The goal of the proposed procedure in this dissertation is the implementation of phasor measurement unit (PMU) based instability detection for islanding prediction procedures using decision tree and neural network modeling. The islanding in the power system define as a separation of the coherent group of generators from the rest of the system due to contingencies, in the case that all generators are coherent together after introducing a fault, it is called stable or non-islanding. The main philosophy of islanding detection in the proposed methodology is to use decision trees and neural network data mining algorithms, performed off-line, to determine the PMU locations, detection parameters, and their triggering values for islanding detection. With the information obtained from accurate system models PMUs can be used online to predict system islanding with high reliability. The proposed approach is proved using a 4000 bus model of the California system. Before data mining was performed, a large number of islanding and non-islanding cases were created for the California model. PMUs data collection was simulated by collecting the voltage and current information in all 500 kV nodes in the system. More than 3000 cases were collected and classified by visual inspection as islanding and non-islanding cases. The proposed neural network and decision tree procedures captured the knowledge for the correct determination of system islanding with a small number of PMUs. / Ph. D.

Yes / In this work, an artificial neural network (ANN) model based on the experimental data was developed to study the performance of vacuum membrane distillation (VMD) desalination process under different operating parameters such as the feed inlet temperature, the vacuum pressure, the feed flow rate and the feed salt concentration. The proposed model was found to be capable of predicting accurately the unseen data of the VMD desalination process. The correlation coefficient of the overall agreement between the ANN predictions and experimental data was found to be more than 0.994. The calculation value of the coefficient of variation (CV) was 0.02622, and there was coincident overlap between the target and the output data from the 3D generalization diagrams. The optimal operating conditions of the VMD process can be obtained from the performance analysis of the ANN model with a maximum permeate flux and an acceptable CV value based on the experiment.

Northwest Montana has seen a significant increase in its population in the past twenty years. The increase in population, and associated development, is thought to be associated with "amenity migration"; people moving to an area to exploit the recreational opportunities that are unique to that area. Impervious surfaces can serve as a suitable proxy for tracking the spread of various anthropogenic influences on an ecosystem; it impacts groundwater recharge, increases overall surface runoff as well as pollution and sediment load, and fragments landscapes. In this study, an Artificial Neural Network model was developed to update NLCD impervious surface product (2011) in Flathead County, Montana. Four Landsat 8 images from 2015 and 2016 were used to characterize imperviousness. This multi-temporal analytical method was designed to reduce the spectral confusion between impervious surface and soil/agricultural lands. We compared the neural network-predicted impervious surface maps with 2011 NLCD. When all four neural network prediction images agreed with a change of 50% or more from the 2011 NLCD map, the average of those four images replaced that pixel from the 2011 imperviousness map. Compared to the ground truth, the method used showed significant promise, with an R2 of 0.73 and RMSE of 0.123. A comparison of the artificial neural network model results and the 2011 NLCD data showed a continuation of urbanization trends; the urban cores of towns in the study remain static while the majority of impervious surface development takes place along the perimeter of urban areas. / Master of Science / Remotely sensed Landsat data can be used to rapidly detect and estimate changes in impervious surface cover. This study used artificial neural networks in conjunction with the National Landcover Database’s 2011 Percent Developed Imperviousness layer and Landsat 8 data from four dates between the summer of 2015 and fall of 2016 to predict impervious surface cover in 2016, by deriving spectral relationships between Landsat data and impervious surfaces. We found that by requiring agreement between the four dates’ neural networks outputs, we eliminated many of the false positives that arose from exposed soil. Using this method, we achieved an R2 of 0.73 and RMSE of .123, sampling only the areas along a rural-urban gradient, in an area with significant seasonal spectral variability.

Yes / In the last few years, there has been growing interest in near-real-time solar data processing, especially for space weather applications. This is due to space weather impacts on both space-borne and ground-based systems, and industries, which subsequently impacts our lives. In the current study, the deep learning approach is used to establish an automated hybrid computer system for a short-term forecast; it is achieved by using the complexity level of the sunspot group on SDO/HMI Intensitygram images. Furthermore, this suggested system can generate the forecast for solar flare occurrences within the following 24 h. The input data for the proposed system are SDO/HMI full-disk Intensitygram images and SDO/HMI full-disk magnetogram images. System outputs are the “Flare or Non-Flare” of daily flare occurrences (C, M, and X classes). This system integrates an image processing system to automatically detect sunspot groups on SDO/HMI Intensitygram images using active-region data extracted from SDO/HMI magnetogram images (presented by Colak and Qahwaji, 2008) and deep learning to generate these forecasts. Our deep learning-based system is designed to analyze sunspot groups on the solar disk to predict whether this sunspot group is capable of releasing a significant flare or not. Our system introduced in this work is called ASAP_Deep. The deep learning model used in our system is based on the integration of the Convolutional Neural Network (CNN) and Softmax classifier to extract special features from the sunspot group images detected from SDO/HMI (Intensitygram and magnetogram) images. Furthermore, a CNN training scheme based on the integration of a back-propagation algorithm and a mini-batch AdaGrad optimization method is suggested for weight updates and to modify learning rates, respectively. The images of the sunspot regions are cropped automatically by the imaging system and processed using deep learning rules to provide near real-time predictions. The major results of this study are as follows. Firstly, the ASAP_Deep system builds on the ASAP system introduced in Colak and Qahwaji (2009) but improves the system with an updated deep learning-based prediction capability. Secondly, we successfully apply CNN to the sunspot group image without any pre-processing or feature extraction. Thirdly, our system results are considerably better, especially for the false alarm ratio (FAR); this reduces the losses resulting from the protection measures applied by companies. Also, the proposed system achieves a relatively high scores for True Skill Statistics (TSS) and Heidke Skill Score (HSS).

Artificial Neural Networks (ANNs) have proved to be useful in approximating non- linear systems in many applications including motion control. ANNs advocated in switched reluctance motor (SRM) control typically have a large number of neurons and several layers which impedes their real time implementation in embedded sys- tems. Real time estimation at high speeds using these ANNs is diffcult due to the high number of operations required to process the ANN controller. An insuffcient availability of time between two sampling intervals limits the available computation time for both processing the neural net and the other functions required for the motor drive. One ideal application of ANNs in SRM control is rotor position estimation. Due to reliability issues, elimination of the rotor position sensors is absolutely required for high volume, high speed and low cost applications of SRM's. ANNs provide a means by which drive designers can implement position sensorless drive technology that is both robust and easily implemented. It is demonstrated that a new and novel ANN configuration can be implemented for accurate rotor position estimation in a sensorless SRM drive. Consisting of just 4 neurons, the neural estimator is the smallest of its kind for SRM rotor position estimation. The breakthrough that provided the reduction was the addition of a non- linear input. Typical input spaces for SRM position neural estimators consist of both current,and fux-linkage. The neural network was trained on-line using these inputs and a third, non-linear input provided by a preprocessed product of the two typical inputs. / Master of Science

This study evaluates the effectiveness of an artificial neural network (ANN) to predict locations of urban change at a countywide level by testing various calibrations of the Land Transformation Model (LTM). It utilizes the Stuttgart Neural Network Simulator (SNNS), a common medium through which ANNs run a back-propagation algorithm, to execute neural net training. This research explores the dynamics of socioeconomic and biophysical variables (derived from the 1990 Comprehensive Plan) and how they affect model calibration for Montgomery County, Virginia. Using NLCD Retrofit Land Use data for 1992 and 2001 as base layers for urban change, we assess the sensitivity of the model with policy-influenced variables from data layers representing road accessibility, proximity to urban lands, distance from urban expansion areas, slopes, and soils. Aerial imagery from 1991 and 2002 was used to visually assess changes at site-specific locations. Results show a percent correct metric (PCM) of 32.843% and a Kappa value of 0.319. A relative operating characteristic (ROC) value of 0.660 showed that the model predicted locations of change better than chance (0.50). It performs consistently when compared to PCMs from a logistic regression model, 31.752%, and LTMs run in the absence of each driving variable ranging 27.971% – 33.494%. These figures are similar to results from other land use and land cover change (LUCC) studies sharing comparable landscape characteristics. Prediction maps resulting from LTM forecasts driven by the six variables tested provide a satisfactory means for forecasting change inside of dense urban areas and urban fringes for countywide urban planning. / Master of Science

yes / A 9 × 18 × 1 feed-forward neural network (NN) model trained using a resilient back-propagation algorithm and early stopping technique is constructed to predict the shear strength of deep reinforced concrete beams. The input layer covering geometrical and material properties of deep beams has nine neurons, and the corresponding output is the shear strength. Training, validation and testing of the developed neural network have been achieved using a comprehensive database compiled from 362 simple and 71 continuous deep beam specimens. The shear strength predictions of deep beams obtained from the developed NN are in better agreement with test results than those determined from strut-and-tie models. The mean and standard deviation of the ratio between predicted capacities using the NN and measured shear capacities are 1·028 and 0·154, respectively, for simple deep beams, and 1·0 and 0·122, respectively, for continuous deep beams. In addition, the trends ascertained from parametric study using the developed NN have a consistent agreement with those observed in other experimental and analytical investigations.

In this work, we propose ModelPred, a framework that helps to understand the impact of changes in training data on a trained model. This is critical for building trust in various stages of a machine learning pipeline: from cleaning poor-quality samples and tracking important ones to be collected during data preparation, to calibrating uncertainty of model prediction, to interpreting why certain behaviors of a model emerge during deployment. Specifically, ModelPred learns a parameterized function that takes a dataset S as the input and predicts the model obtained by training on S. Our work differs from the recent work of Datamodels as we aim for predicting the trained model parameters directly instead of the trained model behaviors. We demonstrate that a neural network-based set function class is capable of learning the complex relationships between the training data and model parameters. We introduce novel global and local regularization techniques to prevent overfitting and we rigorously characterize the expressive power of neural networks (NN) in approximating the end-to-end training process. Through extensive empirical investigations, we show that ModelPred enables a variety of applications that boost the interpretability and accountability of machine learning (ML), such as data valuation, data selection, memorization quantification, and model calibration. / Amazon-Virginia Tech Initiative in Efficient and Robust Machine Learning / Master of Science / Also published as Zeng, Y., Wang, J. T., Chen, S., Just, H. A., Jin, R., & Jia, R. (2023, February). ModelPred: A Framework for Predicting Trained Model from Training Data. In 2023 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) (pp. 432-449). IEEE. https://doi.org/10.1109/SaTML54575.2023.00037 / With the prevalence of large and complicated Artificial Intelligence (AI) models, it is important to build trust in the various stages of a machine learning model pipeline, from cleaning poor-quality samples and tracking important ones to be collected during the training data preparation, to calibrating uncertainty of model prediction during the inference stage, to interpreting why certain behaviors of a model emerge during deployment. In this work, we propose ModelPred, a framework that helps to understand the impact of changes in training data on a trained model. To achieve this, ModelPred learns a parameterized function that takes a dataset S as the input and predicts the model obtained by training on S, thus learning the impact from data on the model efficiently. Our work differs from the recent work of Datamodels [28] as we aim for predicting the trained model parameters directly instead of the trained model behaviors. We demonstrate that a neural network-based set function class is capable of learning the complex relationships between the training data and model parameters. We introduce novel global and local regularization techniques to enhance the generalizability and prevent overfitting. We also rigorously characterize the expressive power of neural networks (NN) in approximating the end-to-end training process. Through extensive empirical investigations, we show that ModelPred enables a variety of applications that boost the interpretability and accountability of machine learning (ML), such as data valuation, data selection, memorization quantification, and model calibration. This greatly enhances the trustworthy of machine learning models.

Stroke is a chronic illness which often impairs survivors for extended periods of time, leaving the individual limited in motor function. The ability to perform daily activities (ADL) is closely linked to motor recovery following a stroke. The objective of this work is to employ surface electromyography (sEMG) gathered through a novel, wearable armband sensor to monitor and quantify ADL performance. The first contribution of this work seeks to develop a relationship between sEMG and and grip aperture, a metric tied to the success of post-stroke individuals’ functional independence. The second contribution of this work aims to develop a deep learning model to classify RTG movements in the home setting using continuous EMG and acceleration data. In contribution one, ten non-disabled participants (10M, 22.5 0.5 years) were recruited. We performed a correlation analysis between aperture and peak EMG value, as well as a one-way non parametric analysis to determine cylinder diameter effect on aperture. In contribution two, one non-disabled participant is instructed to wash a set of dishes. The EMG and acceleration data collected is input into a recurrent neural network (RNN) machine learning model to classify movement patterns. The first contribution’s analysis demonstrated a strong positive correlation between aperture and peak EMG value, as well as a statistically significant effect of diameter (p < 0.001). The RNN model built in contribution two demonstrated high capability at classifying movement at 94% accuracy and an F1-score of 86%. These results demonstrate promising feasibility for long-term, in-home classification of daily tasks. Future applications of this approach should consider extending the procedure to include post-stroke individuals, as this could offer valuable insight into motor recovery within the home setting.

Road accidents pose a significant threat, leading to fatalities and injuries with far-reaching consequences. This study addresses two crucial challenges in road safety: analyzing traffic intersections to enhance safety by predicting potentially risky situations, and monitoring driver activity to prevent distracted driving accidents. Focusing on Virginia's intersections, we thoroughly examine traffic participant interactions to identify and mitigate conflicts, employing graph-based modeling of traffic scenarios to evaluate contributing parameters. Additionally, we leverage graph neural networks to detect and track potential crash situations from intersection videos, offering practical recommendations to enhance intersection safety. To understand the causes of risky behavior, we specifically investigate accidents resulting from distracted driving, which has become more prevalent due to advanced driver assistance systems in semi-autonomous vehicles. For monitoring driver activity inside vehicles, we propose the use of Video Transformers on challenging secondary driver activity datasets, incorporating grayscale and low-quality data to overcome limitations in capturing overall image context. Finally, we validate our predictions by studying attention modules and introducing explainability into the computer vision model. This research contributes to improving road safety by providing comprehensive analysis and recommendations for intersection safety enhancement and prevention of distracted driving accidents. / Master of Science / Road accidents are a serious problem causing numerous deaths and injuries each year. By studying driver behavior, we can uncover common causes of accidents like distracted driving, impaired driving, speeding, and not following traffic rules. New vehicle technologies aim to assist drivers, raising concerns about driver attentiveness. It is crucial for car manufacturers to develop systems that can detect and prevent accidents, especially in semi-autonomous vehicles. This study focuses on intersections in Virginia and examines driver behavior within vehicles to identify and prevent dangerous situations. We create models of different traffic scenarios using graphs/networks and utilize machine learning to identify potential accidents. Our objective is to provide practical recommendations for improving intersection safety. Existing datasets and algorithms for recognizing driver activities often fail to capture common distractions like eating, drinking, and phone use. To address this, we introduce two challenging datasets specifically designed to capture distracted driving activities. Finally, we try to understand the predictions bade by the chosen deep learning model by visualizing the inner workings.