Defenses

Name: Varun Khemani
Title: Prognostics and Secure Health Management of Analog Circuits
Committee Members:
Professor Michael G. Pecht, Chair
Dr. Michael H. Azarian, Co-Chair
Professor Peter Sandborn
Professor Abhijit Dasgupta
Professor Mark Fuge
Professor Pamela Abshire, Dean’s Representative

Date: Friday, May 13th, 2022 Time: 01:00 PM Eastern Time (US and Canada)Location: EGR-2164 (ENGR)
Zoom Link:https://umd.zoom.us/j/96565970442?pwd=MU1xMWJyZUJEbS9zS1Zla21EN3hLdz09
Meeting ID: 965 6597 0442
Passcode: 4VrtYb

Abstract:
Analog circuits are a critical part of industrial circuits and systems. Estimates in the literature show that, even though analog circuits comprise less than 20% of all circuits, they are responsible for more than 80% of faults. Hence, analog circuit prognosis and health management (PHM) is critical to the health of industrial circuits. There are a multitude of ways that any analog circuit can fail, which leads to proportional scaling in the number of possible fault classes with the number of circuit components. Therefore, this research presents an advanced design of experiments-based (DOE) approach to account for components that degrade in an individual and interacting fashion, to narrow down the number of possible fault classes under consideration. A wavelet-based deep-learning approach is developed that can localize the circuit component that is the source of degradation and predict the exact value of the degraded component. This degraded value is used in conjunction with physics-of-failure models to predict when the circuit will fail based on the source of degradation.
Increasing outsourcing in the fabrication of electronic circuits has made them susceptible to the insertion of hardware trojans by untrusted foundries. In many cases, hardware trojans are more destructive than software trojans as they cannot be remedied by a software patch and are impossible to repair. Process reliability trojans are a new class of hardware trojans that are inserted through modification of fabrication parameters and accelerate the aging of circuit components. They are challenging to detect through traditional trojan detection methods as they have zero area footprint i.e., require no insertion of additional circuitry. The PHM approach is modified to detect these hardware trojans in order to incorporate circuit security, resulting in the PSHM framework.
Deep neural networks achieve state-of-the-art performance on classification and regression applications but are a black-box approach, which is a concern for implementation. Wavelets are approximations of cells found in the human visual cortex and cochlea. They were used to develop wavelet scattering networks (WSNs), which were intended to be an interpretable alternative to deep neural networks. WSNs achieve state-of-the-art performance on low to moderately complex datasets but are inferior to deep neural networks for extremely complex datasets. Improvements are made to WSNs to overcome their shortcomings in terms of performance and learnability. Further applications of the research are highlighted for rotating machinery vibration analytics, functional safety online estimation. 

Author: Nehemiah Emaikwu

Committee Members: Professor Reinhard Radermacher, Chair
Professor Ichiro Takeuchi, Co-Chair
Professor Yunho Hwang
Professor Amr Baz
Professor Bao Yang
Professor Peter B. Sunderland

Date: Friday, April 15th, 2022

Time: 1:30 PM

Location: EGR-2164 (DeWALT Conference Room)

Abstract: Elastocaloric solid-state refrigerants rival conventional refrigerants in lower environmental impact, but require significant advancements to gain widespread implementation. Two barriers which prevent adoption are low temperature lift and poor fatigue life. This dissertation addresses those challenges through a single, scalable architecture with the objectives of 1) designing high-performing elastocaloric devices, and 2) maximizing temperature lift. The developed prototype consists of 23 shortened and thermally insulated Ni-Ti tubes in a staggered pattern that exchange energy with the surrounding fluid medium through their external surface areas. They are contained inside of a 3D-printed plastic that provides alignment and restricts heat transfer to other components. A top loader and fixed bottom plate transfer compressive loads to the tubes, and a 3D-printed housing encapsulates all four parts.
Single, two, and three-stage configurations were experimentally investigated. A sensitivity analysis was conducted on the single-stage device and identified fluid-solid ratio, loading/unloading time, and strain as three parameters that could increase temperature span by over 1.5 K each. The combination of these findings resulted in a maximum steady-state temperature span of 16.6 K (9.7 K in heating and 6.8 K in cooling) at 4% strain and under zero load conditions. The temperature lift was increased in the two and three-stage configurations which achieved 20.2 K and 23.2 K, respectively, under similar operating conditions.
Validated 1D numerical models developed for this work confirm that the multi-staging approach positively impacts thermal response, though with decaying significance as the number of banks increase. By minimizing the water volume in the fluid loop, the three-stage device was able to develop a larger lift of 27.4 K. The tubes used in the single and two-stage tests also withstood over 30,000 cycles without failure, showing promising fatigue life behavior and emphasizing the viability of this alternative cooling technology.

Join Zoom Meeting: https://umd.zoom.us/j/4660347520

Meeting ID: 466 034 7520 

Author: Satwik Kommula
Date: Friday, April 15th, 2022 Time: 2-4pm Location: 088-2162 DeWALT

Committee Members:

Dr. Michael H. Azarian, Chair
Professor Peter Sandborn
Professor Patrick McCluskey

Title: Impact of Ripple Current on MLCCs

Abstract: The trend toward miniaturization along with the availability of multilayer ceramic capacitors (MLCCs) in a wide range of voltage ratings and capacitances, makes MLCCs a viable option to be used in applications that were previously reserved for electrolytic capacitors. Capacitors are widely used in filtering applications that involve current flowing through them because of a varying voltage, which is known as ripple current. The power dissipated by the parasitic resistance (ESR) of the MLCCs raises its temperature when current flows through it. Operating under elevated temperatures over long periods of time has the potential to degrade the performance of the MLCC resulting in a catastrophic failure or operating outside the limits. This study analyzes the performance of MLCCs when they are subjected to a varying voltage and compares the effects of different voltage ratings on the degradation of their electrical characteristics during extended exposure to ripple current. The failure mechanism for the degradation in insulation resistance observed in the tested MLCCs is also presented.

Author:  Han Zhou

Date: Wednesday, April 13th, 2022 at 4:00pm

Location: Glenn L. Martin Hall, Room EGR-2164

Committee Members:

Professor Amr Baz, Chair
Professor Bala Balachandran
Professor Nikhil Chopra
Professor Abhijit Dasgupta
Professor Sherif Aggour, Dean’s Representative

Title: ACTIVE NON-RECIPROCAL ACOUSTIC METAMATERIALS

Abstract:

This dissertation presents different configurations of active Acoustic MetaMaterials (AMM) which are proposed in order to control the flow of vibration and acoustic wave propagations in various applications. Distinct among these configurations is a 1-dimensional (1D) periodic array which consists of an assembly of active acoustic unit cells which are provided with programmable piezoelectric elements. By tuning the structural properties of these cells, the 1D array can impede the wave propagation over specific frequency ranges. In order to achieve non-reciprocal acoustic wave transmission of the AMMs, three different methodologies are introduced including active control of the piezoelectric elements using virtual gyroscopic control actions, eigenstructure shaping controller, and finally spatial-temporal modulation algorithm.
Theoretical models are developed to investigate the fundamentals and the underlying physical phenomena associated with all the considered three AMM configurations. Experimental prototypes of all these AMM configurations are built and tested to demonstrate their effectiveness in controlling the propagation of vibration and noise through these materials. Furthermore, the experimental results are used to validate the developed theoretical models. The developed theoretical and experimental approaches are envisioned to be valuable tools in the design of arrays of AMM for various applications which are only limited by our imagination.

Author: Michael Dawson

Committee Members:

Professor Jeffrey Herrmann- Assistant Professor

Yancy Diaz-Mercado

Mark Fuge

Title: Metareasoning Approaches to Thermal Management During Image Processing

Date: Friday, April 15th, 2022 Time: 10:00 AM Location: EGR-2164 (DeWALT Conference Room)

Abstract: Resource-constrained electronic systems are present in many semi- and fully-autonomous systems and are tasked with computationally heavy tasks such as image processing. Without sufficient cooling, these tasks often increase device temperature up to a predetermined maximum, beyond which the task is slowed by the device firmware to maintain the maximum. This is done to avoid decreased processor lifespan due to thermal fatigue or catastrophic processor failure due to thermal overstress. This thesis describes a study that evaluated how well metareasoning can manage the central processing unit (CPU) temperature during image processing (object detection and classification) on two devices: a Raspberry Pi 4B and an NVIDIA Jetson Nano Developer Kit
Three policies which employ metareasoning were developed; one which maintains constant image throughput, one which maintains constant expected detection precision, and a third which combines trades between throughput and precision losses based on a user-defined parameter. All policies used the EfficientDet series of object detectors. Depending on the policy, these networks were either switched between, delayed, or both. This thesis also considered cases that used the system’s built-in throttling policy to control the temperature. 
A policy was also created via reinforcement learning. The policy was able to adjust the detection precision and program throughput based on a set of states corresponding to the possible temperatures, neural networks, and processing delays. 
All three designed metareasoning policies were able to stabilize the device temperature without relying on thermal throttling. Additionally, the policy created through reinforcement learning was able to successfully stabilize the device temperature, though less consistently. These results suggest that a metareasoning-based approach to thermal management in image processing is able to provide a platform-agnostic and programmatic way to comply with temperature constraints.

Author:  Sergio Manuel Ignacio Cofre Martel 

Date: Friday, April 8th, 2022 at 1:00PM

Location: Glenn L. Martin Hall, Room EGR-2164

Committee Members:

Professor Dr. Mohammad Modarres, Chair + Dr. Enrique Lopez Droguett, Co-Chair
Professor Mark Fuge
Professor Katrina Groth
Professor Balakumar Balachandran
Professor Gregory Baecher, Dean’s Representative

Title: A Physics-Informed Neural Network Framework for Big Machinery Data in Prognostics and Health Management for Complex Engineering Systems

Abstract:

Big data analysis and data-driven models (DDMs) have become essential tools in prognostics and health management (PHM). Despite this, several challenges remain to successfully apply these techniques to complex engineering systems (CESs). Indeed, current state-of-the-art applications are treated as black-box algorithms, where research efforts have focused on developing complex DDMs, overlooking or neglecting the importance of the data preprocessing stages prior to training these models. Guidelines to adequately prepare data sets collected from CESs to train DDMs in PHM are frequently unclear or inexistent. Furthermore, these DDMs do not consider prior knowledge on the system’s physics of degradation, which gives little-to-no control over the data interpretation in reliability applications such as maintenance planning.

In this context, this dissertation presents a physics-informed neural network (PINN) architecture for remaining useful life (RUL) estimation based on big machinery data (BMD) collected from sensor monitoring networks (SMNs) in CESs. The main outcomes of this work are twofold. First, a systematic guide to preprocess BMD for diagnostics and prognostics tasks is developed based on expert knowledge and data science techniques. Second, a PINN-inspired PHM framework is proposed for RUL estimation through an open-box approach by exploring the system’s physics of degradation through partial differential equations (PDEs). The PINN-RUL framework aims to discover the system’s underlying physics-related behaviors, which could provide valuable information to create more trustworthy PHM models.

The data preprocessing and RUL estimation frameworks are validated through three case studies, including the C-MAPSS benchmark data set and two data sets corresponding to real CESs. Results show that the proposed preprocessing methodology can effectively generate data sets for supervised PHM models for CESs. Furthermore, the proposed PINN-RUL framework provides an interpretable latent variable that can capture the system’s degradation dynamics. This is a step forward to increase interpretability of prognostic models by mapping the RUL estimation to the latent space and its implementation as a state of health classifier. The PINN-RUL framework is flexible as it allows incorporating available physics-based models (PBMs) to its architecture. As such, this framework takes a step forward in bridging the gap between statistic-based PHM and physics-based PHM methods.

 

Author:  Walter Arias-Ramírez.

Date: Friday, April 8th, 2022 at 11:00AM

Location: Glenn L. Martin Hall, Room EGR-2164

Committee Members:

Professor Dr. Johan Larsson, Chair
Professor Dr. Amir Riaz
Professor James Duncan
Professor Kenneth Kiger
Professor Dr. James D. Baeder,Representative

Title: A multi-fidelity approach to sensitivity estimation in large eddy simulation.

Abstract:

 An approach to compute approximate sensitivities in a large eddy simulation (LES) is proposed and assessed. The multi-fidelity sensitivity analysis (MFSA) solves a linearized mean equation, where the mean equation is based on the LES solution. This requires closure modeling which makes the computed sensitivities approximate. The closure modeling is based on inferring the eddy viscosity from the LES data and predicting the change in turbulence (or the perturbed eddy viscosity) using a simple algebraic model. The method is assessed for the flow over a NACA0012 airfoil at a fixed angle of attack, with the Reynolds number as the varying parameter and the lift, drag, skin friction, and pressure coefficients as the quantities-of-interest. The results show the importance of accurate closure modeling, specifically that treating the eddy viscosity as “frozen” is insufficiently accurate. Also, predictions obtained using the algebraic model for closing the perturbed eddy viscosity are closer to the true sensitivity than results obtained using the fully RANS-based method which is the state-of-the-art and most common method used in industry. The proposed method aims to complement, rather than replace, the current state-of-the-art method in situations in which sensitivities with higher fidelity are required.

Author: Amirhossein Yazdkhasti

Date: Wednesday, April 6th, 2022 at 12:00PM

Location: Glenn L. Martin Hall, Room EGR-0159

Committee Members:

Professor Miao Yu, Chair/Advisor
Professor Amir Baz
Professor Balakumar Balachandran
Professor Nikhil Chopra
Professor Timothy Horiuchi, Dean’s Representative

Title: PASSIVE AND ACTIVE GRADED-INDEX ACOUSTIC METAMATERIALS: SPATIAL AND FREQUENCY DOMAIN MULTIPLEXING

Abstract:

Acoustic metamaterials, similar to their electromagnetic counterparts, are artificial subwavelength materials designed to manipulate sound waves. By tailoring the material’s effective properties such as bulk modulus, mass density, and reflective index, these materials can be designed to achieve unprecedented acoustic waves control and realize functional devices of novel properties. Specifically, high-refractive-index acoustic metamaterials have an effective refractive index much larger than air, enabling wave compression in space and a strong concentration of wave energy. Another type of acoustic metamaterials closely related to high-index acoustic metamaterials is graded-index metamaterials, which can be obtained by gradually varying material compositions or geometry over a volume of high-index acoustic metamaterials.
The overall goal of this dissertation is to achieve a fundamental understanding of passive and active graded-index acoustic metamaterials for spatial and frequency domain multiplexing and explore their applications in far-field acoustic imaging and sonar systems. Three research thrusts have been pursued. In the first thrust,the spatial domain multiplexing of passive graded-index acoustic metamaterials has been investigated for enhancing far-field acoustic imaging. An array of passive graded-index acoustic metamaterials has been designed and developed to achieve a far-field acoustic imaging system. Parametric studies have been carried out to facilitate the performance optimization of the imaging system. The performance of the metamaterial-based imaging system has been investigated and compared to the scenario without the metamaterials. In the second thrust, frequency-domain multiplexing with active graded-index acoustic metamaterials has been investigated. An active graded-index metamaterial system with a number of active unit cells has been designed and fabricated. A fundamental understanding of the frequency multiplexing properties of the metamaterials has been developed through numerical and experimental studies. In the third thrust, the capabilities of an acoustic sensing system with active graded-index metamaterials as an emitter for shape, size, and surface classification have been explored.