficiency of well known ciphers. However, these ciphers have a limit on how much they can be improved. 

In this study, we quantify the effects of Round Unrolling, and Data Width Reduction on area, and power consumption. These are tested on three candidate ciphers: LiCi, ANUII, and QTL. Round Unrolling reduces cipher latency which lessens the contribution of static power for each computation. Results show that round unrolling improves power efficiency by 25.97%, 3%, and 14% in LiCi, ANUII, and QTL, respectively. This comes at 299%, 414%, and 52% increase in area. Data Width Reduction improves area by only using a fraction of the original datapath. Data width reduction was found to be ineffective in reducing area in the candidate ciphers, increasing area by 5.64%, 6%, and 9.2% in LiCi, ANUII, and QTL, respectively. The results allow newer and better lightweight ciphers to be further improved for small scale devices. Round Unrolling can be used for power oriented systems, and while Data Width Reduction did not work on these ciphers, the effect can still be tested on others first before using on area oriented systems.

27 May 2020 (3:00PM)

will be

streamed

online

Ryan Albert G. Antonio

MS Thesis: Post-training Bit-selection Control for Energy-Efficient Hyperdimensional Computing Architecture

(recorded presentation)

Hyperdimensional computing (HDC) is a brain-inspired computing framework that provides simple and convenient methods to perform cognitive tasks like classification. Its foundation lies in the properties of very high dimensional vectors called hypervectors (HV). The first attempts to create an energy-efficient HDC hardware contains massive bit-wise operations. State-of-the-art designs focus on either optimizing computations or developing new devices; however, some of these may directly affect the accuracy performance of the algorithm. 

After a careful investigation of the HDC algorithm, there exist redundant bit locations in the class HVs that do not contribute any significant information during classification. These irrelevant bits can be shut-off to improve its energy efficiency. However, the amount of redundancy is dependent on the cross-similarity of the original data set of a particular application. This thesis presents two major contributions. First are mathematical models relating to the redundancy and the cross-similarity of a given data set. Second, a bit-selection control that is added to the generic hardware designs, which disables the redundant bit-wise operations and improves the overall energy efficiency without sacrificing accuracy. Results show a 10% – 67% energy reduction at the cost of 8% – 20% increase in area and < 10% energy overhead cost. Additionally, mathematical models describing resulting energy savings were also developed. 

4 Jun 2020 (3:00PM)

will be

streamed

online

Daryl L. Peralta

MS Thesis: Next-Best View Policy for 3D Reconstruction

(recorded presentation)

 

Creating 3D models of large structures requires capturing monocular or depth images of the target structure at different viewpoints using aerial drones. The selection of these viewpoints has a significant effect on the quality of the output 3D model. Manually selecting viewpoints or using commonly available flight paths like a circular path often results in insufficient viewpoints and incomplete 3D models. On the other hand, adding more viewpoints results in longer processing time and longer flight path. Recent works have relied on hand-engineered heuristics such as maximizing the information gain to select the Next-Best View (NBV) and acquire an optimal path. In this work, we cast the problem of view planning to a reinforcement learning setting where an agent learns an NBV policy to scan houses optimally by maximizing a reward. We call this learning-based algorithm Scan-RL. To train and evaluate our algorithm, we created Houses3K, a dataset of textured 3D house models. Our experiments show that using Scan-RL, the trained NBV policy can be used to scan houses with fewer number of steps and a shorter distance compared to the baseline circular path. Experimental results using Houses3K demonstrate that a single NBV policy can be used to scan multiple houses including those that were not seen during training.

5 Jun 2020 (9:00AM)

will be

streamed

online

Rangel D. Daroya

 

MS Thesis: REIN: Flexible Mesh Generation from Point Clouds

(recorded presentation)

Efficient 3D object reconstruction is important for several computer vision tasks. Objects in 3D can be digitally represented as a point cloud, an occupancy grid, or a mesh. Lidar sensors often acquire sparse point cloud data. In addition, a point cloud's scattered form and lack of surfaces limits its utility compared to meshes. Occupancy grids are an alternative, but have limited resolution when depicting surfaces and have large memory usage. Meshes have continuous surface information and can represent objects with varying point densities. Existing surface reconstruction methods such as Ball Pivoting Algorithm (BPA) and Poisson Surface Reconstruction (PSR) interpolate from point clouds to produce meshes, but their dependence on point density causes significant performance decline with decreasing number of points. To address surface reconstruction from sparse points, we propose REIN: Recurrent Edge Inference Network. REIN is a neural network that generates meshes from point clouds by sequentially generating edges and faces. The network can produce outputs with information about the general structure of the object. REIN's sequential nature provides continuous feedback on edge generation and results in flexible mesh outputs with varying number of vertices. In this work, we demonstrate the mesh generation improvement of REIN compared to other surface reconstruction methods: BPA and PSR. Experimental results on ShapeNet and ModelNet10 show 81.5% average improvement in Chamfer Distance and 14% average improvement in Point Normal Similarity compared to Ball Pivoting Algorithm and Poisson Surface Reconstruction. Qualitatively, the generated meshes have a closer similarity to the ground truth.

15 Jun 2020

(2:00PM)

will be

streamed

online

Arvin D. Escultero

MS Thesis: Two-Stage LED Driver Based on Buck and LCC Resonant Converter Operating at the Constant Current Frequency 

(recorded presentation)

LEDs are widely used in smart farms because of high efficacy, ease of control and consistent wavelength. In this application, a driver capable of delivering high power constant current with dimming capability is required. To utilize the benefits of LEDs in smart farms, an appropriate driver based on buck and LCC resonant converter was designed, simulated and tested. Resonant converters are growing in popularity in these applications due to their soft switching capability which allows higher frequency of operation without compromising efficiency. When operating at a certain frequency, the LCC functions as a constant current source, but locating the exact frequency of operation to exploit this feature poses a challenge due to component value tolerances. Since resonant converters usually operate at frequencies 100kHz or higher, changes caused by tolerance in the values of the reactive components cause a significant shift from the desired operating frequency. A conventional solution is by using high precision components, but these are more expensive. This driver was designed with a unique feature to track the LCC constant current frequency despite tolerances in the component values. This was done by utilizing resonant frequency tracking techniques and carefully designing the LCC quality factor. This allows constant current output despite the use of low precision components and the absence of a regulating current feedback loop. Dimming capability for the LED output was also available using the buck section PWM which is easily set through the microcontroller. The design procedure was developed and verified with a 120W prototype which was tested on all operating conditions. Results show the driver was capable of delivering constant current, with a dimmable range of 200mA up to 2A, to the LED strings at high efficiency which peaked at 92.88%. The tracking algorithm was used to set the operating frequency which was detected despite the large shift of 19kHz above the nominal calculated value. Output current levels were maintained despite changes in loading conditions brought by changes in temperature from 10oC up to 40oC which proves the driver was operating at the constant current frequency. Lastly, the driver output current was directly proportional to the PWM of the buck section which allows easy controls in setting dimming levels. Overall, this resulted in the design of an efficient open-loop constant current driver with dimming function and creation of an algorithm for tracking the constant current frequency of the LCC resonant converter.