SEARCH · Search NASA
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.
A material sorting system sorts materials utilizing a vision system that implements a machine learning system in order to identify or classify each of the materials, which are then sorted into separate groups based on such an identification or classification determining that the materials are composed of either wrought aluminum or cast aluminum.
A system classifies materials utilizing a vision system that implements a machine learning system, such as a neural network, in order to identify or classify each of the materials as either a monetary coin or not a monetary coin, which may then be sorted into separate groups based on such an identification or classification. Such a system can sort monetary coins from other forms of scrap, which may have been produced from a shredding of end of life vehicles.
A material sorting system sorts materials utilizing multiple stages of classification and sorting, including a vision system that implements a machine learning system in order to identify or classify each of the materials, and Laser Induced Breakdown Spectroscopy to perform a subsequent classification and sorting of the remaining materials.
A material sorting system sorts materials utilizing a vision system that implements a machine learning system in order to identify or classify each of the materials, which are then sorted into separate groups based on such an identification or classification determining that the materials are composed of either wrought aluminum, extruded aluminum, or cast aluminum.
A material sorting system sorts materials utilizing an x-ray fluorescence and/or a vision system that implements a machine learning system in order to identify or classify each of the materials, which are then sorted into separate groups based on such an identification or classification determining that the materials are composed of either wrought aluminum, extruded aluminum, or cast aluminum. The system is capable of sorting between cast aluminum alloys and also between wrought aluminum alloys.
Systems and methods for classifying and sorting of plastic materials utilizing a vision system and one or more sensor systems, which may implement a machine learning system in order to identify or classify each of the materials, which may then be sorted into separate groups based on such an identification or classification.
Frequency control from photovoltaic (PV) power plants has great potential to address the frequency response challenge of the power system with high penetrations of renewable generation. Using model-based approaches to determine the optimal PV headroom reserve, however, requires significant online computation and is intractable for an interconnection level system. This paper proposes a machine learning based strategy, that is suitable for real-time operation, to determine the optimal PV reserve for frequency control. The proposed machine learning algorithm is trained and tested on 1,987 offline simulations of a 60% renewable penetration Western Electricity Coordinating Council (WECC) system. Furthermore, the proposed reserve determination strategy is applied on a realistic 1-day operation profile of the WECC system and demonstrates a savings of more than 40% PV headroom compared to a conservative approach. It is evident that the proposed strategy can efficiently and effectively determine the optimal PV frequency control reserve for realistic interconnection systems.
Standard deep learning algorithms require differentiating large nonlinear networks, a process that is slow and power-hungry. Electronic contrastive local learning networks (CLLNs) offer potentially fast, efficient, and fault-tolerant hardware for analog machine learning, but existing implementations are linear, severely limiting their capabilities. These systems differ significantly from artificial neural networks as well as the brain, so the feasibility and utility of incorporating nonlinear elements have not been explored. Here, we introduce a nonlinear CLLN—an analog electronic network made of self-adjusting nonlinear resistive elements based on transistors. We demonstrate that the system learns tasks unachievable in linear systems, including XOR (exclusive or) and nonlinear regression, without a computer. We find our decentralized system reduces modes of training error in order (mean, slope, curvature), similar to spectral bias in artificial neural networks. The circuitry is robust to damage, retrainable in seconds, and performs learned tasks in microseconds while dissipating only picojoules of energy across each transistor. This suggests enormous potential for fast, low-power computing in edge systems like sensors, robotic controllers, and medical devices, as well as manufacturability at scale for performing and studying emergent learning.
Power system emergency control is generally regarded as the final safety net for grid security and resiliency. Existing emergency control schemes are usually designed off-line based on either the conceived “worst” case scenarios or a few typical operation scenarios. These schemes are facing significant adaptiveness and robustness issues as increasing uncertainties and variations occur in modern electrical grids. To address these challenges, for the first time, this paper proposes a novel adaptive emergency control scheme using deep reinforcement learning (DRL), by leveraging the high-dimensional feature extraction and non-linear generalization capabilities DRL has for complex systems with high-dimensional variations. Furthermore, an open-source platform named DeepGrid has been designed for the first time to assist the DRL development and benchmarking processes in power system emergency control. Details of the platform, DRL, and emergency control schemes that use dynamic braking or under-voltage load shedding are presented. Extensive case studies performed in both two-area four-machine system and IEEE 39-Bus system have demonstrated the excellent performance and robustness of the proposed schemes.
Federated learning (FL) is increasingly deployed on accelerator-rich high-performance computing (HPC) systems, yet the system-level energy cost of privacy-aware FL remains poorly understood, particularly across heterogeneous networking and server-placement options. We present a measurement-driven study of energy–performance trade-offs for FL on GH200-class nodes across three deployment configurations: CPU-Ethernet, CPU-InfiniBand (RDMA-capable), and a DPU-hosted FL server over InfiniBand using a BlueField-3 SmartNIC/DPU. Using NVIDIA FLARE (NVFLARE), we align node-level power telemetry with per-round timing extracted from NVFLARE logs to quantify time-to-solution (TTS), energy-to-solution (ETS), energy-delay product (EDP), and synchronization behavior for three transformer models (ALBERT, DistilBERT, BERT), trained with and without differential privacy (DP). We find that interconnect choice is the dominant driver of runtime and energy: host-managed InfiniBand consistently reduces communication overhead versus Ethernet, yielding lower TTS/ETS/EDP. In contrast, in our NVFLARE deployment, placing the FL server on the DPU does not consistently match CPU-InfiniBand performance and can be slower—especially for larger models—highlighting that server placement alone is not sufficient to guarantee end-to-end gains. Finally, under our fixed-round protocol, DP increases per-round cost and runtime variance; ETS increases largely in proportion to TTS because average node power remains relatively stable across configurations.
Deep reinforcement learning (DRL) algorithms have successfully solved many challenging problems in various power system control scenarios. However, their decision-making process is usually regarded as black-boxes. Furthermore, how DRL models interact with human intelligence remains an open problem. Thus, this paper proposes a policy extraction framework to extract a complex DRL model into an explainable policy. This framework includes three parts: 1) DRL training and data generation. We train an agent for a specific control task and generate data, which contains the control policy of the agent. 2) Policy extraction. We propose an information gain rate based weighted oblique decision tree (IGR-WODT) for DRL policy extraction. 3) Policy evaluation. We define three metrics to evaluate the performance of the proposed approach. A case study for the under-voltage load shedding problem shows that the IGR-WODT presents a performance enhancement compared with DRL, weighted oblique decision tree, and univariate decision tree. The proposed policy extraction method could provide an intuitive explanation of the neural network decision-making process to the dispatchers when making final decisions on power grid operation. Also, the resulted rule-based controller could replace the deep neural network-based controller in many field edge devices with limited computing resources, providing comparable performance.
Reliability and stability of the electric grid today has depended upon operations of the grid which include the protective relay. Today, the electricity sector faces new challenges with the shift of generation resource characteristics away from the traditional “big iron” generation to inverter-based resources (IBR) which shift the physics and assumption used in grid operation and protection. These changing conditions represent new challenges for protective relays (identification of faults) and increased challenges for protection engineers (correct settings and configuration, reduction of mis-operations), both issues recognized in research and industry. Finding new approaches to reduce mis-operations in relaying and new approaches to fault identification is critical to grid operations. Using today’s modern technology of embedded systems, edge computing, machine learning (ML), and communications we can help address challenges and augment and improve on existing power system operations methodologies.
We develop a methodology based on Bayesian inference over Probabilistic Graphical Models (PGMs) to understand and quantify risk in solar-centered grids using targeted measurements and learned system behavior. Being non-prescriptive but, rather, able to infer system behavior and, ultimately, address risk queries from data, our machine learning-type paradigm is tailored for diverse topologies and threat scenarios often associated with distributed energy generation and photovoltaic distributed energy resources (PV-DERs) in particular. We describe algorithmic processes for: (i) learning the structure of PGMs that result from attack-prone PV-DER-proliferated distribution systems, (ii) quantifying cause-effect relationships, and (iii) evaluating risk queries based on diverse evidence. The contributions are illustrated on a residential grid subject to output impairment attacks on its PV-DER infrastructure.
Deep learning (DL) workloads and their performance at scale are becoming important factors to consider as we design, develop and deploy next-generation high-performance computing systems. Since DL applications rely heavily on DL frameworks and underlying compute (CPU/GPU) stacks, it is essential to gain a holistic understanding from compute kernels, models, and frameworks of popular DL stacks, and to assess their impact on science-driven, mission-critical applications. At Oak Ridge Leadership Computing Facility (OLCF), we employ a set of micro and macro DL benchmarks established through the Collaboration of Oak Ridge, Argonne, and Livermore (CORAL) to evaluate the AI readiness of our next-generation supercomputers. In this paper, we present our early observations and performance benchmark comparisons between the Nvidia V100 based Summit system with its CUDA stack and an AMD MI100 based testbed system with its ROCm stack. We take a layered perspective on DL benchmarking and point to opportunities for future optimizations in the technologies that we consider.
Power system protection devices have transitioned over the past few decades from mechanical to analog devices, then to solid state and finally digital. Relays and their associated critical network of equipment have significantly increased in complexity. Even internally, relays have gained significant intricacy, with relatively simple overcurrent or differential functions now being assisted by a myriad of other functions. This is necessary as the grid becomes more complex, but it brings increased difficulty in monitoring and upkeep. Misoperation caused by improper relay settings or malicious actions is a constant challenge faced by all utilities. These improper settings can be difficult to identify and may require exhaustive post-mortem analysis, typically after a major outage event has already occurred. A mechanism is needed for monitoring the behavior of protection systems to validate that they act and perform as expected. This work presents a concept for a machine learning (ML) system capable of validating the performance of protection systems by classifying anomalous events and characterizing protection system responses based solely on available current and voltage measurements. As a first step in its development, an experimental dataset is generated, and a random forest model is implemented with high accuracy in distinguishing four power system scenarios.
We describe a spatial augmented reality system with a tangible user interface used to control computer simulations of complex systems. In spatial augmented reality, the user’s physical space is augmented with projected imagery, blending real objects with projected information, and a tangible user interface enables users to manipulate physical objects as controllers for interactive visualizations. Our system learns ad hoc objects in the user’s environment as fiducial markers (i.e., objects that are visually recognized and tracked). When combined with simulation and visualization tools, these interfaces allow the user to control simulations or ensembles of simulations via physical objects using apt metaphors. While other research has leveraged the use of depth cameras, our system enables the use of standard cameras in readily available smartphones and webcams and has an implementation that runs completely in JavaScript in the web browser. We discuss the prerequisite object-recognition requirements for such tangible user interfaces and describe computer-vision and machine-learning algorithms meeting those requirements. We conclude by presenting example applications, which are also available online.
With the increased reliance on digitization in industrial control systems, the need for effective monitoring techniques has risen dramatically. Specifically, there is now a growing concern about the so-called false data injection (FDI) attacks. These attacks aim to alter the raw sensors’ data to cause malicious outcomes. Any serious FDI algorithm is based on an intimate knowledge of the system and its associated physics models, which renders conventional outlier/anomaly detection techniques almost obsolete in the face of such attacks. Thus, a critical need has emerged to develop a new class of defense methods that are capable of detecting FDI attacks under the assumption that the attacker has a strong familiarity with the system and its physics modeling. This class of defense methods are denoted by model-based defenses which are premised on the assumption that the attacker, while having a good understanding of the system, does not have full privileged access to all proprietary data and historical records of operation. However, (s)he is assumed to be capable of learning system behavior using self-learning techniques during an initial lie-in-wait period. To defend against this scenario, we propose a new model-based randomized window algorithm that searches time-series data for signatures that can serve as classifiers between normal and FDI scenarios. The classifiers are based on the correlations between the dominant degrees of freedom (DOFs) and the less-dominant DOFs (expected to be very sensitive to the system details that are unknown to the attacker). For demonstration, RELAP5 models are employed to calculate representative nuclear reactor behavior during a number of transient scenarios. Finally, falsified data are injected into the RELAP5-simulated behavior, and the proposed signature-identification algorithm is employed to detect the injected data.
This article proposes data-driven control via a nonlinear autoregressive model with exogenous inputs (NARX) for real-time voltage regulation of a modified feeder using reactive power sources. Traditional voltage control strategies rely on rule-based heuristics or optimization techniques, which often require detailed system models and extensive computational resources. The NARX-based controller learns system dynamics from historical data and predicts optimal reactive power dispatch in real-time for voltage correction. The proposed approach is evaluated on a power system feeder model under varying load and network conditions. Simulation results demonstrate that the NARX-based controller achieves improved voltage regulation, offering higher adaptability to system fluctuations. This study highlights the potential of data-driven control for enhancing the reliability of power distribution networks.