Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Learning with errors”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

A safe reinforcement learning algorithm for supervisory control of power plants

Traditional control theory-based methods require tailored engineering for each system and constant fine-tuning. In power plant control, one often needs to obtain a precise representation of the system dynamics and carefully design the control scheme accordingly. Model-free Reinforcement learning (RL) has emerged as a promising solution for control tasks due to its ability to learn from trial-and-error interactions with the environment. It eliminates the need for explicitly modeling the environment’s dynamics, which is potentially inaccurate. However, the direct imposition of state constraints in power plant control raises challenges for standard RL methods. To address this, we propose a chance-constrained RL algorithm based on Proximal Policy Optimization for supervisory control. Our method employs Lagrangian relaxation to convert the constrained optimization problem into an unconstrained objective, where trainable Lagrange multipliers enforce the state constraints. In conclusion, our approach achieves the smallest distance of violation and violation rate in a load-follow maneuver for an advanced Nuclear Power Plant design.

constrained optimization↗
DOE OSTI2024

Online Learning of Effective Turbine Wind Speed in Wind Farms

To develop better wind farm controllers that can meet more complex objectives, methods of modeling the wind turbine wakes at low computational expense are needed. Gaussian process (GP) regression offers a computationally inexpensive framework for learning complex functions from noisy measurements with very few datapoints. In this work, an online learning approach is presented to learn the rotor-averaged wind velocity at downstream wind turbines with GPs, using the available datastream of wind field measurements and wind turbine control set-points. This framework can readily be integrated into model-based controls methods because the model a) is updated online at low computational expense, b) assumes a mathematically favorable Gaussian form, and c) explicitly quantifies the stochastic nature of the wake field so that the trade-off between exploration and exploitation, and the uncertainty in the prediction, can be utilized. We show that a GP-learned model can match true values with errors within 0.5% on average, with as few as 5 training data points.

Gaussian process↗
DOE OSTI2025

A physics-constrained neural ordinary differential equations approach for robust learning of stiff chemical kinetics

The high computational cost associated with solving for detailed chemistry poses a significant challenge for predictive computational fluid dynamics (CFD) simulations of turbulent reacting flows. While deep learning techniques have been explored to develop faster surrogate models, they often fail to integrate reliably with CFD solvers. This instability arises because traditional deep learning approaches optimize for training error without ensuring compatibility with ordinary differential equation (ODE) solvers, resulting in accumulation of errors over time. Recently, neuralODE (NODE) based approaches have been shown to be a promising technique to emulate and accelerate detailed chemistry computations. Here, in the present work, we extend this NODE framework for stiff chemical kinetics by incorporating mass conservation constraints directly into the loss function during training. This ensures that the total mass as well as the individual elemental species masses are conserved in an a-posteriori manner. Proof-of-concept studies are performed with the novel physics-constrained NODE (PC-NODE) approach for homogeneous autoignition of hydrogen-air mixture over a range of composition and thermodynamic conditions. It is demonstrated that the PC-NODE framework not only improves the physical consistency of the resulting data-driven model with respect to mass conservation criteria, but also improves training efficiency. PC-NODE is shown to achieve 2–100× speedup relative to the hydrogen-air detailed chemical mechanism depending on the type of the ODE solver (implicit or explicit) used during autoregressive inference tests. Lastly, a-posteriori studies are performed wherein the trained PC-NODE model is coupled with a CFD solver. It is shown that higher accuracy is achieved with PC-NODE relative to the purely data-driven NODE approach. Moreover, PC-NODE also exhibits robustness and generalizability to unseen initial conditions from within (interpolative capability) as well as outside (extrapolative capability) the training regime.

computational combustion↗
DOE OSTI2024

Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics

Traditional data-driven deep learning models often struggle with high training costs, error accumulation, and poor generalizability in complex physical processes. Physics-informed deep learning (PiDL) addresses these challenges by incorporating physical principles into the model. Most PiDL approaches regularize training by embedding governing equations into the loss function, yet this depends heavily on extensive hyperparameter tuning to weigh each loss term. To this end, we propose to leverage physics prior knowledge by “baking” the discretized governing equations into the neural network architecture via the connection between the partial differential equations (PDE) operators and network structures, resulting in a PDE-preserved neural network (PPNN). This method, embedding discretized PDEs through convolutional residual networks in a multi-resolution setting, largely improves the generalizability and long-term prediction accuracy, outperforming conventional black-box models. The effectiveness and merit of the proposed methods have been demonstrated across various spatiotemporal dynamical systems governed by spatiotemporal PDEs, including reaction-diffusion, Burgers’, and Navier-Stokes equations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Deep nonparametric estimation of intrinsic data structures by chart autoencoders: Generalization error and robustness

Autoencoders have demonstrated remarkable success in learning low-dimensional latent features of high-dimensional data across various applications. Assuming that data are sampled near a low-dimensional manifold, we employ chart autoencoders, which encode data into low-dimensional latent features on a collection of charts, preserving the topology and geometry of the data manifold. Our paper establishes statistical guarantees on the generalization error of chart autoencoders, and we demonstrate their denoising capabilities by considering n noisy training samples, along with their noise-free counterparts, on a d-dimensional manifold. By training autoencoders, we show that chart autoencoders can effectively denoise the input data with normal noise. We prove that, under proper network architectures, chart autoencoders achieve a squared generalization error in the order of n–$\frac{2}{d+2}$log 4 n, which depends on the intrinsic dimension of the manifold and only weakly depends on the ambient dimension and noise level. We further extend our theory on data with noise containing both normal and tangential components, where chart autoencoders still exhibit a denoising effect for the normal component. As a special case, our theory also applies to classical autoencoders, as long as the data manifold has a global parametrization. Furthermore, our results provide a solid theoretical foundation for the effectiveness of autoencoders, which is further validated through several numerical experiments.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Implementation and (Inverse Modified) Error Analysis for Implicitly Templated ODE-Nets

We focus on learning unknown dynamics from data using ODE-nets templated on implicit numerical initial value problem solvers. First, we perform inverse modified error analysis of the ODE-nets using unrolled implicit schemes for ease of interpretation. It is shown that training an ODE-net using an unrolled implicit scheme returns a close approximation of an inverse modified differential equation (IMDE). In addition, we establish a theoretical basis for hyperparameter selection when training such ODE-nets, whereas current strategies usually treat numerical integration of ODE-nets as a black box. We thus formulate an adaptive algorithm which monitors the level of error and adapts the number of (unrolled) implicit solution iterations during the training process, so that the error of the unrolled approximation is less than the current learning loss. This helps accelerate training while maintaining accuracy. Several numerical experiments are performed to demonstrate the advantages of the proposed algorithm compared to nonadaptive unrollings and validate the theoretical analysis. Here, we also note that this approach naturally allows for incorporating partially known physical terms in the equations, giving rise to what is termed “gray box” identification.

ODE-nets↗
DOE OSTI2026

Data driven investigation to understand the influence of total solids on biological biogas upgrading

In situ biogas upgrading achieves CO 2 conversion to CH 4 via hydrogenotrophic methanogenesis; however, gas-liquid mass transfer constraints limit the upgrading performance. Recognizing that optimization studies often underrepresent the effects of total solids (TS) and organic loading rate (OLR), this study undertook a holistic, statistics driven assessment of operating conditions for in situ H 2 assisted biogas upgrading, centering the analysis on TS and OLR. A dataset of 31 studies was compiled and comprised 99 observations. A rigorous analytical framework was employed, combining data standardization, fixed- and random-effects (REML) weighted regressions with cluster-robust errors, stratified analyses, and machine learning. Mixed-effects meta regression indicated that TS was the main factor explaining differences of methane fraction (CH 4 %) when considering the between studies heterogeneity. Focusing on a near-stoichiometric subset (H 2 /CO 2 ≈ 4:1), TS remained significant. Stratified results showed a stronger negative relationship between TS and CH 4 % in UASB reactors than in CSTRs, with a negative effect under mesophilic conditions and no significant effect under thermophilic conditions. A Random Forest model corroborated the statistical findings, consistently ranking H 2 /CO 2 ratio, OLR, TS, and hydrogen injection rate (HIR) as the most influential predictors. These findings delineate trends across increasing TS levels, particularly between 1% and 10%, and provide preliminary insights for TS above 15% in in situ biogas upgrading. They further provide insights for the influence of TS by reactor type and temperature, thereby advancing the evidence base for implementing biological CO 2 conversion to CH 4 in practice.

In situ biogas upgrading↗
DOE OSTI2025

Curiosity driven exploration to optimize structure–property learning in microscopy

Rapidly determining structure–property correlations in materials is an important challenge in better understanding fundamental mechanisms and greatly assists in materials design. In microscopy, imaging data provides a direct measurement of the local structure, while spectroscopic measurements provide relevant functional property information. Deep kernel active learning approaches have been utilized to rapidly map local structure to functional properties in microscopy experiments, but are computationally expensive for multi-dimensional and correlated output spaces. Here, we present an alternative lightweight curiosity algorithm which actively samples regions with unexplored structure–property relations, utilizing a deep-learning based surrogate model for error prediction. We show that the algorithm outperforms random sampling for predicting properties from structures, and provides a convenient tool for efficient mapping of structure–property relationships in materials science.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Chaotic neural dynamics facilitate probabilistic computations through sampling

Cortical neurons exhibit highly variable responses over trials and time. Theoretical works posit that this variability arises potentially from chaotic network dynamics of recurrently connected neurons. Here, we demonstrate that chaotic neural dynamics, formed through synaptic learning, allow networks to perform sensory cue integration in a sampling-based implementation. We show that the emergent chaotic dynamics provide neural substrates for generating samples not only of a static variable but also of a dynamical trajectory, where generic recurrent networks acquire these abilities with a biologically plausible learning rule through trial and error. Furthermore, the networks generalize their experience in the stimulus-evoked samples to the inference without partial or all sensory information, which suggests a computational role of spontaneous activity as a representation of the priors as well as a tractable biological computation for marginal distributions. These findings suggest that chaotic neural dynamics may serve for the brain function as a Bayesian generative model.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Analysis of Lobe Power Calculator and Indication System with Physics and Cycle Based Models

The Advanced Test Reactor (ATR) at INL measures reactor power through various methods, two of them being thermal and Nitrogen-16 (N-16) activity. Water power calculator (WPC) is a thermal power system that measures flow and temperature to determine thermal quadrant powers. The N-16 system utilizes a beta detector that outputs Nitrogen activation levels to calculate lobe power through an algorithm called Lobe Power Calculation and Indication system (LPCIS). The LPCIS utilizes the N-16 system and the WPC system to determine reactor core power levels. The WPC provides accurate calculations of quadrant and total reactor thermal power. With the use of WPC measurements, thermal-to-N-16 (T2N) power ratios are produced to determine if the two indication systems agree on core power. Relative magnitude equations are used to utilize N-16 coefficients and multipliers to improve the indications of the LPCIS. These correct indications are crucial for maintaining safety limits because operators rely on this information for decision making. Currently, the LPCIS system uses linear equations and matrices to calculate lobe power through multipliers and coefficients. Advancements in technology and system upgrades have increased the accuracy of power readings by making the system more dynamic. The new proposed coefficient and multiplier method implements a cycle specific and physics-based model intended for changing coefficients during operation. This calibration experiment focused on power splitting, outer shim and neck shim, as well as fuel burning into the reactor digital acquisition system (RDAS) weighting factors. Results demonstrated that the physics learning method yields a smaller error margin inside of the desired power range for the data set 166-A. Continuing to improve the physics-based model will help improve the power accuracy of the LPCIS system.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2026

NMF-Based Anomaly Detection in CMS 2D Tracking Occupancy Histograms

The CMS experiment relies on Data Quality Monitoring (DQM) to ensure that recorded collision data are suitable for physics analysis. During LHC Run 3, each run contains many lumisections and tracking monitoring elements, making offline inspection challenging, especially for localized detector effects that may appear only for short periods of time. This poster presents an unsupervised machine-learning approach to identify anomalous lumisections in CMS tracking occupancy histograms using Non-Negative Matrix Factorization (NMF). The workflow uses offline CMS DQMIO tracking histograms retrieved with the CMS DIALS API and organized as two-dimensional occupancy maps for each lumisection. After selecting stable lumisections, the occupancy maps are normalized and arranged into a non-negative data matrix. The NMF model learns a compact set of basis patterns describing normal tracking occupancy. Each lumisection is then reconstructed from these learned components, and the reconstruction error is used as an anomaly score. Large residuals indicate occupancy patterns that deviate from normal detector behavior and are flagged for further inspection. This NMF-based approach provides a fast and interpretable way to flag lumisections whose tracking occupancy patterns differ from normal detector behavior. Preliminary studies show sensitivity to known tracking anomalies, and ongoing work is focused on validating the method across additional Run 3 Pixel and Strip detector issues.

Rodríguez Ramos, Iliomar [Puerto Rico U., Mayaguez↗
DOE OSTI2025

An Experiment with LLMs as Database Design Tutors: Persistent Equity and Fairness Challenges in Online Learning

As large language models (LLMs) continue to evolve, their capacity to replace humans as their surrogates is also improving. As increasing numbers of intelligent tutoring systems (ITSs) are embracing the integration of LLMs for digital tutoring, questions are arising as to how effective they are and if their hallucinatory behaviors diminish their perceived advantages. One critical question that is seldom asked if the availability, plurality, and relative weaknesses in the reasoning process of LLMs are contributing to the much discussed digital divide and equity and fairness in online learning. In this paper, we present an experiment with database design theory assignments and demonstrate that while their capacity to reason logically is improving, LLMs are still prone to serious errors. We demonstrate that in online learning and in the absence of a human instructor, LLMs could introduce inequity in the form of “wrongful” tutoring that could be devastatingly harmful for learners, which we call ignorant bias, in increasingly popular digital learning. We also show that significant challenges remain for STEM subjects, especially for subjects for which sound and free online tutoring systems exist. Based on the set of use cases, we formulate a possible direction for an effective ITS for online database learning classes of the future.

Jamil, Hasan M. (ORCID:0000000231243780)↗
NASA NTRS2012

Improved Snow Mapping Accuracy with Revised MODIS Snow Algorithm

The MODIS snow cover products have been used in over 225 published studies. From those reports, and our ongoing analysis, we have learned about the accuracy and errors in the snow products. Revisions have been made in the algorithms to improve the accuracy of snow cover detection in Collection 6 (C6), the next processing/reprocessing of the MODIS data archive planned to start in September 2012. Our objective in the C6 revision of the MODIS snow-cover algorithms and products is to maximize the capability to detect snow cover while minimizing snow detection errors of commission and omission. While the basic snow detection algorithm will not change, new screens will be applied to alleviate snow detection commission and omission errors, and only the fractional snow cover (FSC) will be output (the binary snow cover area (SCA) map will no longer be included).

Riggs, George↗
DOE OSTI2021

Analysis of Lobe Power Calculator and Indication System with Physics and Cycle Based Models

The Advanced Test Reactor (ATR) at INL measures reactor power through two methods, thermal and Nitrogen-16 (N-16) activity. Water power calculator (WPC) is a thermal power system that measures flow and temperature to determine the thermal quadrant powers. The N-16 system utilizes a beta chamber detector that outputs reactivity levels to calculate lobe power through an algorithm called lobe power calculation and indication system (LPCIS). The LPCIS utilizes the N-16 system and the WPC system to determine reactor core power levels. The WPC provides accurate calculations of quadrant and total reactor thermal power. With the use of WPC measurements, thermal-to-N-16 (T2N) power ratios are produced to determine if the two indication systems agree on core power. Relative magnitude equations are used to utilize N-16 coefficients and multipliers to improve the indications of the LPCIS. This is crucial for maintaining safety limits. Currently, the LPCIS system uses linear equations and matrices to calculate lobe power through multipliers and coefficients. Advancements in technology and system upgrades have assisted system engineers at INL with the objective to reach a more dynamic system. In return the system demonstrates an increase in the accuracy of power reading while maintaining safety margins. The new proposed coefficient and multiple method implements a cycle specific, and a physics-based model intended for changing coefficients during operation. This calibration experiment focuses on power splitting, outer shim and neck shim, as well as fuel burning into the RDAS weighting factors. Results show that the Physics learning method yields a smaller error margin inside of the desired power range for the data set 166-A. This proved true for both constrained and unconstrained testing. This is most likely due to the physics data fitting approach that resulted in favorable coefficients and multipliers for 166-A. Continuing to improve the physics-based model will help improve the power accuracy of the LPCIS system.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2024

An Approach to Dynamic Human Reliability Analysis and Its Data Collection Framework

Human reliability analysis (HRA) is a method for evaluating human errors in a variety of complex systems such as nuclear power plants, military systems, aircraft, and chemical plants. Most HRA methods currently used by regulatory institutes or utilities are called static HRA and are carried out by simple worksheets or simple calculators. To date, there are many unsolved or intrinsic challenges in static HRA. For example, existing static HRA does not realistically model and evaluate human actions as they would be performed at actual systems. There is no method with HRA to objectively estimate the time required for human actions despite being essential to HRA processes. In addition, many HRA methods still rely on a dataset generated prior to the 1980s, from unrelated industry experience or simply from expert judgment. Accordingly, this study attempted to research how to overcome the challenges of existing HRA via dynamic risk assessment (a.k.a., simulation-based or computation-based risk assessment) techniques. First, this study developed a dynamic HRA method, named as PRocedure-based Investigation Method of EMRALD Risk Assessment – HRA (PRIMERA-HRA). The PRIMERA-HRA mainly concentrates on providing HRA analysts with specific guidelines on how to reasonably model human actions, assign human reliability data and evaluate output of simulation within a dynamic probabilistic risk assessment tool, called as Event Modeling Risk Assessment using Linked Diagrams (EMRALD). Second, this study also developed a module for performance shaping factors (i.e., the key concept in HRA quantification) applicable to dynamic HRA, then implemented it based on PRIMERA-HRA within the EMRALD tool. Third, this study developed an HRA data collection framework to support dynamic HRA, called as Simplified Human Error Experimental Program (SHEEP). Originally, the SHEEP study aimed to support static HRA and its data collection, but recently extended the scope to the new technologies such as dynamic HRA or HRA for advanced reactors. SHEEP focuses on the use of data collected from simplified simulators to complement—but not replace—data collection studies using full-scope simulators and actual operators. To date, many experiments were conducted under the SHEEP framework. Multiple analyses, such as human performance analysis, human error analysis, task complexity analysis, learning effect analysis and time distribution analysis, were also carried out using the collected data. Then, based on the major insights, an approach to inferring full-scope data based on simplified simulator data was proposed. The PRIMERA-HRA and SHEEP research are expected to evaluate human actions more realistically than existing static HRA, provide an opportunity to collect more HRA data with reasonable cost and labor, then contribute to enhance the quality of HRA.

99 - GENERAL AND MISCELLANEOUS↗
DOE OSTI2024

An Approach to Dynamic Human Reliability Analysis and Its Data Collection Framework

Human reliability analysis (HRA) is a method for evaluating human errors in a variety of complex systems such as nuclear power plants, military systems, aircraft, and chemical plants. Most HRA methods currently used by regulatory institutes or utilities are called static HRA and are carried out by simple worksheets or simple calculators. To date, there are many unsolved or intrinsic challenges in static HRA. For example, existing static HRA does not realistically model and evaluate human actions as they would be performed at actual systems. There is no method with HRA to objectively estimate the time required for human actions despite being essential to HRA processes. In addition, many HRA methods still rely on a dataset generated prior to the 1980s, from unrelated industry experience or simply from expert judgment. Accordingly, this study attempted to research how to overcome the challenges of existing HRA via dynamic risk assessment (a.k.a., simulation-based or computation-based risk assessment) techniques. First, this study developed a dynamic HRA method, named as PRocedure-based Investigation Method of EMRALD Risk Assessment – HRA (PRIMERA-HRA). The PRIMERA-HRA mainly concentrates on providing HRA analysts with specific guidelines on how to reasonably model human actions, assign human reliability data and evaluate output of simulation within a dynamic probabilistic risk assessment tool, called as Event Modeling Risk Assessment using Linked Diagrams (EMRALD). Second, this study also developed a module for performance shaping factors (i.e., the key concept in HRA quantification) applicable to dynamic HRA, then implemented it based on PRIMERA-HRA within the EMRALD tool. Third, this study developed an HRA data collection framework to support dynamic HRA, called as Simplified Human Error Experimental Program (SHEEP). Originally, the SHEEP study aimed to support static HRA and its data collection, but recently extended the scope to the new technologies such as dynamic HRA or HRA for advanced reactors. SHEEP focuses on the use of data collected from simplified simulators to complement—but not replace—data collection studies using full-scope simulators and actual operators. To date, many experiments were conducted under the SHEEP framework. Multiple analyses, such as human performance analysis, human error analysis, task complexity analysis, learning effect analysis and time distribution analysis, were also carried out using the collected data. Then, based on the major insights, an approach to inferring full-scope data based on simplified simulator data was proposed. The PRIMERA-HRA and SHEEP research are expected to evaluate human actions more realistically than existing static HRA, provide an opportunity to collect more HRA data with reasonable cost and labor, then contribute to enhance the quality of HRA.

99 - GENERAL AND MISCELLANEOUS↗
DOE OSTI2022

Using probabilistic solar power forecasts to inform flexible ramp product procurement for the California ISO

How can independent system operators (ISOs) take advantage of probabilistic solar forecasts to lower generation costs and improve reliability of power systems? We discuss one three-step approach for doing so, focusing on how such forecasts might help the California Independent System Operator (CAISO) prepare unexpected net load ramps, where net load equals gross demand minus wind and solar production. First, we enhance an existing solar forecasting system to provide well-calibrated hours-ahead probabilistic forecasts. We then relate the degree of uncertainty reflected in the forecasted prediction intervals (independent variables) to error distributions for net load ramp forecasts for the CAISO real-time market (dependent variable) using machine learning and quantile regression. Projected ramp forecast errors conditioned on solar uncertainty are translated into flexible ramp requirements that therefore reflect real-time meteorological and solar conditions, improving on typical ISO procedures. Detailed descriptions are provided on the quantile regression and kth-nearest neighbor categorization methods for accomplishing that translation. Finally, a multiple time-scale look-ahead market simulation model is applied to a 118-bus IEEE Reliability Test System, modified to represent the CAISO generation mix and demand distributions. The model runs quantify how solar-conditioned ramp requirements can, first, decrease operating costs by reducing requirements compared to often conservative unconditional methods and, second, decrease generation scarcity events and consequently improve reliability by increasing flexibility requirements at times when unconditional forecast-based requirements understate actual ramp uncertainty. Solar-conditioned ramp requirements are found to reduce generation operating costs by about 2% for the test system (which would be equivalent to over $\$100$ million per year for a CAISO-size system).

14 SOLAR ENERGY↗