Machine learning enabled reduced-order scenario generation for stochastic analysis of solar power forecasts
Not provided.
Energy & Fuels↗
SEARCH · Engineering Papers
Results for “probabilistic machine 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.
Multi-Fidelity Learning for Distribution System Voltage Probabilistic Analysis with High Penetration of PVs
This paper proposes a multi-fidelity learning approach for distribution voltage probabilistic analysis with high penetration of PVs. Unlike the existing machine learning-based approaches that require a large number of high fidelity data to achieve satisfactory results, our approach strategically leverage massive low fidelity data from inaccurate model simulations and limited high fidelity historical data. The key idea is to use low-fidelity data to establish an initial model and then the high-fidelity data to calibrate and correct the constructed low-fidelity model. This allows us to fuse low- and high-fidelity data, yielding a high fidelity prediction model. Results obtained from a realistic feeder in US with 80% penetration of PVs show that the proposed approach can achieve a similar accuracy to the one with a large number of high fidelity data. This significantly highlights the advantages of the proposed method as compared to existing data-hungry machine learning methods. Different levels of fidelity data and their impacts are also investigated.
distribution system↗
Day-Ahead Probabilistic Forecasting of Net-Load and Demand Response Potentials with High Penetration of Behind-the-Meter Solar-plus-Storage
The goal of this project is to develop advanced methods for day-ahead net-load forecasting, by leveraging the state-of-the-art machine learning techniques. The developed models produce both point and probabilistic forecasts for a variety of use cases, and are versatile to work with different types of data sets. The innovation lies in the novel design of the architectures, leveraging the most recent advances in machine learning that have not been explored in power systems, accompanied by techniques in the broader artificial intelligence fields such as fuzzy systems. This project has achieved the following accomplishments: (1) preprocessing of over 10 data sets covering varying geographical regions, time horizons, and system levels, which form a robust foundation for training and evaluating forecasting models across a wide range of realistic grid scenarios; (2) development of an interactive web app that enables exploratory analysis of load and generation data, and supports better understanding of data trends, anomalies, and correlations, facilitating model development and stakeholder engagement; (3) implementation of over 10 benchmark models for point and probabilistic forecasting, which include a mix of conventional machine learning methods and state-of-the-art deep learning approaches, providing a comprehensive baseline for performance comparison and validation of the proposed models; (4) development of a fuzzy system based gradient boosting model, tailored for small (less than 3 years) data sets, which achieves a mean absolute percentage error (MAPE) of 4% for point forecasting and a 20% improvement in average pinball loss for probabilistic forecasting; (5) development of a Transformer (a state-of-the-art deep learning architecture) based neural network model, tailored for large (3 years or more) data sets, which achieves a MAPE of 2% for point forecasting and a 20% improvement in average pinball loss for probabilistic forecasting; (6) development of a methodology for quantifying DR potential, and extensions of the previous models for multi-target forecasting of net load and DR potential, which achieve a MAPE of 10% for DR potential.
24 POWER TRANSMISSION AND DISTRIBUTION↗
Probabilistic Programming for Transportable Source Characterization and Uncertainty Quantification of the North Korean Nuclear Tests 2006–2017
Here, we introduce a transportable technique to determine the yield and depth of burial (DOB) from seismic source spectra of underground nuclear explosions. We demonstrate this technique on the six declared North Korean nuclear tests. This approach derives source spectra in absolute units from regional phase (Pg) amplitudes by correcting the observations for geometric spreading, attenuation, and site amplification. We couple the source spectra and explosion source models with a probabilistic programming framework that integrates deep learning techniques and Bayesian modeling. This approach permits the exchange of information across various data categories to quantify both the data and model uncertainty. This technique stands out as an innovative use of broad‐area propagation models, making it transportable across various geologic settings. This method proves to be effective in scenarios with diverse and/or limited observational data, even when the source depth is unknown. We present new independent estimates of absolute yield and DOB that are consistent with the prior assessments, underscoring the potential of this method in enhancing transportable nuclear explosion monitoring capabilities.
58 GEOSCIENCES↗
Ensemble‐Based, Large‐Eddy Reconstruction of Wind Turbine Inflow in a Near‐Stationary Atmospheric Boundary Layer Through Generative Artificial Intelligence
ABSTRACT To validate the second‐by‐second dynamics of turbines in field experiments, it is necessary to accurately reconstruct the winds going into the turbine. Current time‐resolved inflow reconstruction techniques estimate wind behavior in unobserved regions using relatively simple spectral‐based models of the atmosphere. Here, we develop a technique for time‐resolved inflow reconstruction that is rooted in a large‐eddy simulation model of the atmosphere. Our “large‐eddy reconstruction” technique blends observations and atmospheric model information through a diffusion model machine learning algorithm, allowing us to generate probabilistic ensembles of reconstructions for a single 10‐min observational period. Our generated inflows can be used directly by aeroelastic codes or as inflow boundary conditions in a large‐eddy simulation. We verify the second‐by‐second reconstruction capability of our technique in three synthetic field campaigns, finding positive Pearson correlation coefficient values () between ground‐truth and reconstructed streamwise velocity, as well as smaller positive correlation coefficient values for unobserved fields (spanwise velocity, vertical velocity, and temperature). We validate our technique in three real‐world case studies by driving large‐eddy simulations with reconstructed inflows and comparing to independent inflow measurements. The reconstructions are visually similar to measurements, follow desired power spectra properties, and track second‐by‐second behavior ().
17 WIND ENERGY↗
Mesoscale informed parameter estimation through machine learning: A case-study in fracture modeling
Scale bridging is a critical need in computational sciences, where the modeling community has developed accurate physics models from first principles, of processes at lower length and time scales that influence the behavior at the higher scales of interest. However, it is not computationally feasible to incorporate all of the lower length scale physics directly into upscaled models. This is an area where machine learning has shown promise in building emulators of the lower length scale models, which incur a mere fraction of the computational cost of the original higher fidelity models. We demonstrate the use of machine learning using an example in materials science estimating continuum scale parameters by emulating, with uncertainties, complicated mesoscale physics. Additionally, we describe a new framework to emulate the fine scale physics, especially in the presence of microstructures, using machine learning, and showcase its usefulness by providing an example from modeling fracture propagation. Our approach can be thought of as a data-driven dimension reduction technique that yields probabilistic emulators. Our results show well-calibrated predictions for the quantities of interests in a low-strain simulation of fracture propagation at the mesoscale level. Furthermore, on average, we achieve ~10% relative errors on time-varying quantities like total damage and maximum stresses. Successfully replicating mesoscale scale physics within the continuum models is a crucial step towards predictive capability in multi-scale problems.
36 MATERIALS SCIENCE↗
Time-Resolved Line Shapes of Single Quantum Emitters via Machine Learned Photon Correlations
Solid-state single-photon emitters (SPEs) are quantum light sources that combine atomlike optical properties with solid-state integration and fabrication capabilities. SPEs are hindered by spectral diffusion, where the emitter’s surrounding environment induces random energy fluctuations. Timescales of spectral diffusion span nanoseconds to minutes and require probing single emitters to remove ensemble averaging. Photon correlation Fourier spectroscopy (PCFS) can be used to measure time-resolved single emitter line shapes, but is hindered by poor signal-to-noise ratio in the measured correlation functions at early times due to low photon counts. Here, we develop a framework to simulate PCFS correlation functions directly from diffusing spectra that match well with experimental data for single colloidal quantum dots. We use these simulated datasets to train a deep ensemble autoencoder machine learning model that outputs accurate, noiseless, and probabilistic reconstructions of the noisy correlations. Using this model, we obtain reconstructed time-resolved single dot emission line shapes at timescales as low as 10 ns, which are otherwise completely obscured by noise. This enables PCFS to extract optical coherence times on the same timescales as Hong-Ou-Mandel two-photon interference, but with the advantage of providing spectral information in addition to estimates of photon indistinguishability. Further, our machine learning approach is broadly applicable to different photon correlation spectroscopy techniques and SPE systems, offering an enhanced tool for probing single emitter line shapes on previously inaccessible timescales.
74 ATOMIC AND MOLECULAR PHYSICS↗
Machine Learning Derived Dynamic Operating Reserve Requirements in High-Renewable Power Systems
Accurately forecasting wind and solar power output poses challenges for deeply decarbonized electricity systems. Grid operators must commit resources to provide reserves to ensure reliable operations in the face of forecast errors, a process which can increase fuel consumption and emissions. We apply neural network-based machine learning to expand the usefulness of median point forecast data by creating probabilistic distributions of short-term uncertainty in demand, wind, and solar forecasts that adapt to prevailing grid conditions. Machine learning derived estimates of forecast errors compare favorably to estimates based on incumbent methods. Reserves derived from machine learning are usually smaller than values derived using incumbent methods, which enables fuel savings during most hours. Machine learning reserves are generally larger than incumbent reserves during times of higher forecast error, potentially improving system reliability. Performance is tested using multi-stage production simulation modeling of the California Independent System Operator (CAISO) system. Machine learning reserves provide production cost and greenhouse gas (GHG) emission reductions of approximately 0.3% relative to historical 2019 requirements. Savings in the 2030 timeframe are highly dependent on battery storage capacity. At lower levels of battery capacity, savings of 0.4% from machine learning reserves are shown. Significant quantities of battery storage are expected to be added to meet California's resource adequacy needs and GHG reduction targets. Addition of these batteries saturate reserve needs and results in minimal within-hour balancing costs in 2030.
24 POWER TRANSMISSION AND DISTRIBUTION↗
EFIT-Prime: Probabilistic and physics-constrained reduced-order neural network model for equilibrium reconstruction in DIII-D
We introduce EFIT-Prime, a novel machine learning surrogate model for EFIT (Equilibrium FIT) that integrates probabilistic and physics-informed methodologies to overcome typical limitations associated with deterministic and ad hoc neural network architectures. EFIT-Prime utilizes a neural architecture search-based deep ensemble for robust uncertainty quantification, providing scalable and efficient neural architectures that comprehensively quantify both data and model uncertainties. Physically informed by the Grad–Shafranov equation, EFIT-Prime applies a constraint on the current density J tor and a smoothness constraint on the first derivative of the poloidal flux, ensuring physically plausible solutions. Furthermore, the spatial location of the diagnostics is explicitly incorporated in the inputs to account for their spatial correlation. Extensive evaluations demonstrate EFIT-Prime's accuracy and robustness across diverse scenarios, most notably showing good generalization on negative-triangularity discharges that were excluded from training. Timing studies indicate an ensemble inference time of 15 ms for predicting a new equilibrium, offering the possibility of plasma control in real-time, if the model is optimized for speed.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
LandScan Mosaic
The LandScan program at Oak Ridge National Laboratory (ORNL), in collaboration with the National Geospatial-Intelligence Agency (NGA), continues to deliver the most accurate and up to date global, high resolution gridded population data. Additionally, the latest advancements in the LandScan HD methodology led to reduced latency in development of rapid updates for geopolitical events. With momentum towards reporting more up to date population estimates, feedback from the user community expressed interest in reporting population estimates in ranges - whether to express a level of uncertainty or confirm to leadership and stakeholders the modeled data are estimates. Building upon the need to understand uncertainty or confidence in the modeled data and report ranges at the global scale, LandScan Mosaic was developed. LandScan Mosaic represents the next generation of high-resolution population modeling, building upon the established success of previous LandScan HD iterations. While LandScan HD employed a deterministic big data fusion approach, LandScan Mosaic enhances this methodology by integrating advanced machine learning techniques to impute missing, yet crucial, population model parameters. This advancement allows for probabilistic modeling of building occupancy and population distribution, incorporating uncertainty quantification through Monte Carlo sampling methods. By combining big data fusion with machine learning-driven imputation and stochastic modeling, LandScan Mosaic provides a more comprehensive and robust representation of population dynamics. LandScan Mosaic will be following the in the footsteps of its longstanding counterpart LandScan Global and releasing a global gridded population raster, at the 3-arcsecond resolution. This technical report documents the current stage of development of LandScan Mosaic, detailing the methodologies and data sources behind the modeling. Stakeholders are encouraged to use this document as an authoritative reference for insight into Mosaic’s data development processes. However, readers should note that LandScan Mosaic remains in a late-stage research and development phase, and methodologies and data presented here are subject to refinements ahead of the anticipated global release in Summer 2025. Feedback and inquiries from users and stakeholders are welcomed as we continue to refine and enhance this important population resource.
97 MATHEMATICS AND COMPUTING↗
Logical Activation Functions v.1.1
SAND2024-01501O Logical Activation Functions software is a PyTorch implementation from the paper, "Logical Activation Functions for Training Arbitrary Probabilistic Boolean Logic." The activation functions approximate logit-space marginalization of probabilistic truth tables from probabilistic interpretations of inputs. They also provide a general methodology to approximate logical relationships between abstract antecedents and consequents for machine learning architectures. They do not target any specific application or use-case. By training probabilistic truth tables, these activation functions can capture more expressive relationships in a neural network than typical elementwise activation functions. This code is only designed for a single compute node with a GPU and is limited to machine learning architectures than can fit within the memory of a single GPU. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.
Duersch, Jed↗
Unified architecture for data-driven metadata tagging of building automation systems
This article presents a Unified Architecture (UA) for automated point tagging of Building Automation System (BAS) data, based on a combination of data-driven approaches. Advanced energy analytics applications—including fault detection and diagnostics and supervisory control—have emerged as a significant opportunity for improving the performance of our built environment. Effective application of these analytics depends on harnessing structured data from the various building control and monitoring systems, but typical BAS implementations do not employ any standardized metadata schema. While standards such as Project Haystack and Brick Schema have been developed to address this issue, the process of structuring the data, i.e., tagging the points to apply a standard metadata schema, has, to date, been a manual process. This process is typically costly, labor-intensive, and error-prone. In this work we address this gap by proposing a UA that automates the process of point tagging by leveraging the data accessible through connection to the BAS, including time-series data and the raw point names. The UA intertwines supervised classification and unsupervised clustering techniques from machine learning and leverages both their deterministic and probabilistic outputs to inform the point tagging process. Furthermore, we extend the UA to embed additional input and output data-processing modules that are designed to address the challenges associated with the real-time deployment of this automation solution. We test the UA on two datasets for real-life buildings: (i) commercial retail buildings and (ii) office buildings from the National Renewable Energy Laboratory (NREL) campus. We report the proposed methodology correctly applied 85–90% and 70–75% of the tags in each of these test scenarios, respectively for two significantly different building types used for testing UA's fully-functional prototype. The proposed UA, therefore, offers promising approach for automatically tagging BAS data as it reaches close to 90% accuracy. Further building upon this framework to algorithmically identify the equipment type and their relationships is an apt future research direction to pursue.
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
Machine-Learning-Based Adaptive Thinning of CrIS Radiances to Improve Global Tropical Cyclone Analysis and Forecasts
This work is focused on optimizing the assimilation of hyperspectral infrared (IR) radiances from the Cross-track Infrared Sounder (CrIS) with the goal of improving the representation of tropical cyclones (TCs) in global analyses and forecasts. Current operational assimilation systems rely on subsampling IR radiances on a regular thinning grid. A new and improved adaptive methodology based on machine learning (ML) recognizes TCs from geostationary satellite imagery and is implemented in the Goddard Earth Observing System (GEOS) model and data assimilation framework. The ML methodology is extensively trained on existing TC data sets and creates for each TC a dynamic mask, based on the evolving shape and life cycle of that specific event. Once a TC mask is created, a switch is then activated in the data assimilation system to alter the thinning, ingesting more CrIS radiances within the moving mask, thus increasing the TC sampling. After the TC dissipates, the assimilation of CrIS radiances reverts to normal data density. Results of TC segmentation provided by a state-of-the-art generative machine learning model known as the Denoising Diffusion Probabilistic Model (DDPM) are compared to the previously used U-Net model. The new approach surpasses the performance of the previously developed one. The methodology is applied to both clear-sky and cloud-cleared radiances. Benefits from the latter methodology, particularly in improving the structure of TCs and the intensity forecasts, are presented.
Oreste Reale↗
Systems and methods for improved aircraft safety
Methods and system for improved aircraft safety are described herein. A machine learning model may be trained using historical flight data for a number of previously flown flights where an automation function autonomously disengaged. The trained machine learning model may be used to generate probabilistic alert rules. The probabilistic alert rules may be used by a computing device onboard an aircraft to provide contextual information to flight crew relating to engagement status for one or more automation functions of the aircraft.
Sherry, Lance↗
Applying AI/ML Techniques to U.S. Nuclear Operating Experience Program
Idaho National Laboratory (INL) has provided technical assistance to the U.S. Nuclear Regulatory Commission (NRC) in reliability and risk analysis including the operating experience (OpE) program since the 1980s. The U.S. nuclear OpE program provides input parameters to the NRC Standardized Plant Analysis Risk models and the industry probabilistic risk assessment (PRA) models. While earlier PRA focuses were on at-power, internal event analysis, the risks from external hazards and during low power shutdown (LPSD) operation could be significant and the needs to develop LPSD PRA and external hazards PRA are on the rise. One issue in developing LPSD PRA is the reasonable estimation of shutdown initiative event (SDIE) frequencies. INL has developed and is maintaining an SDIE database for the NRC. However, this database is based on the reviewing of Licensee Event Reports (LERs), which is believed to be only a subset of “actual” shutdown initiating events occurred in the industry. This paper investigates a new approach to identify and characterize shutdown initiating events from the Institute of Nuclear Power Operations (INPO) industry database using machine learning techniques. The main process in this approach is to find out the relationship between key words in event descriptions and the SDIE categories as in the NRC SDIE database. The relationship can then be applied to the INPO database and search for SDIEs.
99 GENERAL AND MISCELLANEOUS↗
Operational Probabilistic Tools for Solar Uncertainty (OPTSUN) (Final Project Report for DOE Solar Forecasting II Project)
Increasing levels of solar PV can challenge system operations and may require novel methods to operate the power system reliably and efficiently. Power system operating plans generally use deterministic forecasts, in which the variable energy resources are represented by the expected value for each interval of the decision horizon. Probabilistic forecasts are relatively new but have the potential to address the shortfalls of deterministic forecasts. However, understanding how best to use such forecasts is still a key gap in industry and was the focus of this project. The project had three workstreams. In a forecasting workstream, improvements were made to baseline probabilistic forecasts using a number of new approaches such as machine learning methods and improved input data. In a design workstream, advanced simulation tools used these forecasts to investigate newly proposed reserve determination methods. Lastly, in a demonstration workstream a scheduling management platform (SMP) was developed to leverage probabilistic forecasts in a modular and customizable manner. In order to study the benefits that could be accrued, the project team collaborated with three utility partners (Duke Energy, Southern Company and Hawaiian Electric) to deliver improved probabilistic forecasts for each region and to model each region in case studies using advanced production cost modeling tools. Different methods to determine operating reserve requirements from probabilistic forecasts were developed, simulated, and tested across each region. The benefits of using these newly proposed methods varied by utility, but, in general, using probabilistic forecasts as well as historical data to set the reserve requirements seems to improve reliability related results, with less risk of reserve or supply shortfalls. The cost implications were not always straightforward; in some cases the new methods could show a reduction in expected operating costs, but often the increase in reserves associated with better risk mitigation using probabilistic forecasts could result in an increase in operating costs in the simulations. The SMP tool was developed to process probabilistic forecasts from their initial receipt through to scheduling decisions. This open-source tool consists of several modules for scenario development, reserve requirements calculation, and visualization. The SMP tool was demonstrated to a wide range of operators and stakeholders at all three utilities and further improved based on their feedback. The tool will be available on www.epri.com/optsun. The proposed probabilistic information-based reserve determination approaches have the potential to be implemented by different regions to ensure an economic and reliable power system operation on power systems integrating increasing levels of variable renewable resources. The innovative yet practical methods developed in this project demonstrated tangible benefits from using probabilistic forecasts beyond just study-based assessments to include three unique balancing areas. The demonstrated benefits across the multiple utility environments, are expected to provide system operators in all regions the confidence required and a platform to adopt the new forecasting and operating methods.
14 SOLAR ENERGY↗
Predicting peak day and peak hour of electricity demand with ensemble machine learning
Battery energy storage systems can be used for peak demand reduction in power systems, leading to significant economic benefits. Two practical challenges are 1) accurately determining the peak load days and hours and 2) quantifying and reducing uncertainties associated with the forecast in probabilistic risk measures for dispatch decision-making. In this study, we develop a supervised machine learning approach to generate 1) the probability of the next operation day containing the peak hour of the month and 2) the probability of an hour to be the peak hour of the day. Guidance is provided on preparation and augmentation of data as well as selection of machine learning models and decision-making thresholds. The proposed approach is applied to the Duke Energy Progress system and successfully captures 69 peak days out of 72 testing months with a 3% exceedance probability threshold. On 90% of the peak days, the actual peak hour is among the 2 h with the highest probabilities.
25 ENERGY STORAGE↗