Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Distance Learning”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 343 records · Page 19

Interpretable Machine Learning for Characterizing Electric Vehicle Charging Behavior: Insights from Real-World Data

As electric vehicle (EV) adoption rises globally, concerns about the impact on aging electrical grids grow, particularly regarding the charging behavior of EV drivers. This study analyzes real-world driving and charging data from Ford battery electric vehicles (BEVs) collected between 2018 and 2019 to develop interpretable models that characterize charging behavior and quantify influencing factors. Prior research has relied on assumptions regarding driver behavior, often overlooking actual charging patterns. By employing generalized linear mixed models (GLMMs), this work offers insights into how various elements, such as next trip distance and state of charge (SOC), influence charging decisions. The dataset comprises over three million park-trip pairs from 1,997 vehicles, revealing that features related to driving behavior significantly dictate charging behavior, while infrastructure and regional factors have lesser impacts. The findings suggest that existing simulation models may oversimplify EV charging behavior assumptions. This work utilizes real-world EV driving and charging data to train interpretable models that describe charging behavior and quantify the factors most associated with how drivers use charging infrastructure. This research underscores the need for interpretable, data-driven methodologies to inform future EV infrastructure planning and grid management.

29 - ENERGY PLANNING, POLICY AND ECONOMY↗

The Evaluation of Machine Learning Techniques for Isotope Identification Contextualized by Training and Testing Spectral Similarity

Precise gamma-ray spectral analysis is crucial in high-stakes applications, such as nuclear security. Research efforts toward implementing machine learning (ML) approaches for accurate analysis are limited by the resemblance of the training data to the testing scenarios. The underlying spectral shape of synthetic data may not perfectly reflect measured configurations, and measurement campaigns may be limited by resource constraints. Consequently, ML algorithms for isotope identification must maintain accurate classification performance under domain shifts between the training and testing data. To this end, four different classifiers (Ridge, Random Forest, Extreme Gradient Boosting, and Multilayer Perceptron) were trained on the same dataset and evaluated on twelve other datasets with varying standoff distances, shielding, and background configurations. A tailored statistical approach was introduced to quantify the similarity between the training and testing configurations, which was then related to the predictive performance. Wilcoxon signed-rank tests revealed that the OVR-wrapped XGB significantly outperformed the other algorithms, with confidence levels of 99.0% or above for the 133Ba, 60Co, 137Cs, and 152Eu sources. The findings from this work are significant as they outline techniques to promote the development of robust ML-based approaches for isotope identification.

domain adaptation↗

SU-E-J-206: A Comparison of Different Hardware Design Approaches for Feature-Supported Optical Head-Tracking with Respect to Angular Dependencies

Purpose: Recent research has shown that optical features significantly improve marker-less optical head-tracking for cranial radiotherapy. Simulations, however, showed that these optical features, which are used to derive tissue thickness, depend on the incident angle of the IR scanning laser beam and the perspective of the camera analyzing the reflective patterns. We present an experimental analysis determining which is the most robust optical setup concerning angular influences. Methods: In three consecutive experiments, the incident angle of the laser (1), the perspective of the camera (2) or both simultaneously (3, ‘inBeam’-perspective) were changed with respect to the target. We analyzed how this affects feature intensity. These intensities were determined from seven concentric regions of interest (ROIs) around the laser spot. Two targets were used: a tissue-like silicone phantom and a human's forehead. Results: For each experiment, the feature intensity generally decreases with increasing angle. We found that the optical properties of the silicone phantom do not fit the properties of human skin. Furthermore, the angular influence of the laser on the features is significantly higher than the perspective of the camera. With the ‘inBeam’- perspective, the smoothest decays of feature intensity were found. We suppose that this is because of a fixed relationship between both devices. This smoothness, suggesting a predictable functional relationship, may simplify angle compensation for machine learning algorithms. This is particularly prominent for the medial ROIs. The inner ROIs highly depend on the angle and power of the laser. The outer ROIs show less angular dependency but the signal strength is critically low and prone to artifacts. Therefore and because of the smooth decays, medial ROIs are a suitable tradeoff between susceptibility, signal-noise-ratio and distance to the center of the laser spot. Conclusion: For tissue thickness correlated feature acquisition, the medial ROIs with the ‘inBeam’-setup provide most valuable features This work was supported by Varian Medical Systems Inc. (Palo Alto, CA, USA). In addition, this work was supported by the Graduate School for Computing in Medicine and Life Sciences funded by Germany's Excellence Initiative [DFG GSC 235/1].

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Jacobian-scaled K-means clustering for physics-informed segmentation of reacting flows

This work introduces Jacobian-scaled K-means (JSK-means) clustering, which is a physicsinformed clustering strategy centered on the K-means framework. The method allows for the injection of underlying physical knowledge into the clustering procedure through a distance function modification: instead of leveraging conventional Euclidean distance vectors, the JSKmeans procedure operates on distance vectors scaled by matrices obtained from dynamical system Jacobians evaluated at the cluster centroids. The goal of this work is to show how the JSKmeans algorithm - without modifying the input dataset - produces clusters that capture regions of dynamical similarity, in that the clusters are redistributed towards high-sensitivity regions in phase space and are described by similarity in the source terms of samples instead of the samples themselves. The algorithm is demonstrated on a complex reacting flow simulation dataset (a channel detonation configuration), where the dynamics in the thermochemical composition space are known through the highly nonlinear and stiff Arrhenius-based chemical source terms. Interpretations of cluster partitions in both physical space and composition space reveal how JSK-means shifts clusters produced by standard K-means towards regions of high chemical sensitivity (e.g., towards regions of peak heat release rate near the detonation reaction zone). Furthermore, the findings presented here illustrate the benefits of utilizing Jacobian-scaled distances in clustering techniques, and the JSK-means method in particular displays promising potential for improving former partition-based modeling strategies in reacting flow (and other multi-physics) applications.

Clustering↗

What did we learn from gamma-ray burst 080319B?

The optical and gamma-ray observations of GRB 080319B allow us to provide a broad-brush picture for this remarkable burst. The data indicate that the prompt optical and gamma-ray photons were possibly produced at the same location but by different radiation processes: synchrotron and synchrotron self-Compton, respectively (but we note that this interpretation of the gamma-ray data faces some difficulties). We find that the burst prompt optical emission was produced at a distance of 10{sup 16.3} cm by an ultrarelativistic source moving at Lorentz factor of -500. A straightforward inference is that about 10 times more energy must have been radiated at tens of GeV than that released at 1 MeV. Assuming that the GRB outflow was baryonic and the gamma-ray source was shock-heated plasma, the collimation-corrected kinetic energy of the jet powering GRB 080319B was larger than 10{sup 52.3} erg. The decay of the early afterglow optical emission (up to 1 ks) is too fast to be attributed to the reverse-shock crossing the GRB ejecta but is consistent with the expectations for the 'large-angle' emission released during the burst. The pure power-law decay of the optical afterglow flux from 1 ks to 10 d is most naturally identified with the (synchrotron) emission from the shock propagating into a wind-like medium. However, the X-ray afterglow requires a departure from the standard blast-wave model.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Cavity contour segmentation in chest radiographs using supervised learning and dynamic programming

Purpose: Efficacy of tuberculosis (TB) treatment is often monitored using chest radiography. Monitoring size of cavities in pulmonary tuberculosis is important as the size predicts severity of the disease and its persistence under therapy predicts relapse. The authors present a method for automatic cavity segmentation in chest radiographs. Methods: A two stage method is proposed to segment the cavity borders, given a user defined seed point close to the center of the cavity. First, a supervised learning approach is employed to train a pixel classifier using texture and radial features to identify the border pixels of the cavity. A likelihood value of belonging to the cavity border is assigned to each pixel by the classifier. The authors experimented with four different classifiers:k-nearest neighbor (kNN), linear discriminant analysis (LDA), GentleBoost (GB), and random forest (RF). Next, the constructed likelihood map was used as an input cost image in the polar transformed image space for dynamic programming to trace the optimal maximum cost path. This constructed path corresponds to the segmented cavity contour in image space. Results: The method was evaluated on 100 chest radiographs (CXRs) containing 126 cavities. The reference segmentation was manually delineated by an experienced chest radiologist. An independent observer (a chest radiologist) also delineated all cavities to estimate interobserver variability. Jaccard overlap measure Ω was computed between the reference segmentation and the automatic segmentation; and between the reference segmentation and the independent observer's segmentation for all cavities. A median overlap Ω of 0.81 (0.76 ± 0.16), and 0.85 (0.82 ± 0.11) was achieved between the reference segmentation and the automatic segmentation, and between the segmentations by the two radiologists, respectively. The best reported mean contour distance and Hausdorff distance between the reference and the automatic segmentation were, respectively, 2.48 ± 2.19 and 8.32 ± 5.66 mm, whereas these distances were 1.66 ± 1.29 and 5.75 ± 4.88 mm between the segmentations by the reference reader and the independent observer, respectively. The automatic segmentations were also visually assessed by two trained CXR readers as “excellent,” “adequate,” or “insufficient.” The readers had good agreement in assessing the cavity outlines and 84% of the segmentations were rated as “excellent” or “adequate” by both readers. Conclusions: The proposed cavity segmentation technique produced results with a good degree of overlap with manual expert segmentations. The evaluation measures demonstrated that the results approached the results of the experienced chest radiologists, in terms of overlap measure and contour distance measures. Automatic cavity segmentation can be employed in TB clinics for treatment monitoring, especially in resource limited settings where radiologists are not available.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

3D liver segmentation using multiple region appearances and graph cuts

Purpose: Efficient and accurate 3D liver segmentations from contrast-enhanced computed tomography (CT) images play an important role in therapeutic strategies for hepatic diseases. However, inhomogeneous appearances, ambiguous boundaries, and large variance in shape often make it a challenging task. The existence of liver abnormalities poses further difficulty. Despite the significant intensity difference, liver tumors should be segmented as part of the liver. This study aims to address these challenges, especially when the target livers contain subregions with distinct appearances. Methods: The authors propose a novel multiregion-appearance based approach with graph cuts to delineate the liver surface. For livers with multiple subregions, a geodesic distance based appearance selection scheme is introduced to utilize proper appearance constraint for each subregion. A special case of the proposed method, which uses only one appearance constraint to segment the liver, is also presented. The segmentation process is modeled with energy functions incorporating both boundary and region information. Rather than a simple fixed combination, an adaptive balancing weight is introduced and learned from training sets. The proposed method only calls initialization inside the liver surface. No additional constraints from user interaction are utilized. Results: The proposed method was validated on 50 3D CT images from three datasets, i.e., Medical Image Computing and Computer Assisted Intervention (MICCAI) training and testing set, and local dataset. On MICCAI testing set, the proposed method achieved a total score of 83.4 ± 3.1, outperforming nonexpert manual segmentation (average score of 75.0). When applying their method to MICCAI training set and local dataset, it yielded a mean Dice similarity coefficient (DSC) of 97.7% ± 0.5% and 97.5% ± 0.4%, respectively. These results demonstrated the accuracy of the method when applied to different computed tomography (CT) datasets. In addition, user operator variability experiments showed its good reproducibility. Conclusions: A multiregion-appearance based method is proposed and evaluated to segment liver. This approach does not require prior model construction and so eliminates the burdens associated with model construction and matching. The proposed method provides comparable results with state-of-the-art methods. Validation results suggest that it may be suitable for the clinical use.

60 APPLIED LIFE SCIENCES↗

PANTHER: Trajectory Analysis

We want to organize a body of trajectories in order to identify, search for, classify and predict behavior among objects such as aircraft and ships. Existing comparison functions such as the Fréchet distance are computationally expensive and yield counterintuitive results in some cases. We propose an approach using feature vectors whose components represent succinctly the salient information in trajectories. These features incorporate basic information such as total distance traveled and distance between start/stop points as well as geometric features related to the properties of the convex hull, trajectory curvature and general distance geometry. Additionally, these features can generally be mapped easily to behaviors of interest to humans that are searching large databases. Most of these geometric features are invariant under rigid transformation. We demonstrate the use of different subsets of these features to identify trajectories similar to an exemplar, cluster a database of several hundred thousand trajectories, predict destination and apply unsupervised machine learning algorithms.

42 ENGINEERING↗

Optimal Coordination of Electric Vehicles for Grid Services using Deep Reinforcement Learning

Recent research has shown the effectiveness of reinforcement learning (RL) in coordinating electric vehicles (EVs) with vehicle-to-grid capabilities for grid services. However, many of these studies rely on lookup table and deep Q-network techniques, which can be impractical when dealing with continuous states and actions. In addition, existing RL designs inadequately account for battery aging effects, EV user satisfaction, uncertain departure and arrival time, and trip distance, which may compromise effective coordination. This paper aims to bridge these gaps by developing an innovative deep deterministic policy gradient-based RL framework for optimal coordination of EVs. Case studies were carried out using a test system with 100 EVs, and numerical analysis results showed that the proposed RL framework can effectively coordinate EVs to maximize economic benefits and user satisfaction while ensuring the expected battery lifespan.

Das, Avijit↗

SU-D-BRB-02: Patient-Specific Rectal Toxicity Predictor Based Plan Quality Control for Prostate Stereotactic Body Radiation Therapy (SBRT)

Purpose: To develop a patient-specific rectal toxicity predictor guided plan quality control tool for prostate SBRT plans. Methods: For prostate SBRT cases, four segments of rectal walls including peri-prostatic anterior rectal wall, peri-prostatic lateral rectal walls, peri-prostatic posterior rectal wall and rectum superior to prostate are identified as organs at risk and the circumference of rectal wall receiving more than 39 Gy (CRW39) and 24 Gy (CRW24) are rectal toxicity predictors. In this new geometry-dosimetry model, a patient geometry descriptor, differential circumference of rectal wall (dCRW) is used as model input geometry parameters and plan dosimetric endpoints CRW39 and CRW24 are output dosimetric parameters. Linear models are built to correlate dCRW to both CRW39 and CRW24 and established with both a linear regression method and a modified bagging ensemble machine learning method. 27 SBRT prostate cases are retrospectively studied from a dose-escalated clinical trial research. 20 prescribed 50 Gy SBRT cases are recruited to train the model and the other rescaled 7 cases are used to evaluated model feasibility and accuracy. Results: Each solved linear coefficient sequence related to CRW39 or CRW24 is a one-dimensional decreasing function of the distance from the PTV boundary, indicating that the different locations of each rectal circumference have different contributions to each particular dosimetric endpoint. The fitting errors for those trained 20 prostate SBRT cases are small with mean values of 2.39%, 2.45% relative to the endpoint values for SBRT rectal toxicity predictor CRW39 and CRW24 respectively. 1 out of 7 evaluation plans is identified as poor quality plan. After re-planning, the CRW39 and CRW24 can be reduced by 3.34% and 3%, without sacrificing PTV coverage. Conclusion: The proposed patient geometry-plan toxicity predictor model for SBRT plans can be successfully applied to plan quality control for prostate SBRT cases.

60 APPLIED LIFE SCIENCES↗

SU-D-204-01: A Methodology Based On Machine Learning and Quantum Clustering to Predict Lung SBRT Dosimetric Endpoints From Patient Specific Anatomic Features

Purpose: To develop a data-mining methodology based on quantum clustering and machine learning to predict expected dosimetric endpoints for lung SBRT applications based on patient-specific anatomic features. Methods: Ninety-three patients who received lung SBRT at our clinic from 2011–2013 were retrospectively identified. Planning information was acquired for each patient, from which various features were extracted using in-house semi-automatic software. Anatomic features included tumor-to-OAR distances, tumor location, total-lung-volume, GTV and ITV. Dosimetric endpoints were adopted from RTOG-0195 recommendations, and consisted of various OAR-specific partial-volume doses and maximum point-doses. First, PCA analysis and unsupervised quantum-clustering was used to explore the feature-space to identify potentially strong classifiers. Secondly, a multi-class logistic regression algorithm was developed and trained to predict dose-volume endpoints based on patient-specific anatomic features. Classes were defined by discretizing the dose-volume data, and the feature-space was zero-mean normalized. Fitting parameters were determined by minimizing a regularized cost function, and optimization was performed via gradient descent. As a pilot study, the model was tested on two esophageal dosimetric planning endpoints (maximum point-dose, dose-to-5cc), and its generalizability was evaluated with leave-one-out cross-validation. Results: Quantum-Clustering demonstrated a strong separation of feature-space at 15Gy across the first-and-second Principle Components of the data when the dosimetric endpoints were retrospectively identified. Maximum point dose prediction to the esophagus demonstrated a cross-validation accuracy of 87%, and the maximum dose to 5cc demonstrated a respective value of 79%. The largest optimized weighting factor was placed on GTV-to-esophagus distance (a factor of 10 greater than the second largest weighting factor), indicating an intuitively strong correlation between this feature and both endpoints. Conclusion: This pilot study shows that it is feasible to predict dose-volume endpoints based on patient-specific anatomic features. The developed methodology can potentially help to identify patients at risk for higher OAR doses, thus improving the efficiency of treatment planning. R01-184173.

60 APPLIED LIFE SCIENCES↗

WaveDP

We developed a robust deep learning model designed to effectively reduce background noise and measure signal arrival times from seismic waveforms. This model processes a 57-second, three-component seismogram to predict the probability of Primary (P) and Secondary (S) waves for each timestamp, while also generating a denoised seismogram. Training was conducted using the benchmark STEAD dataset, which includes globally distributed earthquake signals recorded at local distances ranging from 0 to 350 kilometers.

Chai, Chengping [Oak Ridge National Laboratory (OR↗

Forensic analysis of the microbiome of phones and shoes

Background: Microbial interaction between human-associated objects and the environments we inhabit may have forensic implications, and the extent to which microbes are shared between individuals inhabiting the same space may be relevant to human health and disease transmission. In this study, two participants sampled the front and back of their cell phones, four different locations on the soles of their shoes, and the floor beneath them every waking hour over a 2-day period. A further 89 participants took individual samples of their shoes and phones at three different scientific conferences. Results: Samples taken from different surface types maintained significantly different microbial community structures. The impact of the floor microbial community on that of the shoe environments was strong and immediate, as evidenced by Procrustes analysis of shoe replicates and significant correlation between shoe and floor samples taken at the same time point. Supervised learning was highly effective at determining which participant had taken a given shoe or phone sample, and a Bayesian method was able to determine which participant had taken each shoe sample based entirely on its similarity to the floor samples. Both shoe and phone samples taken by conference participants clustered into distinct groups based on location, though much more so when an unweighted distance metric was used, suggesting sharing of low-abundance microbial taxa between individuals inhabiting the same space. In conclusion, correlations between microbial community sources and sinks allow for inference of the interactions between humans and their environment.

59 BASIC BIOLOGICAL SCIENCES↗

Self-Supervised Anomaly Detection via Neural Autoregressive Flows with Active Learning

Many self-supervised methods have been proposed with the target of image anomaly detection. These methods often rely on the paradigm of data augmentation with predefined transformations such as flipping, cropping, and rotations. However, it is not straightforward to apply these techniques for non-image data, such as time series or tabular data, while the performance of the existing deep approaches has been under our expectation on tasks beyond images. In this work, we propose a novel active learning (AL) scheme that relied on neural autoregressive flows (NAF) for self-supervised anomaly detection, specifically on small-scale data. Unlike other generative models such as GANs or VAEs, flow-based models allow to explicitly learn the probability density and thus can assign accurate likelihoods to normal data which makes it usable to detect anomalies. The proposed NAF-AL method is achieved by efficiently generating random samples from latent space and transforming them into feature space along with likelihoods via invertible mapping. The samples with lower likelihoods are selected and further checked by outlier detection using Mahalanobis distance. The augmented samples incorporating with normal samples are used for training a better detector so as to approach decision boundaries. Compared with random transformations, NAF-AL can be interpreted as a likelihood-oriented data augmentation that is more efficient and robust. Extensive experiments show that our approach outperforms existing baselines on multiple time series and tabular datasets, and a real-world application in advanced manufacturing, with significant improvement on anomaly detection accuracy and robustness over the state-of-the-art.

Zhang, Jiaxin↗

Variability in Performance of a Machine Learning Seismicity Catalog: Central Italy, 2016–2017

Machine learning (ML) catalogs contain many more earthquakes than routine catalogs, but their performance in phase picking and earthquake detection has not been fully evaluated. We develop station‐level detection probabilities using logistic regression and combine them across a seismic network to compute spatial magnitude‐of‐completeness fields. We apply this approach to two catalogs from the 2016–2017 Central Italy sequence that were constructed from the same seismic network, one routine and one ML‐based. At the station level, the ML picker increases detection sensitivity by identifying smaller magnitude events and detecting earthquakes at greater distances. Spatially, the magnitude of completeness decreases substantially, with median values shifting from 1.6 to 0.5 for P waves and from 1.7 to 0.5 for S waves. However, the ML catalog also shows greater variability in station‐level performance than the routine catalog. These results demonstrate that ML‐based improvements in detectability are widespread but spatially nonuniform, highlighting their benefits, their limitations, and the potential for further improvements.

15 GEOTHERMAL ENERGY↗

How to sustain a CO 2 -thermosiphon in a partially saturated geothermal reservoir: Lessons learned from field experiment and numerical modeling

CO 2 has been proposed as a working fluid for geothermal energy production because of its ability to establish a self-sustaining CO 2 thermosiphon, taking advantage of the strong temperature dependence of CO 2 density. To test the concept of CO 2 heat extraction, in January 2015 a CO 2 thermosiphon was operated at the SECARB Cranfield Site, Cranfield, Mississippi, where a brine-saturated sand at a depth of 3.2 km has been under near continuous CO 2 flood since December 2009 as part of a U.S. Department of Energy demonstration of CO 2 sequestration, resulting in a partially saturated reservoir surrounding a well pair. The lateral distance between the producer and injector was 112 m at reservoir depth, a distance considered pre-commercial in scale, but great enough that thermal breakthrough was still not significant after several years of injection. Instead of producing power with a turbine, heat was extracted heat from recirculated fluid using a heat exchanger and portable chiller. The well field and surface equipment were instrumented to compare field observations with predicted responses from numerical models. Thermosiphon flow could be initiated by venting, but thereafter flow rate steadily declined, indicating that the thermosiphon was not sustainable. To model the system, the capability of T2Well, a fully coupled wellbore/reservoir numerical simulator, was expanded to enable simulation of the entire loop of fluid circulation in the fully-coupled system consisting of the injection/production wells, the reservoir, and the surface devices (heat exchanger, flow-rate regulator etc.). Combined with the newly developed TOUGH2 equation of state module called EOS7CMA, the enhanced T2 Well was used prior to the field experiment to simulate the circulation of a CO 2 -H 2 O-CH 4 mixture in a model geothermal system patterned after the Cranfield demonstration test. The model predicted that a sustainable thermosiphon could be achieved. After the field thermosiphon did not achieve the pre-test prediction of flow rates and thermosiphon sustainability, the numerical model was modified to improve realism and calibrate certain processes; it was then able to reproduce the major phenomena observed in the field. In a series of sensitivity studies, many factors were found that could potentially contribute to the failing of a sustainable thermosiphon. These factors could be categorized as two types: factors that increase the resistance to flow and factors that increase heat loss of the working fluid. In conclusion, the lessons learned can be applied to both future modeling and to achieving CO 2 -based geothermal reservoir exploitation.

15 GEOTHERMAL ENERGY↗

A Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter (PICC) Tip Detection

A peripherally inserted central catheter (PICC) is a thin catheter that is inserted via arm veins and threaded near the heart, providing intravenous access. The final catheter tip position is always confirmed on a chest radiograph (CXR) immediately after insertion since malpositioned PICCs can cause potentially life-threatening complications. Although radiologists interpret PICC tip location with high accuracy, delays in interpretation can be significant. In this study, we proposed a fully-automated, deep-learning system with a cascading segmentation AI system containing two fully convolutional neural networks for detecting a PICC line and its tip location. A preprocessing module performed image quality and dimension normalization, and a post-processing module found the PICC tip accurately by pruning false positives. Our best model, trained on 400 training cases and selectively tuned on 50 validation cases, obtained absolute distances from ground truth with a mean of 3.10 mm, a standard deviation of 2.03 mm, and a root mean squares error (RMSE) of 3.71 mm on 150 held-out test cases. This system could help speed confirmation of PICC position and further be generalized to include other types of vascular access and therapeutic support devices.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Nuclear reactor safety and efficiency can be enhanced through the development of accurate and fast methods for prediction of reactor transient (RT) states. Physics informed neural networks (PINNs) leverage deep learning methods to provide an alternative approach to RT modeling. Applications of PINNs in monitoring of RTs for operator support requires near real-time model performance. However, as with all machine learning models, development of a PINN involves time-consuming model training. Here, we show that a transfer learning (TL-PINN) approach achieves significant performance gain, as measured by reduction of the number of iterations for model training. Using point kinetic equations (PKEs) model with six neutron precursor groups, constructed with experimental parameters of the Purdue University Reactor One (PUR-1) research reactor, we generated different RTs with experimentally relevant range of variables. The RTs were characterized using Hausdorff and Fréchet distance. We have demonstrated that pre-training TL-PINN on one RT results in up to two orders of magnitude acceleration in prediction of a different RT. The mean error for conventional PINN and TL-PINN models prediction of neutron densities is smaller than 1%. We have developed a correlation between TL-PINN performance acceleration and similarity measure of RTs, which can be used as a guide for application of TL-PINNs.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗