SEARCH · Engineering Papers
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.
Informatics-driven approaches, such as machine learning and sequential experimental design, have shown the potential to drastically impact next-generation materials discovery and design.
The intersection between classical data assimilation methods and novel machine learning techniques has attracted significant interest in recent years. Here, we explore another promising solution in which diffusion models are used to formulate a robust nonlinear ensemble filter for sequential data assimilation. Unlike standard machine learning methods, the proposed ensemble score filter (EnSF) is completely training free and can efficiently generate a set of analysis ensemble members. Here, in this study, we apply the EnSF to a surface quasigeostrophic model and compare its performance against the popular local ensemble transform Kalman filter (LETKF), which makes Gaussian assumptions in the analysis step. Numerical tests demonstrate that EnSF maintains stable performance in the absence of localization and for a variety of experimental settings. We find that while LETKF maintains optimal performance in the case of linear observations of the entire state and a perfect model, EnSF shows improvements over LETKF when nonlinear observations are assimilated and the system is subject to unexpected model errors. A spectral decomposition of the analysis results in this nonlinear observation regime shows that the largest improvements over LETKF occur at large scales (small wavenumbers), where LETKF lacks sufficient ensemble spread. Overall, this initial application of EnSF to a geophysical model of intermediate complexity motivates further development of the algorithm for more realistic problems.
Abstract Additive manufacturing (AM) has gained notoriety for offering advantages over traditional manufacturing methods, such as increased design complexity and flexibility. However, it has not found widespread use beyond rapid prototyping. One hindrance to the acceptance of AM processes in industry is the time and cost of fabrication per component. While metal AM by itself can be inexpensive, extra manufacturing steps in the form of subtractive manufacturing (SM) may need to be performed to reach final part tolerances, leading to hybrid additive-subtractive manufacturing (HASM) of a part, which increases time and cost. A potential area to reduce cost is through increasing the efficiency of the HASM process by conducting additive and subtractive manufacturing simultaneously. Usually, HASM is performed in a process where AM is completed in one machine or cell and transferred to another machine or cell for SM in a sequential assembly line process. This efficiency decreases part cost, but high aspect ratio parts or parts with internal geometry that require interleaved additive deposition and machining cannot be produced. One unexplored solution to simultaneous HASM that allows for interleaved operations is to operate the deposition head and machining spindle concurrently within the same machine envelope, known as concurrent HASM (CHASM). In this type of process, both AM and SM occur simultaneously on a batch of small parts or a single large part, maintaining a high efficiency without sacrificing the full range of complex geometries that AM allows for. A potential approach to the single-machine method could be to combine a robot and mill within the same envelope. A challenge to this approach, however, is control of both systems. Most machine controllers have limited external communication or, if a robot has been integrated, only offer movement of either the robot or mill at any given time. As a result, systems must pause either the AM or SM process to switch between them rather than working simultaneously. The present work investigates the positional accuracy of such a CHASM system comprised of a robotic arm and a 3-axis mill. Open-loop tests with limited communication between machines are performed on the system to verify positional error during concurrent robot-mill movements. Under certain conditions, it is demonstrated that position error can stay within 2 mm for the duration of a single layer; however, these tests show that, generally, the open-loop positioning performance of the system is inadequate for CHASM without part-specific hand-tuning of parameters. Based on these results, a set of requirements for successful robot-CNC CHASM is proposed for future integrations.
The primary intent of PAR-FIT (Particle Filter based Inference Testing) is to provide hard inductive evidence that a machine learning model is capable and proven for an individual test input. By examining training data used to form the underlying model functional correlation, an estimate of the reliability that a model will make the correct prediction can be made. The Sequential Probability Ratio Test is used to derive a qualitative evaluation for reliability based on hypothesis testing. The PAR-FIT framework achieves this by implementing a particle filter and the sequential probability ratio test algorithms on the machine learning model training data to determine relevancy of new individual test samples to the training dataset. The kernel function evaluates the local proximity and density of training data used to derive a prediction outcome. Particles are used to probabilistically determine which training data to evaluate for proximity. For test samples that are within a close proximity to and surrounded by multiple training data points, the evaluated reliability of the prediction is high. For test samples that are anomalies not represented by the training dataset, in low density data clusters, or are far from existing data points, the evaluated reliability is low as insufficient training evidence exists to suggest the model is capable of making the correct prediction. Sequential Probability Ratio Test is further used to determine when a hypothesis on whether a signal can be rejected or accepted for use. The ratio test collects sequence information from the particle filter to test whether the signal is anomalous or normal via hypothesis testing of the underlying distributions.
Inverter reliability remains one of the most persistent challenges limiting the performance, availability, and economic viability of utility‑scale photovoltaic (PV) plants. Industry data consistently show that inverters account for the highest share of corrective maintenance events and unplanned outages across PV fleets. These failures result in energy losses, increased O&M costs, and reduced confidence in long‑term solar asset performance. Motivated by these challenges, this project—Extending Component Lifetime and Improving Inverter Reliability (ECLAIIR)—was undertaken to systematically investigate inverter degradation and failure mechanisms, develop predictive maintenance capabilities, and establish data‑driven pathways to improve service life and reduce the Levelized Cost of Energy (LCOE) for large‑scale PV systems. The primary goal of the project was to identify pre‑failure signatures in string inverters using both lab‑based accelerated lifetime testing and field‑based data and to develop predictive maintenance algorithms that can anticipate inverter faults before they occur. Through collaboration with inverter testing laboratory, solar PV plant owner, and failure‑analysis experts, the project advanced the technical understanding of inverter reliability. By instrumenting inverters with thermistors, humidity sensors, power‑quality meters, and acoustic sensors, the research established how multiple sensing modalities can reliably detect deviations from normal behavior hours to days before failure. These findings substantially enhance scientific understanding of inverter failure kinetics and provide the PV industry with the most comprehensive cross‑OEM characterization of early‑stage failure indicators reported to date. Technically, the project demonstrated the effectiveness of predictive maintenance by developing and validating the PreDICT (Predictive Diagnostics of PV Inverters Using Condition Monitoring and Trend Analysis) framework—a multi‑layer diagnostic architecture combining peer‑to‑peer analytics, historical trend modeling, and advanced machine‑learning techniques such as the Sequential Conditional Variational Autoencoder (SCVAE). This predictive model achieved more than 90% accuracy in detecting pre‑failure conditions and provided up to four days of lead time before inverter failure in field scenarios. Economically, the project’s LCOE analysis showed that predictive maintenance can reduce lifetime energy losses and minimize corrective maintenance interventions. Modeling indicated that, depending on inverter failure rates and replacement timelines, predictive maintenance can significantly reduce LCOE impacts associated with inverter downtime: from as high as 19.4% under conventional maintenance strategies to 0.1%–10.17% when predictive analytics are adopted. These results confirm that predictive maintenance is both technically feasible and economically advantageous for utilities and plant operators. The project’s findings also have broad public benefit. By improving inverter reliability and reducing downtime, predictive maintenance directly increases electricity generation from existing PV assets. Enhanced reliability lowers operational costs for utilities, which can translate over time into lower energy costs for consumers. Furthermore, the project’s technical publications, conference presentations, and industry workshops ensure that knowledge gained is shared broadly across the solar industry, supporting workforce development and enabling utilities of all sizes to adopt modern asset‑health monitoring practices. The retrofitting case study and service‑life prediction framework further support informed decision‑making for aging PV fleets, helping operators extend system life and reduce electronic waste. In summary, the ECLAIIR project significantly advanced the state of knowledge on inverter degradation, demonstrated the technical and economic value of predictive maintenance, and delivered actionable tools and insights that support more reliable, cost‑effective, and sustainable PV plant operation. The outcomes of this project will continue to inform utility practices, guide inverter design improvements, and strengthen the long‑term performance of solar assets nationwide.
Entanglement is fundamental to quantum information science and technology, yet controlling and manipulating entanglement—so-called entanglement engineering—for arbitrary quantum systems remains a formidable challenge. There are two difficulties: the fragility of quantum entanglement and its experimental characterization. We develop a model-free deep reinforcement-learning (RL) approach to entanglement engineering, in which feedback control together with weak continuous measurement and partial state observation is exploited to generate and maintain desired entanglement. We employ quantum optomechanical systems with linear or nonlinear photon–phonon interactions to demonstrate the workings of our machine-learning-based entanglement engineering protocol. In particular, the RL agent sequentially interacts with one or multiple parallel quantum optomechanical environments, collects trajectories, and updates the policy to maximize the accumulated reward to create and stabilize quantum entanglement over an arbitrary amount of time. The machine-learning-based model-free control principle is applicable to the entanglement engineering of experimental quantum systems in general.
Traditional transmission line protection relies on predictable synchronous-based fault signatures, which frequently fail under the non-standard, current-limited fault characteristics of Inverter-Based Resources (IBRs). This study investigates how to achieve secure, communication-free fault isolation in IBR-dominated weak grids without relying on opaque, computationally heavy "black-box" machine learning algorithms. To address this, we propose a novel, standalone, and inherently interpretable data-driven protection framework. Unlike centralized methods requiring multi-terminal communication, this decentralized approach relies solely on local measurements using a hierarchical linear-kernel Support Vector Machine (SVM). The methodology decomposes the protection task into four sequential stages that mimic traditional protection elements: fault detection and fault direction identification, fault type classification, zone classification, and location estimation. This multi-stage architecture allows for specialized feature engineering at each stage, combining high computational efficiency with logic traceability. The framework's end-to-end performance was validated via C-code and PSCAD/EMTDC co-simulation, utilizing a real-world utility network and an OEM black-box IBR model. The proposed relay achieves 97.2% overall accuracy and provides a reliable trip decision within a 2.5-cycle window. The results confirm 100% accuracy in fundamental fault detection, reliable zone selectivity across low to moderate fault resistances, and robust security against non-fault transients, proving its immediate viability for integration into commercial numerical relays.
United States critical infrastructure faces new cyber threats from adversarial nation-state actors in the form of malware-free attacks. Traditional cybersecurity techniques use rules-based methods to identify indicators of compromise on networks, often missing these sophisticated attacks. Our approach leverages multiple state of the art machine learning models in a pipeline to identify abnormal network events through sequential analysis. We combine both device and packet-level information into individual events to characterize anomalous network actions. The model is trained and tested on real network traffic from the Idaho National Lab High Performance Computing (HPC) with greater than 98% precision. It is capable of flagging malicious tactics used by adversaries in malware-free attacks, severe changes to the network, and abnormal user activity by network devices.
Python scripts for physics informed machine learning for time-dependent problems. The code implements hard constrained sequential physics-informed neural networks (HCSPINNs) using JAX library. The details of the method and implementation can be found in the following paper: Roy, P., & Castonguay, S. (2024). Exact Enforcement of Temporal Continuity in Sequential Physics- Informed Neural Networks. arXiv preprint arXiv:2403.03223. (https://arxiv.org/abs/2403.03223)
Hydrogen bonded electrolytes that exhibit accelerated proton transport via sequential reactive hops have drawn interest for their promise in clean energy applications. Molecular dynamics simulations of these electrolytes offer the opportunity to uncover microscopic mechanistic details that could be used to design and tune the properties of candidate electrolyte technologies. However, accurately modeling the proton transfer reactions and transport properties that give rise to high charge conductivites in these electrolytes proves computationally challenging because of the need to perform lengthy condensed phase simulations, treating both the electronic and nuclear degrees of freedom quantum mechanically. In this paper, we demonstrate that such a modeling task can be efficiently achieved with the use of density functional theory (DFT)-trained machine learning potentials (MLP) to accelerate path integral molecular dynamics (PIMD) simulations. We highlight the practical utility of this approach by using it to benchmark how closely PIMD simulations employing different DFT exchange–correlation functionals reproduce the composition-dependent densities, diffusion coefficients, and electrical conductivities of mixtures consisting of imidazole and levulinic acid. Even with the speedup afforded by our MLPs, PIMD simulations remain quite expensive. Furthermore, in order to render PIMD more computationally tractable, we introduce and benchmark the accuracy of a ring polymer contraction approach that leverages a computationally efficient short-range MLP to accelerate our PIMD simulations by an additional factor of four.
We propose to address the challenge of the vast parameter space associated with perovskite manufacturing optimization, by developing a machine learning (ML)-assisted optimization framework for a scalable perovskite PV manufacturing tool. This framework will be interpretable, sequential, and rapidly adaptable to upgraded systems (e.g., via transfer learning). The tool is an open-air rapid spray plasma process (RSPP) of perovskite films, which has already been established at Stanford and is a unique platform to test and deploy the proposed ML-guided framework because the RSPP technique is able to conduct optimization experiments with a high throughput, and easily adjust a wide range of process variables.
Manual labeling for machine learning tasks such as image classification is tedious and labor-intensive; as a result, scientific datasets suitable for deep learning applications are scarce and limited. While data augmentation techniques have shown promise for extending image datasets, very little work has been done to understand the impact of combining multiple augmentation methods sequentially or the limits of their effectiveness when combined. Our work addresses this gap by examining how standard and combinatorial data augmentation affects the performance of machine learning models when trained on small datasets for label classification tasks. For our analysis, we generate single, double and quadruple-augmented datasets for a microscopy image classification task using six standard augmentation methods, and compare the resultant improvements observed in binary classification accuracy with three standard image classification models (DenseNet169, MobileNetV2, ResNet101V2). Our experiments show a non-monotonic relationship between the number of simultaneous augmentation methods and classification accuracy, indicating that there is a trade-off between the degree of augmentation and the model performance. These findings suggest that the optimal number of augmentation methods will vary by domain and use case. We also find that the order in which augmentation methods are applied to a limited dataset matters when combining augmentation schemes, with our use case showing performance differences up to 2.6% when the augmentation order is reversed for double-augmented datasets. Our work offers insights to the limits of data augmentation when working on image classification tasks with limited datasets.
A coloring of a graph is an assignment of colors to vertices such that no two neighboring vertices have the same color. The need for memory-efficient coloring algorithms is motivated by their application in computing clique partitions of graphs arising in quantum computations where the objective is to map a large set of Pauli strings into a compact set of unitaries. We present Picasso, a randomized memory-efficient iterative parallel graph coloring algorithm with theoretical sublinear space guarantees under practical assumptions. The parameters of our algorithm provide a trade-off between coloring quality and resource consumption. To assist the user, we also propose a machine learning model to predict the coloring algorithm’s parameters considering these trade-offs. We provide a sequential and a parallel implementation of the proposed algorithm. We perform an experimental evaluation on a 64-core AMD CPU equipped with 512 GB of memory and an Nvidia A100 GPU with 40GB of memory. For a small dataset where existing coloring algorithms can be executed within the 512 GB memory budget, we show up to 68× memory savings. On massive datasets we demonstrate that GPU-accelerated Picasso can process inputs with 49.5× more Pauli strings (vertex set in our graph) and 2,478× more edges than state-of-the-art parallel approaches.
A key step in almost all scientific endeavors is answering the question: Given this data I already collected, what new data do I expect will yield the most useful information toward my scientific objective? The area of (sequential) experimental design has long been investigating answers to this question, but in recent years techniques from the machine learning subfield of active learning are increasingly applied. Researchers need a simple software tool for active learning applied to experimental design that can easily integrate into their existing workflows. This computer code, Dial, provides a microservice in ORNL's INTERSECT ecosystem for active learning applied to experimental design. By being part of the INTERSECT ecosystem, Dial is simple to integrate into any INTERSECT-based workflow. Dial provides multiple backend options, where a backend is an implementation of a specific active learning method. Users can select the backend that performs best for their application. Developers can also add new backends as needed. At its core, Dial receives a set of pre-existing measurements and input parameter bounds and then recommends one or more new sets of parameters to measure. Dial also includes interfaces to other microservices in the INTERSECT ecosystem so that it can be incorporated into INTERSECT campaigns. Dial provides a simple, yet powerful interface to convert automated INTERSECT workflows into autonomous workflows that adapt based on the results that are obtained. A shared microservice for active learning prevents duplicated effort by each application team implementing its own adaptive design of experiments tool.
The performance and stability of perovskite solar cells are strongly governed by the crystallization behavior of their active layer. In two-step sequential deposition, early-stage film formation plays a decisive role in determining final phase purity and device quality. Guided by a data-driven analysis of nearly 39 000 devices in the FAIR perovskite database, we identified solvent-mediated quenching and thermal processing as key variables affecting power conversion efficiency (PCE), particularly in two-step fabrication. Here, to investigate these effects in real time, we designed and implemented a custom-built, temperature-controlled spin-coating system, enabling precise thermal modulation during precursor deposition. Using this platform, we performed in situ GIWAXS measurements to study the crystallization dynamics of FA 0.5 MA 0.5 PbI 3 films over a temperature range of 30°C–90°C. Our results reveal a non-monotonic relationship between spin-coating temperature and α-phase formation, governed by the interplay between precursor interdiffusion, PbI 2 crystallinity, and δ-phase suppression. The custom thermal control enabled us to isolate and quantify these competing effects during the earliest stages of film formation, providing mechanistic insight into how spin-coating temperature governs both phase purity and kinetic pathways in two-step perovskite systems. Temperature-dependent SEM and photovoltaic device measurements further demonstrate that early-stage crystallization pathways directly translate into differences in morphology, charge-transport continuity, and device performance. These findings inform targeted strategies for optimizing deposition protocols to balance rapid nucleation, phase stability, and device performance.
Recent advances in differentiable modeling, a genre of physics-informed machine learning that trains neural networks (NNs) together with process-based equations, have shown promise in enhancing hydrological models' accuracy, interpretability, and knowledge-discovery potential. Current differentiable models are efficient for NN-based parameter regionalization, but the simple explicit numerical schemes paired with sequential calculations (operator splitting) can incur numerical errors whose impacts on models' representation power and learned parameters are not clear. Implicit schemes, however, cannot rely on automatic differentiation to calculate gradients due to potential issues of gradient vanishing and memory demand. Here we propose a “discretize-then-optimize” adjoint method to enable differentiable implicit numerical schemes for the first time for large-scale hydrological modeling. The adjoint model demonstrates comprehensively improved performance, with Kling–Gupta efficiency coefficients, peak-flow and low-flow metrics, and evapotranspiration that moderately surpass the already-competitive explicit model. Therefore, the previous sequential-calculation approach had a detrimental impact on the model's ability to represent hydrological dynamics. Furthermore, with a structural update that describes capillary rise, the adjoint model can better describe baseflow in arid regions and also produce low flows that outperform even pure machine learning methods such as long short-term memory networks. The adjoint model rectified some parameter distortions but did not alter spatial parameter distributions, demonstrating the robustness of regionalized parameterization. Despite higher computational expenses and modest improvements, the adjoint model's success removes the barrier for complex implicit schemes to enrich differentiable modeling in hydrology.
CCSI Toolset 3.22 Release Highlights The Sequential Design of Experiments user interface was updated to resolve an issue where the results would fail to plot in some cases (e.g., Non-Uniform Space Filling designs). The Machine Learning/Artificial Intelligence module was updated to support Keras 3 and to reflect changes made to dependencies’ syntax. A check was added to ensure PSUADE is installed and available at FOQUS startup. If PSUADE is not installed, a link to the FOQUS documentation is displayed and FOQUS is closed. The copyright year was updated to include 2024 in places where it had not previously been updated. Typographical errors were corrected to improve clarity in variable names and documentation. The FOQUS documentation was updated to reflect the fact that ALAMO can have two executables and indicates the correct executable to add to the Settings path. SimSinter was updated to version 3.1.0. This version removed gPROMS support and included security updates.
This study presents a physics-informed neural network (PINN) framework for reactive transport modeling for simulating fast bimolecular reactions in porous media. Accurate characterization of cAhemical interactions and product formation in surface and subsurface environments is essential for advancing critical mineral extraction and related geoscience applications. The proposed methodology sequentially addresses the flow and diffusion–reaction subproblems. The flow field is computed using a mixed formulation, while the diffusion–reaction system is modeled via two uncoupled tensorial diffusion equations reformulated in terms of chemical invariants. PINNs are employed to solve the governing equations, enabling data-efficient, mesh-free prediction of chemical concentration fields. The framework is validated through a series of benchmark problems involving flow in heterogeneous porous media. Initial verification is conducted using patch tests for the flow field, followed by validation of the transport problem with emphasis on preserving non-negativity of concentrations. The complete fast bimolecular reaction scenario is then solved, yielding spatial distributions of reactants and product species. Results demonstrate that the PINNs-based approach effectively captures sharp, mixing-limited reaction fronts and dispersive mixing behavior, offering reliable predictions of reactive plume evolution. These capabilities are crucial for evaluating long-term subsurface behavior in applications such as fluid storage, energy extraction, and efficient extraction of critical minerals.