Publications

We propose the use of balanced iterative reducing and clustering using hierarchies (BIRCH) combined with linear regression to predict the reduced Young's modulus and hardness of highly heterogeneous materials from a set of nanoindentation experiments. We first use BIRCH to cluster the dataset according to its mineral compositions, which are derived from the spectral matching of energy-dispersive spectroscopy data through the modular automated processing system (MAPS) platform. We observe that grouping our dataset into five clusters yields the best accuracy as well as a reasonable representation of mineralogy in each cluster. Subsequently, we test four types of regression models, namely linear regression, support vector regression, Gaussian process regression, and extreme gradient boosting regression. The linear regression and Gaussian process regression provide the most accurate prediction, and the proposed framework yields R2 = 0.93 for the test set. Although the study is needed more comprehensively, our results shows that machine learning methods such as linear regression or Gaussian process regression can be used to accurately estimate mechanical properties with a proper number of grouping based on compositional data.

The characterization of mixtures of non-interacting, spectroscopically similar quantum components has important applications in chemistry, biology, and materials science. We introduce an approach based on quantum tracking control that allows for determining the relative concentrations of constituents in a quantum mixture, using a single pulse which enhances the distinguishability of components of the mixture and has a length that scales linearly with the number of mixture constituents. To illustrate the method, we consider two very distinct model systems: mixtures of diatomic molecules in the gas phase, as well as solid-state materials composed of a mixture of components. A set of numerical analyses are presented, showing strong performance in both settings.

Any program tasked with the evaluation and acquisition of algorithms for use in deployed scenarios must have an impartial, repeatable, and auditable means of benchmarking both candidate and fielded algorithms. Success in this endeavor requires a body of representative sensor data, data labels indicating the proper algorithmic response to the data as adjudicated by subject matter experts, a means of executing algorithms under review against the data, and the ability to automatically score and report algorithm performance. Each of these capabilities should be constructed in support of program and mission goals. By curating and maintaining data, labels, tests, and scoring methodology, a program can understand and continually improve the relationship between benchmarked and fielded performance of acquired algorithms. A system supporting these program needs, deployed in an environment with sufficient computational power and necessary security controls is a powerful tool for ensuring due diligence in evaluation and acquisition of mission critical algorithms. This paper describes the Seascape system and its place in such a process.

It is impossible in practice to comprehensively test even small software programs due to the vastness of the reachable state space; however, modern cyber-physical systems such as aircraft require a high degree of confidence in software safety and reliability. Here we explore methods of generating test sets to effectively and efficiently explore the state space for a module based on the Traffic Collision Avoidance System (TCAS) used on commercial aircraft. A formal model of TCAS in the model-checking language NuSMV provides an output oracle. We compare test sets generated using various methods, including covering arrays, random, and a low-complexity input paradigm applied to 28 versions of the TCAS C program containing seeded errors. Faults are triggered by tests for all 28 programs using a combination of covering arrays and random input generation. Complexity-based inputs perform more efficiently than covering arrays, and can be paired with random input generation to create efficient and effective test sets. A random forest classifier identifies variable values that can be targeted to generate tests even more efficiently in future work, by combining a machine-learned fuzzing algorithm with more complex model oracles developed in model-based systems engineering (MBSE) software.

This paper reports on Catalyst usability and initial adoption by SPARC analysts. The use case approach highlights the analysts' perspective. Impediments to adoption can be due to deficiencies in software capabilities, or analysts may identify mundane inconveniences and barriers that prevent them from fully leveraging Catalyst. With that said, for many analyst tasks Catalyst provides enough relative advantage that they have begun applying it in their production work, and they recognize the potential for it to solve problems they currently struggle with. The findings in this report include specific issues and minor bugs in ParaView Python scripting, which are viewed as having straightforward solutions, as well as a broader adoption analysis.

We present a procedure for randomly generating realistic steady-state contingency scenarios based on the historical outage data from a particular event. First, we divide generation into classes and fit a probability distribution of outage magnitude for each class. Second, we provide a method for randomly synthesizing generator resilience levels in a way that preserves the data-driven probability distributions of outage magnitude. Finally, we devise a simple method of scaling the storm effects based on a single global parameter. We apply our methods using data from historical Winter Storm Uri to simulate contingency events for the ACTIVSg2000 synthetic grid on the footprint of Texas.

OpenMP 5.0 added support for reductions over explicit tasks. This expands the previous reduction support that was limited primarily to worksharing and parallel constructs. While the scope of a reduction operation in a worksharing construct is the scope of the construct itself, the scope of a task reduction can vary. This difference requires syntactical means to define the scope of reductions, e.g., the task_reduction clause, and to associate participating tasks, e.g., the in_reduction clause. Furthermore, the disassociation of the number of threads and the number of tasks creates space for different implementations in the OpenMP runtime. In this work, we provide insights into the behavior and performance of task reduction implementations in GCC/g++ and LLVM/Clang. Our results indicate that task reductions are well supported by both compilers, but their performance differs in some cases and is often determined by the efficiency of the underlying task management.

Stochasticity is ubiquitous in the world around us. However, our predominant computing paradigm is deterministic. Random number generation (RNG) can be a computationally inefficient operation in this system especially for larger workloads. Our work leverages the underlying physics of emerging devices to develop probabilistic neural circuits for RNGs from a given distribution. However, codesign for novel circuits and systems that leverage inherent device stochasticity is a hard problem. This is mostly due to the large design space and complexity of doing so. It requires concurrent input from multiple areas in the design stack from algorithms, architectures, circuits, to devices. In this paper, we present examples of optimal circuits developed leveraging AI-enhanced codesign techniques using constraints from emerging devices and algorithms. Our AI-enhanced codesign approach accelerated design and enabled interactions between experts from different areas of the micro-electronics design stack including theory, algorithms, circuits, and devices. We demonstrate optimal probabilistic neural circuits using magnetic tunnel junction and tunnel diode devices that generate an RNG from a given distribution.

Interest in the application of DC Microgrids to distribution systems have been spurred by the continued rise of renewable energy resources and the dependence on DC loads. However, in comparison to AC systems, the lack of natural zero crossing in DC Microgrids makes the interruption of fault currents with fuses and circuit breakers more difficult. DC faults can cause severe damage to voltage-source converters within few milliseconds, hence, the need to quickly detect and isolate the fault. In this paper, the potential for five different Machine Learning (ML) classifiers to identify fault type and fault resistance in a DC Microgrid is explored. The ML algorithms are trained using simulated fault data recorded from a 750 VDC Microgrid modeled in PSCAD/EMTDC. The performance of the trained algorithms are tested using real fault data gathered from an operational DC Microgrid located on the Kirtland Air Force Base. Of the five ML algorithms, three could detect the fault and determine the fault type with at least 99% accuracy, and only one could estimate the fault resistance with at least 99% accuracy. By performing a self-learning monitoring and decision making analysis, protection relays equipped with ML algorithms can quickly detect and isolate faults to improve the protection operations on DC Microgrids.

We present a field-deployable microfluidic immunoassay device in response to the need for sensitive, quantitative, and high-throughput protein detection at point-of-need. The portable microfluidic system facilitates eight magnetic bead-based sandwich immunoassays from raw samples in 45 minutes. An innovative bead actuation strategy was incorporated into the system to automate multiple sample process steps with minimal user intervention. The device is capable of quantitative and sensitive protein analysis with a 10 pg/ml detection limit from interleukin 6-spiked human serum samples. We envision the reported device offering ultrasensitive point-of-care immunoassay tests for timely and accurate clinical diagnosis.

Iterative methods for solving linear systems serve as a basic building block for computational science. The computational cost of these methods can be significantly influenced by the round-off errors that accumulate as a result of their implementation in finite precision. In the extreme case, round-off errors that occur in practice can completely prevent an implementation from satisfying the accuracy and convergence behavior prescribed by its underlying algorithm. In the exascale era where cost is paramount, a thorough and rigorous analysis of the delay of convergence due to round-off should not be ignored. In this paper, we use a small model problem and the Jacobi iterative method to demonstrate how the Coq proof assistant can be used to formally specify the floating-point behavior of iterative methods, and to rigorously prove the accuracy of these methods.

This paper presents a run-to-run (R2R) controller for mechanical serial sectioning (MSS). MSS is a destructive material analysis process which repeatedly removes a thin layer of material and images the exposed surface. The images are then used to gain insight into the material properties and often to construct a 3-dimensional reconstruction of the material sample. Currently, an experience human operator selects the parameters of the MSS to achieve the desired thickness. The proposed R2R controller will automate this process while improving the precision of the material removal. The proposed R2R controller solves an optimization problem designed to minimize the variance of the material removal subject to achieving the expected target removal. This optimization problem was embedded in an R2R framework to provide iterative feedback for disturbance rejection and convergence to the target removal amount. Since an analytic model of the MSS system is unavailable, we adopted a data-driven approach to synthesize our R2R controller from historical data. The proposed R2R controller is demonstrated through simulations. Future work will empirically demonstrate the proposed R2R through experiments with a real MSS system.

Abstract not provided.

Abstract not provided.

Metasurface lenses are fabricated using membrane projection lithography following a CMOS-compatible process flow. The lenses are 10-mm in diameter and employ 3-dimensional unit cells designed to function in the mid-infrared spectral range.

Incorrect modeling of control characteristics for inverter-based resources (IBRs) can affect the accuracy of electric power system studies. In many distribution system contexts, the control settings for behind-the-meter (BTM) IBRs are unknown. This paper presents an efficient method for selecting a small number of time series samples from net load meter data that can be used for reconstructing or classifying the control settings of BTM IBRs. Sparse approximation techniques are used to select the time series samples that cause the inversion of a matrix of candidate responses to be as well-conditioned as possible. We verify these methods on 451 actual advanced metering infrastructure (AMI) datasets from loads with BTM IBRs. Selecting 60 15-minute granularity time series samples, we recover BTM control characteristics with a mean error less than 0.2 kVAR.

Variable energy resources (VERs) like wind and solar are the future of electricity generation as we gradually phase out fossil fuel due to environmental concerns. Nations across the globe are also making significant strides in integrating VERs into their power grids as we strive toward a greener future. However, integration of VERs leads to several challenges due to their variable nature and low inertia characteristics. In this paper, we discuss the hurdles faced by the power grid due to high penetration of wind power generation and how energy storage system (ESSs) can be used at the grid-level to overcome these hurdles. We propose a new planning strategy using which ESSs can be sized appropriately to provide inertial support as well as aid in variability mitigation, thus minimizing load curtailment. A probabilistic framework is developed for this purpose, which takes into consideration the outage of generators and the replacement of conventional units with wind farms. Wind speed is modeled using an autoregressive moving average technique. The efficacy of the proposed methodology is demonstrated on the WSCC 9-bus test system.

This work explores deriving transmissibility functions for a missile from a measured location at the base of the fairing to a desired location within the payload. A pressure on the outside of the fairing and the rocket motor’s excitation creates an acceleration at a measured location and a desired location. Typically, the desired location is not measured. In fact, it is typical that the payload may change, but measured acceleration at the base of the fairing is generally similar to previous test flights. Given this knowledge, it is desired to use a finite-element model to create a transmissibility function which relates acceleration from the previous test flight’s measured location at the base of the fairing to acceleration at a location in the new payload. Four methods are explored for deriving this transmissibility, with the goal of finding an appropriate transmissibility when both the pressure and rocket motor excitation are equally present. These methods are assessed using transient results from a simple example problem, and it is found that one of the methods gives good agreement with the transient results for the full range of loads considered.

Growing interest in renewable energy sources has led to an increased installation rate of distributed energy resources (DERs) such as solar photovoltaics (PVs) and wind turbine generators (WTGs). The variable nature of DERs has created several challenges for utilities and system operators related to maintaining voltage and frequency. New grid standards are requiring DERs to provide voltage regulation across distribution networks. Volt-Var Curve (VVC) control is an autonomous grid-support function that provides voltage regulation based on the relationship between voltage and reactive power. This paper evaluates the performance of a WTG operating with VVC control. The evaluation of the model involves a MATLAB/Simulink simulation of a distribution system. For this simulation the model considers three WTGs and a variable load that creates a voltage event.

Hyperspectral Computed Tomography (HCT) Data is often visualized using dimension reduction algorithms. However, these methods often fail to adequately differentiate between materials with similar spectral signatures. Previous work showed that a combination of image preprocessing, clustering, and dimension reduction techniques can be used to colorize simulated HCT data and enhance the contrast between similar materials. In this work, we evaluate the efficacy of these existing methods on experimental HCT data and propose new improvements to the robustness of these methods. We introduce an automated channel selection method and compare the Feldkamp, Davis, and Kress filtered back-projection (FBP) algorithm with the maximum-likelihood estimation-maximization (MLEM) algorithm in terms of HCT reconstruction image quality and its effect on different colorization methods. Additionally, we propose adaptations to the colorization process that eliminate the need for a priori knowledge of the number distinct materials for material classification. Our results show that these methods generalize to materials in real-world experimental HCT data for both colorization and classification tasks; both tasks have applications in industry, medicine, and security, wherever rapid visualization and identification is needed.

The focus of this study is on spectral equivalence results for higher-order tensor product finite elements in the H(curl), H(div), and L2 function spaces. For certain choices of the higher-order shape functions, the resulting mass and stiffness matrices are spectrally equivalent to those for an assembly of lowest-order edge-, face- or interior-based elements on the associated Gauss–Lobatto–Legendre (GLL) mesh.

This work presents an experimental investigation of the deformation and breakup of water drops behind conical shock waves. A conical shock is generated by firing a bullet at Mach 4.5 past a vertical column of drops with a mean initial diameter of 192 µm. The time-resolved drop position and maximum transverse dimension are characterized using backlit stereo videos taken at 500 kHz. A Reynolds-Averaged Navier Stokes (RANS) simulation of the bullet is used to estimate the gas density and velocity fields experienced by the drops. Classical correlations for breakup times derived from planar-shock/drop interactions are evaluated. Predicted drop breakup times are found to be in error by a factor of three or more, indicating that existing correlations are inadequate for predicting the response to the three-dimensional relaxation of the velocity and thermodynamic properties downstream of the conical shock. Next, the Taylor Analogy Breakup (TAB) model, which solves a transient equation for drop deformation, is evaluated. TAB predictions for drop diameter calculated using a dimensionless constant of C2 = 2, as compared to the accepted value of C2 = 2/3, are found to agree within the confidence bounds of the ensemble averaged experimental values for all drops studied. These results suggest the three-dimensional relaxation effects behind conical shock waves alter the drop response in comparison to a step change across a planar shock, and that future models describing the interaction between a drop and a non-planar shock wave should account for flow field variations.

The installation of digital sensors, such as advanced meter infrastructure (AMI) meters, has provided the means to implement a wide variety of techniques to increase visibility into the distribution system, including the ability to calibrate the utility models using data-driven algorithms. One challenge in maintaining accurate and up-to-date distribution system models is identifying changes and event occurrences that happen during the year, such as customers who have changed phases due to maintenance or other events. This work proposes a method for the detection of phase change events that utilizes techniques from an existing phase identification algorithm. This work utilizes an ensemble step to obtain predicted phases for windows of data, therefore allowing the predicted phase of customers to be observed over time. The proposed algorithm was tested on four utility datasets as well as a synthetic dataset. The synthetic tests showed the algorithm was capable of accurately detecting true phase change events while limiting the number of false-positive events flagged. In addition, the algorithm was able to identify possible phase change events on two real datasets.

This paper describes how the performance of motion primitive-based planning algorithms can be improved using reinforcement learning. Specifically, we describe and evaluate a framework that autonomously improves the performance of a primitive-based motion planner. The improvement process consists of three phases: exploration, extraction, and reward updates. This process can be iterated continuously to provide successive improvement. The exploration step generates new trajectories, and the extraction step identifies new primitives from these trajectories. These primitives are then used to update rewards for continued exploration. This framework required novel shaping rewards, development of a primitive extraction algorithm, and modification of the Hybrid A* algorithm. The framework is tested on a navigation task using a nonlinear F-16 model. The framework autonomously added 91 motion primitives to the primitive library and reduced average path cost by 21.6 s, or 35.75% of the original cost. The learned primitives are applied to an obstacle field navigation task, which was not used in training, and reduced path cost by 16.3 s, or 24.1%. Additionally, two heuristics for the modified Hybrid A* algorithm are designed to improve effective branching factor.

Neuromorphic computing promises revolutionary improvements over conventional systems for applications that process unstructured information. To fully realize this potential, neuromorphic systems should exploit the biomimetic behavior of emerging nanodevices. In particular, exceptional opportunities are provided by the non-volatility and analog capabilities of spintronic devices. While spintronic devices that emulate neurons have been previously proposed, they require complementary metal-oxide semiconductor (CMOS) technology to function. In turn, this significantly increases the power consumption, fabrication complexity, and device area of a single neuron. This work reviews three previously proposed CMOS-free spintronic neurons designed to resolve this issue.