Search NASA⌕ Search

SEARCH · Search NASA

Results for “Online 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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Deep reinforcement learning with online data augmentation to improve sample efficiency for intelligent HVAC control

Deep Reinforcement Learning (DRL) has started showing success in real-world applications such as building energy optimization. Much of the research in this space utilized simulated environments to train RL-agent in an offline mode. Very few research have used DRL-based control in real-world systems due to two main reasons: 1) sample efficiency challenge---DRL approaches need to perform a lot of interactions with the environment to collect sufficient experiences to learn from, which is difficult in real systems, and 2) comfort or safety related constraints---user's comfort must never or at least rarely be violated. In this work, we propose a novel deep Reinforcement Learning framework with online Data Augmentation (RLDA) to address the sample efficiency challenge of real-world RL. We used a time series Generative Adversarial Network (TimeGAN) architecture as a data generator. We further evaluated the proposed RLDA framework using a case study of an intelligent HVAC control. With a ≈28% improvement in the sample efficiency, RLDA framework lays the way towards increased adoption of DRL-based intelligent control in real-world building energy management systems.

Kurte, Kuldeep↗
DOE OSTI2022

Reinforcement learning for online adaptation of model predictive controllers: Application to a selective catalytic reduction unit

Here we present a novel application of reinforcement learning (RL) for online dynamic tuning of model predictive controllers (MPC). Applying a state-action-reward-state-action (SARSA) algorithm for temporal difference learning with a control-specific reward function improves the error tracking performance of a standard MPC formulation. The proposed RL approach is also readily adaptable to other MPCs, or entirely different control approaches. Practical details for the implementation of the RL-MPC algorithm are also presented. The proposed algorithm is applied to a case study of controlling nitrogen oxide (NO x ) emissions in an industrial selective catalytic reduction (SCR) unit, a control problem characterized by significant nonlinearity and time delay. Along with an RL-MPC formulation for NOx control, another MPC is proposed to mitigate ammonia slip and decrease ammonia consumption in the SCR. Results showing the efficacy of the RL-MPC for NO x control through learning and implementation on the nonlinear SCR dynamic model are presented.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

A neural-network-enhanced parameter-varying framework for multi-objective model predictive control applied to buildings

Management of the electrical grid is becoming more complex due to the increased penetration of alternative energy generation technologies and a broadening diversity of electric loads. This complexity creates challenges in balancing demand and generation that can increase the potential for grid instabilities. One effective way to address this issue is to leverage previously unexploited demand flexibility through advanced control strategies. In this work, we propose an advanced control method, called adaptive neural parameter-varying model predictive control (ANPV-MPC), to control the temperature and energy consumption of a building via its Heating, Ventilation, and Air Conditioning system. ANPV-MPC combines key ideas in parameter-varying control, adaptive control, and online learning strategies to bridge the gap between computationally efficient linear model predictive control and more accurate nonlinear model predictive control. The novelty in ANPV-MPC is the use of a physics-inspired Bayesian neural network to estimate the coefficients of the parameter-varying linear control model. The Bayesian neural network additionally provides uncertainty estimates, triggering online training to capture evolving building system conditions. We show that ANPV-MPC can approximate the building system dynamics with a 28.39% higher accuracy than traditional linear model predictive control, resulting in 36.23% better control performance without increasing complexity of the optimal control problem. ANPV-MPC also adapts in real time to previously unseen conditions using online learning, further improving its performance.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2025

Towards a Robust Adaptive Digital Twin for Fusion Applications

The development of a digital twin system for fusion applications is essential for enhancing the prediction, analysis, and optimization of complex plasma processes. Machine learning (ML), particularly deep learning has demonstrated strong capabilities in modeling such highly nonlinear and intricate systems. However, two critical challenges limit the deployment of deep learning-based digital twins: Uncertainty Quantification (UQ) and data drift. UQ is vital for ensuring trustworthy predictions, especially in decision-support scenarios. Additionally, data-driven models are often sensitive to changes in the underlying data distribution, such as shot-to-shot variations in fusion experiments, which can lead to performance degradation over time. To address these challenges, we are developing an uncertainty-aware, adaptive digital twin framework. Our approach incorporates deep learning models enhanced with Gaussian Process approximations for predictive uncertainty estimation, coupled with an online learning mechanism that enables continuous model adaptation to new experimental data. This adaptive capability allows the data driven models to respond effectively to evolving plasma behaviors and equipment conditions. Specifically, to mitigate the effects of shot-to-shot drift, our system updates itself incrementally as new data becomes available, improving both robustness and fidelity. Our vision is to evolve this data driven model into a self-sustaining digital twin system that leverages UQ based feedback to continuously refine itself and potentially support real-time decision making. This presentation will cover a brief background on uncertainty quantification for ML, our ongoing effort on development of UQ capabilities for ML, our data science pipeline from data collection to model development and analysis and online learning framework for modeling coil deflection at DIII-D. I will also briefly touch upon opportunities and challenges in development of digital twin framework.

Sammuli, Brian [General Atomics]↗
DOE OSTI2025

Risk-Aware Reinforcement Learning Framework for User-Centric O-RAN

The evolution of Open Radio Access Networks (O-RAN) presents an opportunity to enhance network performance by enabling dynamic orchestration of configuration and optimization parameters (COPs) through online learning methods. However, leveraging this potential requires overcoming the limitations of traditional cell-centric RAN architectures, which lack the necessary flexibility. On the other hand, despite their recent popularity, the practical deployment of online learning frameworks, such as Deep Reinforcement Learning (DRL)-based COP optimization solutions, remains limited due to their risk of deteriorating network performance during the exploration phase. In this article, we propose and analyze a novel risk-aware DRL framework for user-centric RAN (UC-RAN), which offers both the architectural flexibility and COP optimization to exploit this flexibility. We investigate and identify UC-RAN COPs that can be optimized via a soft actor-critic algorithm implementable as an O-RAN application (rApp) to jointly maximize latency satisfaction, reliability satisfaction, area spectral efficiency, and energy efficiency. We use the offline learning on UC-RAN to reliably accelerate DRL training, thus minimizing the risk of DRL deteriorating cellular network performance. Results show that our proposed solution approaches near-optimal performance in just a few hundred iterations with a decrease in risk score by a factor of ten.

6G and beyond↗
DOE OSTI2023

Machine-learning-based, online estimation of ceramic’s microstructure upon the laser spot brightness during laser sintering

The ceramic microstructure strongly influences its properties. During manufacturing, the online monitoring of microstructure is critical to ensure the desired material properties. So far, the microstructure on the relevant scale is usually characterized offline using scanning electron microscopy (SEM), which is time and cost-consuming. In this work, we demonstrate a cost-effective, machine learning (ML)-based approach to simulate the SEM micrographs in real-time from the laser spot brightness. We experimentally observed a strong correlation between the laser spot brightness and the corresponding microstructure at the exact locations. The brightness values obtained from thermal emission images and the corresponding SEM micrographs were used in the training datasets. The ML algorithm was a style-based conditional generative adversarial network (CGAN). After training, the ML model could generate high-fidelity microstructure images within 0.1 seconds based on in-situ captured brightness at the laser sintering spot. We used the average grain sizes as the metric to evaluate the accuracy of the ML-predicted micrographs. Here, the ML-predicted microstructures were in good agreement, with less than 5% in difference from the real SEM images. In conclusion, we demonstrate the cost-effective, online microstructure estimation during laser sintering with a simple setup (a camera, a regular computer, and the ML model).

08 HYDROGEN↗
DOE OSTI2025

Scalable 3D reconstruction for X-ray single particle imaging with online machine learning

X-ray free-electron lasers offer unique capabilities for measuring the structure and dynamics of biomolecules, helping us understand the basic building blocks of life. Notably, high-repetition-rate free-electron lasers enable single particle imaging, where individual, weakly scattering biomolecules are imaged under near-physiological conditions with the opportunity to access fleeting states that cannot be captured in cryogenic or crystallized conditions. Existing X-ray single particle reconstruction algorithms, which estimate the particle orientation for each image independently, are slow and memory-intensive when handling the massive datasets generated by emerging free-electron lasers. Here, we introduce X-RAI (X-Ray single particle imaging with Amortized Inference), an online reconstruction framework that estimates the structure of 3D macromolecules from large X-ray single particle datasets. X-RAI consists of a convolutional encoder, which amortizes pose estimation over large datasets, as well as a physics-based decoder, which employs an implicit neural representation to enable high-quality 3D reconstruction in an end-to-end, self-supervised manner. We demonstrate that X-RAI achieves state-of-the-art performance for small-scale datasets in simulation and challenging experimental settings and demonstrate its unprecedented ability to process large datasets containing millions of diffraction images in an online fashion. These abilities signify a paradigm shift in X-ray single particle imaging towards real-time reconstruction.

Computer science↗
DOE OSTI2026

Microsecond-latency feedback at a particle accelerator by online reinforcement learning on hardware

The commissioning and operation of future large-scale scientific experiments will challenge current tuning and control methods. Reinforcement learning (RL) algorithms are a promising solution due to their ability to dynamically adapt to changing environments and consider delayed consequences. In many real-world applications, RL policies must produce actions in real time, often within microseconds to milliseconds, imposing significant constraints on system latency and computational overhead that conventional machine learning libraries are not designed to handle. To control phenomena in real time at these timescales, RL needs to be deployed on-the-edge, namely on dedicated hardware located near the system it controls, without relying on a host CPU or cloud-based inference. In this work we present the design and deployment of an experience accumulator system in a particle accelerator. In this system, deep-RL algorithms run using hardware acceleration and act within a few microseconds, enabling the use of RL for control of phenomena like beam instabilities. The training uses the collected data offline to reduce the number of operations carried out on the acceleration hardware. The proposed architecture was tested in real experimental conditions at the Karlsruhe research accelerator, a synchrotron light source, where the system was used to control artificially induced horizontal betatron oscillations in real-time, with a control loop period of just 2.7 μs. The results showed a performance comparable to the commercial feedback system available at the accelerator, demonstrating the viability and potential of this approach. Due to the self-learning and reconfiguration capability of this implementation, a seamless application to other control problems is possible. Applications range from particle accelerators to large-scale research and industrial facilities.

FPGA↗
DOE OSTI2022

Large Scale Caching and Streaming of Training Data for Online Deep Learning

The training of deep neural network models on large data remains a difficult problem, despite progress towards scalable techniques. In particular, there is a mismatch between the random but predetermined order in which AI flows select training samples and the streaming I/O patterns for which traditional HPC data storage (e.g., parallel file systems) are designed. In addition, as more data are obtained, it is feasible neither simply to train learning models incrementally, due to catastrophic forgetting (i.e., bias towards new samples), nor to train frequently from scratch, due to prohibitive time and/or resource constraints. In this paper, we study data management techniques that combine caching and streaming with rehearsal support in order to enable efficient access to training samples in both offline training and continual learning. We revisit state-of-art streaming approaches based on data pipelines that transparently handle prefetching, caching, shuffling, and data augmentation, and discuss the challenges and opportunities that arise when combining these methods with data-parallel training techniques. We also report on preliminary experiments that evaluate the I/O overheads involved in accessing the training samples from a parallel file system (PFS) under several concurrency scenarios, highlighting the impact of the PFS on the design of the data pipelines.

data pipelines↗
DOE OSTI2025

Towards Online Machine Learning in DUNE Data Acquisition

Processing the large volumes of data produced by liquid argon time projection chamber (LArTPC) experiments presents a significant challenge, especially those at the scale of DUNE. This is a particular challenge when aiming to trigger on low-energy neutrinos from core-collapse supernovae, which are typically buried in a high-rate radiological background. To enable real-time event selection suitable for such rare signals, we are developing machine learning based data filtering methods. In order to demonstrate the feasibility of this approach, we implemented such pipeline using the ICEBERG detector at Fermilab as a small-scale LArTPC, with a focus on identifying Michel electrons as a proxy for low-energy neutrino interactions. This poster will present the current status of integrating these machine learning models into the data acquisition (DAQ) system of this detector.

Dalager, Olivia [Fermilab]↗
DOE OSTI2020

Robust Online Sequential RVFLNs for Data Modeling of Dynamic Time-Varying Systems with Application of an Ironmaking Blast Furnace

In a world where the increasing complexity of modern industrial processes brings difficulties for accurate mathematical modeling, taking advantage of data has become an efficient solution to complex dynamic process modeling issue. In this paper, we develop a novel robust online sequential version of random vector functional-link networks (RVFLNs) for data-driven modeling of dynamic time-varying system and applied it in a blast furnace (BF) ironmaking process. First, to overcome the time-varying dynamics of process and to enable the RVFLNs to learn online with avoiding data saturation, an improved online sequential version of RVFLNs (OS-RFVLNs) is first presented by online sequential learning with forgetting factor. This improved OS-RVFLNs algorithm is not only suitable for the real-time and large data transfer situation, but also can adjust the sensitivity of the algorithm to different samples with the help of the introduced forgetting factor. Second, since the output weights of the improved OS-RVFLNs as well as other RVFLNs algorithms are obtained by the least squares approach, a robustness problem may occur when the training dataset is contaminated with various outliers. To solve this problem, a Cauchy distribution weighted M-estimator is introduced to improve the robustness of the improved OS- RVFLNs. For this proposed robust OS-RVFLNs (R-OS- RVFLNs), since the weights of different outlier data are properly determined by the Cauchy distribution function, their corresponding contribution on modeling can be properly distinguished. Thus robust and better modeling results can be achieved. Experiments using actual industrial data of BF ironmaking process and comparative studies have demonstrated that the proposed method produces a better estimation accuracy and stronger robustness than other methods.

Blast furnace (BF), Modelling, Dynamic systems↗
DOE OSTI2025

Learning with Adaptive Conservativeness for Distributionally Robust Optimization: Incentive Design for Voltage Regulation

Information asymmetry between the Distribution System Operator (DSO) and Distributed Energy Resource Aggregators (DERAs) obstructs designing effective incentives for voltage regulation. To capture this effect, we employ a Stackelberg game-theoretic framework, where the DSO seeks to overcome the information asymmetry and refine its incentive strategies by learning from DERA behavior over multiple iterations. We introduce a model-based online learning algorithm for the DSO, aimed at inferring the relationship between incentives and DERA responses. Given the uncertain nature of these responses, we also propose a distributionally robust incentive design model to control the probability of voltage regulation failure and then reformulate it into a convex problem. This model allows the DSO to periodically revise distribution assumptions on uncertain parameters in the decision model of the DERA. Finally, we present a gradient-based method that permits the DSO to adaptively modify its conservativeness level, measured by the size of a Wasserstein metric-based ambiguity set, according to historical voltage regulation performance. The effectiveness of our proposed method is demonstrated through numerical experiments.

adaptation models↗
DOE OSTI2024

Learning with Adaptive Conservativeness for Distributionally Robust Optimization: Incentive Design for Voltage Regulation: Preprint

Information asymmetry between the Distribution System Operator (DSO) and Distributed Energy Resource Aggregators (DERAs) obstructs designing effective incentives for voltage regulation. To capture this effect, we employ a Stackelberg game-theoretic framework, where the DSO seeks to overcome the information asymmetry and refine its incentive strategies by learning from DERA behavior over multiple iterations. We introduce a model-based online learning algorithm for the DSO, aimed at inferring the relationship between incentives and DERA responses. Given the uncertain nature of these responses, we also propose a distributionally robust incentive design model to control the probability of voltage regulation failure and then reformulate it into a convex problem. This model allows the DSO to periodically revise distribution assumptions on uncertain parameters in the decision model of the DERA. Finally, we present a gradient-based method that permits the DSO to adaptively modify its conservativeness level, measured by the size of a Wasserstein metric-based ambiguity set, according to historical voltage regulation performance. The effectiveness of our proposed method is demonstrated through numerical experiments.

distribution system operator↗
DOE OSTI2021

Efficient Learning of Accurate Surrogates for Simulations of Complex Systems

Machine learning methods are increasingly deployed to construct surrogate models for complex physical systems at a reduced computational cost. However, the predictive capability of these surrogates degrades in the presence of noisy, sparse, or dynamic data. We introduce an online learning method empowered by optimizer-driven sampling that has two advantages over current approaches: it ensures that all local extrema (including endpoints) of the model response surface are included in the training data and it employs a continuous validation and update process in which surrogates undergo retraining when their performance falls below a validity threshold. We find, using benchmark functions, that optimizer-directed sampling generally outperforms traditional sampling methods in terms of accuracy around local extrema even when the scoring metric is biased towards assessing overall accuracy. Finally, the application to dense nuclear matter demonstrates that highly accurate surrogates for a nuclear equation of state model can be reliably auto-generated from expensive calculations using few model evaluations.

97 MATHEMATICS AND COMPUTING↗