Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Network performance forecasting”

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 109 records · Page 6

Application of Ensemble Sensitivity Analysis to Observation Targeting for Short-term Wind Speed Forecasting

The operators of electrical grids, sometimes referred to as Balancing Authorities (BA), typically make critical decisions on how to most reliably and economically balance electrical load and generation in time frames ranging from a few minutes to six hours ahead. At higher levels of wind power generation, there is an increasing need to improve the accuracy of 0- to 6-hour ahead wind power forecasts. Forecasts on this time scale have typically been strongly dependent on short-term trends indicated by the time series of power production and meteorological data from a wind farm. Additional input information is often available from the output of Numerical Weather Prediction (NWP) models and occasionally from off-site meteorological towers in the region surrounding the wind generation facility. A widely proposed approach to improve short-term forecasts is the deployment of off-site meteorological towers at locations upstream from the wind generation facility in order to sense approaching wind perturbations. While conceptually appealing, it turns out that, in practice, it is often very difficult to derive significant benefit in forecast performance from this approach. The difficulty is rooted in the fact that the type, scale, and amplitude of the processes controlling wind variability at a site change from day to day if not from hour to hour. Thus, a location that provides some useful forecast information for one time may not be a useful predictor a few hours later. Indeed, some processes that cause significant changes in wind power production operate predominantly in the vertical direction and thus cannot be monitored by employing a network of sensors at off-site locations. Hence, it is very challenging to determine the type of sensors and deployment locations to get the most benefit for a specific short-term forecast application. Two tools recently developed in the meteorological research community have the potential to help determine the locations and parameters to measure in order to get the maximum positive impact on forecast performance for a particular site and short-term look-ahead period. Both tools rely on the use of NWP models to assess the sensitivity of a forecast for a particular location to measurements made at a prior time (i.e. the look-ahead period) at points surrounding the target location. The fundamental hypothesis is that points and variables with high sensitivity are good candidates for measurements since information at those points are likely to have the most impact on the forecast for the desired parameter, location and look-ahead period. One approach is called the adjoint method (Errico and Vukicevic, 1992; Errico, 1997) and the other newer approach is known as Ensemble Sensitivity Analysis (ESA; Ancell and Hakim 2007; Torn and Hakim 2008). Both approaches have been tested on large-scale atmospheric prediction problems (e.g. forecasting pressure or precipitation over a relatively large region 24 hours ahead) but neither has been applied to mesoscale space-time scales of winds or any other variables near the surface of the earth. A number of factors suggest that ESA is better suited for short-term wind forecasting applications. One of the most significant advantages of this approach is that it is not necessary to linearize the mathematical representation of the processes in the underlying atmospheric model as required by the adjoint approach. Such a linearization may be especially problematic for the application of short-term forecasting of boundary layer winds in complex terrain since non-linear shifts in the structure of boundary layer due to atmospheric stability changes are a critical part of the wind power production forecast problem. The specific objective of work described in this paper is to test the ESA as a tool to identify measurement locations and variables that have the greatest positive impact on the accuracy of wind forecasts in the 0- to 6-hour look-ahead periods for the wind generation area of California's Tehachapi Pass during the warm (high generation) season. The paper is organized as follows. Section 2 highlights the methodology, Section 3 presents results, and Section 4 concludes with a summary and brief discussion of future work.

17 WIND ENERGY↗

Occlusion-Perturbed Deep Learning for Probabilistic Solar Forecasting via Sky Images

Solar forecasting is shifting to the probabilistic paradigm due to the inherent uncertainty within the solar resource. Input uncertainty quantification is one of the widely used and best-performing ways to model solar uncertainty. However, compared to other sources of inputs, such as numerical weather prediction models, pure sky image-based probabilistic solar forecasting lags behind. In this research, an occlusion-perturbed convolutional neural network, named the PSolarNet, is developed. The PSolarNet provides very short-term deterministic forecasts, forecast scenarios, and probabilistic forecasts of the global horizontal irradiance from sky image sequences. Case studies based on 6 years of open-source data show that the developed PSolarNet is able to generate accurate 10-minute ahead deterministic forecasts with a 5.62% normalized root mean square error, realistic and diverse forecast scenarios with a 0.966 average correlation with the actual time series, and reliable and sharp probabilistic forecasts with a 2.77% normalized continuous ranked probability score.

Bayesian model averaging↗

Occlusion-Perturbed Deep Learning for Probabilistic Solar Forecasting via Sky Images: Preprint

Solar forecasting is shifting to the probabilistic paradigm due to the inherent uncertainty within the solar resource. Input uncertainty quantification is one of the widely used and best-performing ways to model solar uncertainty. However, compared to other sources of inputs, such as numerical weather prediction models, pure sky image-based probabilistic solar forecasting lags behind. In this research, an occlusion-perturbed convolutional neural network, named the PSolarNet, is developed. The PSolarNet provides very short-term deterministic forecasts, forecast scenarios, and probabilistic forecasts of the global horizontal irradiance from sky image sequences. Case studies based on 6 years of open-source data show that the developed PSolarNet is able to generate accurate 10-minute ahead deterministic forecasts with a 5.62% normalized root mean square error, realistic and diverse forecast scenarios with a 0.966 average correlation with the actual time series, and reliable and sharp probabilistic forecasts with a 2.77% normalized continuous ranked probability score.

Bayesian model averaging↗

Real-time streamflow forecasting: AI vs. Hydrologic insights

In this paper, we propose a set of simple benchmarks for the evaluation of data-based models for real-time streamflow forecasting, such as those developed with sophisticated Artificial Intelligence (AI) algorithms. The benchmarks are also data-based and provide context to judge incremental improvements in the performance metrics from the more complicated approaches. The benchmarks include temporal and spatial persistence, persistence corrected for baseflow and streamflow, as well as river distance weighted runoff obtained from space-time distributed rainfall. In the development of the benchmarks, we use basic hydrologic insights such as flow aggregation by the river network, scale-dependence in basin response, streamflow partitioning into quick flow and baseflow, water travel time, and rainfall averaging by the basin width function. The study uses 140 streamflow gauges in Iowa that cover a range of basin scales between 7 and 37,000 km 2 . The data cover 17 years. This work demonstrates that the proposed benchmarks can provide good performance according to several commonly used metrics. For example, streamflow forecasting at half of the test locations across years achieves a Kling-Gupta Efficiency (KGE) score of 0.6 or higher at one-day ahead lead time, and 20% of cases reach the KGE of 0.8 or higher. The proposed benchmarks are easy to implement and should prove useful for developers of data-based as well as physics-based hydrologic models and real-time data assimilation techniques.

54 ENVIRONMENTAL SCIENCES↗

Intelligent load pattern modeling and denoising using improved variational mode decomposition for various calendar periods

This paper proposes a novel approach to identify the load patterns considering calendar impact and denoising process, which advances the state of art of load identification and load forecasting to a more flexible and adaptive extent by revealing the correlation between load profile and the substantial influence factors. With the rapid development and the promising application of energy intemet and grid modernization, the accurate load modeling and forecasting especially by load pattern modeling (LPM) have become more and more critical to observe and estimate load profile so as to optimize the operation efficiency of the power grid. Traditional methods mostly focus on the static LPM itself which are unable to account the non-linear and highly dynamic load fluctuations in order to balance the demand of the power grid during various calendar period (VCP) scheduling. In this paper, a load pattern modeling method for power grid considering the impact of typical calendar effects is proposed using a combination of improved variational mode decomposition (IVMD) and deep belief network (DBN) algorithm. Firstly, an index of load curve curvature (LCC) is proposed to improve the selection of value K in the conventional VMD algorithm. The value K of the intrinsic modality is determined by adopting LCC to better quantify the average instantaneous frequency of the modal decomposition of the VMD and to reduce the elevated stochasticity of k. Secondly, an optimization formulation is developed based on adaptive segmented loop decomposition of the typical correlation coefficient to suppress the edge effect in VMD decomposition, which would enable the reconstructed curve to incorporate the variabilities from the initial data. Finally, the IVMD and DBN algorithms are jointly implemented to optimize the accuracy of VCP load pattern modeling curve regression. Numerical examples have been performed with the results proving that the proposed method significantly outperforms the state-of-art in reducing the modeling error of VCP load curve, improving the overall forecasting accuracy, and reducing load dispatching cost.

deep belief network↗

Studying the Impact of Distributed Solar PV on Power Systems Using Integrated Transmission and Distribution Models

Rapid growth of distributed energy resources has prompted increasing interest in integrated Transmission (T) and Distribution (D) modeling. This paper presents the results of a distributed generation from solar photovoltaics (DGPV) impact assessment study that was performed using a synthetic T &D model. The primary objective of the study was to present a new approach for DGPV impact assessment, where along with detailed models of transmission and distribution networks, consumer loads were modeled using the physics of end-use equipment, and DGPV was geographically dispersed and connected to the secondary distribution networks. The study highlights (i) how a lack of DGPV forecasting can increase the Area Control Error (ACE) at the transmission level for high penetration levels; and (ii) how capturing transmission voltage changes using integrated T &D can change simulated distribution voltage profiles and voltage regulator operations between integrated T &D and distribution-only simulations.

14 SOLAR ENERGY↗

PopGNN: Graph Neural Network-Based Flexible Future Population Forecasting Model

Accurate population forecasts is important to plan critical infrastructure and services, from housing and education to healthcare and transport. However, traditional population prediction studies have only employed traditional machine learning models limited to capture complex spatial interdependencies and patterns. Althogh recently computer vision-based framework was introduced with with promising accuracy, it has critical limitations for real-world planning applications: it function only at fixed spatial resolutions, restricting their use in diverse boundaries such as census tracts, neighborhoods, or administrative zones. Therefore, this study suggests a Graph Neural Network (GNN)-based population prediction framework, called PopGNN. This model recorded remarkable performance compared with state-of-the-art models and traditional baseline models in the grid and administrative boundaries. Furthermore, our framework achieved comparable predictive accuracy to a computer vision-based model in both the South Korea and Tennessee case studies. Consequently, this study is valuable in that a single model can provide accurate population forecasts that address diverse planning demands, ranging from granular grid-level estimates for precise service allocation and facility location planning to aggregate administrative-level forecasts for macro-scale regional policy and resource distribution.

97 MATHEMATICS AND COMPUTING↗

Predicting Storm Outages Through New Representations of Weather and Vegetation

This paper presents new developments in an outage prediction model (OPM) for an electric distribution network in the Northeastern United States and assesses their significance to the OPM performance. The OPM uses regression tree models fed by numerical weather prediction outputs, spatially distributed information on soil, vegetation, electric utility assets, and historical power outage data to forecast the number and spatial distribution of outages across the power distribution grid. New modules introduced hereby consist in 1) a storm classifier based on weather variables; 2) a multimodel optimization of regression tree output; and 3) a post-processing routine for more accurately describing tree-leaf conditions. Model implementations are tested through leave-one-storm-out cross-validations performed on 120 storms of varying intensity and characteristics. The results demonstrate that the median absolute percentage error of the new OPM version decreased from 130% to 59% for outage predictions at the service territory level, and the OPM skills for operational forecasts are consistent with the skills based on historical storm analyses.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Reinforcement learning based automated history matching for improved hydrocarbon production forecast

History matching aims to find a numerical reservoir model that can be used to predict the reservoir performance. An engineer and model calibration (data inversion) method are required to adjust various parameters/properties of the numerical model in order to match the reservoir production history. In this study, we develop deep neural networks within the reinforcement learning framework to achieve automated history matching that will reduce engineers’ efforts, human bias, automatically and intelligently explore the parameter space, and remove the need of large set of labeled training data. To that end, a fast-marching-based reservoir simulator is encapsulated as an environment for the proposed reinforcement learning. The deep neural-network-based learning agent interacts with the reservoir simulator within reinforcement learning framework to achieve the automated history matching. Reinforcement learning techniques, such as discrete Deep Q Network and continuous Deep Deterministic Policy Gradients, are used toth, used to train the learning agents. The continuous actions enable the Deep Deterministic Policy Gradients to explore more states at each iteration in a a learning episode; consequently, a better history matching is achieved using this algorithm as compared to Deep Q Network. For simplified dual-target composite reservoir models, the best history-matching performances of the discrete and continuous learning methods in terms of normalized root mean square errors are 0.0447 and 0.0038, respectively. Furthermore, our study shows that continuous action space achieved by the deep deterministic policy gradient drastically outperforms deep Q network.

42 ENGINEERING↗

Explore Spatio‐Temporal Learning of Large Sample Hydrology Using Graph Neural Networks

Abstract Streamflow forecasting over gauged and ungauged basins play a vital role in water resources planning, especially under the changing climate. Increased availability of large sample hydrology data sets, together with recent advances in deep learning techniques, has presented new opportunities to explore temporal and spatial patterns in hydrological signatures for improving streamflow forecasting. The purpose of this study is to adapt and benchmark several state‐of‐the‐art graph neural network (GNN) architectures, including ChebNet, Graph Convolutional Network (GCN), and GraphWaveNet, for end‐to‐end graph learning. We explicitly represent river basins as nodes in a graph, learn the spatiotemporal nodal dependencies, and then use the learned relations to predict streamflow simultaneously across all nodes in the graph. The efficacy of the developed GNN models is investigated using the Catchment Attributes and MEteorology for Large‐sample Studies (CAMELS) data set under two settings, fixed graph topology (transductive learning), and variable graph topology (inductive learning), with the latter applicable to prediction in ungauged basins (PUB). Results indicate that GNNs are generally robust and computationally efficient, achieving similar or better performance than a baseline model trained using the long short‐term memory (LSTM) network. Further analyses are conducted to interpret the graph learning process at the edge and node levels and to investigate the effect of different model configurations. We conclude that graph learning constitutes a viable machine learning‐based method for aggregating spatiotemporal information from a multitude of sources for streamflow forecasting

Sun, Alexander Y.↗

Anticipating Technical Expertise and Capability Evolution in Research Communities Using Dynamic Graph Transformers

The ability to anticipate global technical expertise and capability evolution trends is essential for national and global security, especially in safety-critical domains such as nuclear nonproliferation (NN) and rapidly emerging fields like artificial intelligence (AI). Here, in this work, we extend traditional statistical relational learning approaches (e.g., link prediction in collaboration networks) and formulate a problem of anticipating technical expertise and capability evolution using dynamic heterogeneous graph representations. We develop novel capabilities to forecast collaboration patterns, authorship behavior, and technical capability evolution at different granularities (e.g., scientist and institution levels) in two distinct research fields. We implement a dynamic graph transformer (DGT) neural architecture, which pushes the state-of-the-art graph neural network models by: 1) forecasting heterogeneous (rather than homogeneous) nodes and edges; and 2) relying on both discrete- and continuous-time inputs. We demonstrate that our DGT models predict collaboration, partnership, and expertise patterns with 0.26, 0.73, and 0.53 mean reciprocal rank values for AI and 0.48, 0.93, and 0.22 for NN domains. DGT model performance exceeds the best-performing static graph baseline models by 30%–80% across AI and NN domains. Our findings demonstrate that DGT models boost inductive task performance when previously unseen nodes appear in the test data for the domains with emerging collaboration patterns (e.g., AI). Specifically, models accurately predict which established scientists will collaborate with early career scientists and vice versa in the AI domain.

97 MATHEMATICS AND COMPUTING↗

Grey-box and ANN-based building models for multistep-ahead prediction of indoor temperature to implement model predictive control

Model-based predictive control (MPC) strategies for heating, ventilation, and air-conditioning (HVAC) systems present an opportunity to lower building energy consumption and operational costs. Such approaches rely on the development of a model to precisely forecast building thermal dynamics, such as room air temperature or heating/cooling rate, and make control-related decisions. The control-oriented modeling of building energy systems should be accurate in predicting indoor conditions and present low computational complexity. These features are the key challenge of implementing advanced control methods such as MPC. Extant studies on building modeling for MPC have focused on step-ahead forecasting techniques to forecast building thermal dynamics, while multistep-ahead forecasting is essential. Moreover, machine learning model suitable in case of the domain-based engineering expertise are also not available. To this aim, we perform a comparative analysis of the grey-box model based on a resistance-capacitance (RC) thermal network and a machine learning model composed of an artificial neural network (ANN) for multistep-ahead prediction of building thermal dynamics using current and historical data. Actual experimental data obtained from the Flexible Research Platform (FRP) in Oak Ridge National Laboratory (US) are used for estimation and validation purposes. The average root mean squared error (RMSE) of the grey-box and ANN models are 0.89 °C and 1.02°C, respectively. Finally, the results indicate that the grey-box model outperforms the ANN model in the considered validation periods in terms of accuracy and prediction stability.

42 ENGINEERING↗

A Statistical Interpolation Code for Ocean Analysis and Forecasting

Abstract We present a data assimilation package for use with ocean circulation models in analysis, forecasting, and system evaluation applications. The basic functionality of the package is centered on a multivariate linear statistical estimation for a given predicted/background ocean state, observations, and error statistics. Novel features of the package include support for multiple covariance models, and the solution of the least squares normal equations either using the covariance matrix or its inverse—the information matrix. The main focus of this paper, however, is on the solution of the analysis equations using the information matrix, which offers several advantages for solving large problems efficiently. Details of the parameterization of the inverse covariance using Markov random fields are provided and its relationship to finite-difference discretizations of diffusion equations are pointed out. The package can assimilate a variety of observation types from both remote sensing and in situ platforms. The performance of the data assimilation methodology implemented in the package is demonstrated with a yearlong global ocean hindcast with a 1/4° ocean model. The code is implemented in modern Fortran, supports distributed memory, shared memory, multicore architectures, and uses climate and forecasts compliant Network Common Data Form for input/output. The package is freely available with an open source license from www.tendral.com/tsis/ .

Srinivasan, Ashwanth↗

Solar Irradiance Nowcasting Case Studies near Sacramento

The Sun4Cast solar power forecasting system, designed to predict solar irradiance and power generation at solar farms, is composed of several component models operating on both the nowcasting (0–6 h) and day-ahead forecast horizons. The different nowcasting models include a statistical forecasting model (StatCast), two satellite-based forecasting models [the Cooperative Institute for Research in the Atmosphere Nowcast (CIRACast) and the Multisensor Advection-Diffusion Nowcast (MADCast)], and a numerical weather prediction model (WRF-Solar). It is important to better understand and assess the strengths and weaknesses of these short-range models to facilitate further improvements. To that end, each of these models, including four WRF-Solar configurations, was evaluated for four case days in April 2014. For each model, the 15-min average predicted global horizontal irradiance (GHI) was compared with GHI observations from a network of seven pyranometers operated by the Sacramento Municipal Utility District (SMUD) in California. Each case day represents a canonical sky-cover regime for the SMUD region and thus represents different modeling challenges. The analysis found that each of the nowcasting models perform better or worse for particular lead times and weather situations. StatCast performs best in clear skies and for 0–1-h forecasts; CIRACast and MADCast perform reasonably well when cloud fields are not rapidly growing or dissipating; and WRF-Solar, when configured with a high-spatial-resolution aerosol climatology and a shallow cumulus parameterization, generally performs well in all situations. Further research is needed to develop an optimal dynamic blending technique that provides a single best forecast to energy utility operators.

54 ENVIRONMENTAL SCIENCES↗

Evaluating Probability of Containment Effectiveness at a GCS Sites using integrated assessment modeling approach with Bayesian decision Networks

Improved scientific and engineering understanding of the behavior of geologic CO2 storage together with established regulatory framework and incentive structures raise the prospects for accelerated, large-scale deployment of this greenhouse gas emissions reduction approach. Incentive structures call for the establishment of appropriate verification and accounting approaches to support claims of the integrity of a geologic storage complex and to justify taking credit for long-term storage. In this study, we present a framework for assessing the probability of containment effectiveness over the lifetime of a geologic carbon storage site (e.g., after 70 years of injection and post-injection site performance) using forward stochastic model realizations based on site characterization data and using a monitoring-informed Bayesian network based on hypothetical detectability from surface seismic surveys over the site injection and post-injection phases. The National Risk Assessment Partnership’s open-source Integrated Assessment Model (NRAP-Open-IAM) was utilized to develop an ensemble of 10,000 a priori stochastic forecasts of CO2 containment. Those simulations were used to train the Bayesian network model to estimate the prior probabilities of the CO2 leakage mass into overlying, monitorable aquifers considering the uncertainties in the reservoir properties, permeability of potentially leaky wells and the overlying aquifers. The conditional probabilities in the Bayesian network were either learned from the NRAP-Open-IAM simulations or derived from the predefined detection thresholds for the monitoring method. Observations obtained from monitoring, over time during the site operation phases were then used to generate updated posterior probabilities of containment (and any loss from containment) in the Bayesian network by propagating the prior probabilities through the conditional probabilities. We demonstrate how to construct and use the Bayesian network for verifying the long-term storage complex effectiveness informed by monitoring based on the NRAP-Open-IAM simulations previously developed for the FutureGen 2.0 site. This approach may have relevance for stake holders to demonstrate secure geologic storage, provide a defensible, probabilistic approach to claim credit for geologic storage, and to estimate the likelihood that any fraction of the claimed credit may need to be refunded to the creditor based on available monitoring information.

Bayesian network, Risk assessment, Monitoring, car↗

Analytics-at-scale of Sensor Data for Digital Monitoring in Nuclear Plants (3 rd Annual Report)

Nuclear power plants collect and store large volumes of heterogeneous data from various components and systems. With recent advances in machine learning (ML) techniques, these data can be leveraged to develop diagnostic and short-term forecasting models to better predict future equipment condition. Maintenance operations can then be planned in advance whenever degraded performance is predicted, thus resulting in fewer unplanned outages and the optimization of maintenance activities. This enables lower maintenance costs and improves the overall economics of nuclear power. This report primarily focuses on developing a short-term forecasting process that leverages a feature selection process to distill large volumes of heterogeneous data and predict specific equipment parameters. A variety of feature selection methods, including Shapley Additive Explanations (SHAP) and variance inflation factor (VIF), were used to select the optimal features as inputs for three ML methods: long short-term memory (LSTM) networks, support vector regression (SVR), and random forest (RF). Each combination of model and input features was used to predict a pump bearing temperature both 1 and 24 hours in advance, based on actual plant system data. The optimal inputs for the LSTM and SVR were selected using the SHAP values, while the optimal input for the RF consisted solely of the response variable itself. Each model produced similar 1-hour-ahead predictions, with root mean square errors (RMSEs) of roughly 0.006. For the 24-hour-ahead predictions, differences could be seen between LSTM, SVR, and RF, as reflected by model performances of 0.036 ± 0.014, 0.0026 ± 0, and 0.063 ± 0.004 RMSE, respectively. As big data and continuous online monitoring become more widely available, the proposed feature selection process can be used for many applications beyond the prediction of process parameters within nuclear infrastructure. This report summarizes the Fiscal Year 2021 research progress encompassing the (1) data cleaning and feature selection necessary for ML applications; (2) development of short-term forecasting models to predict future plant process parameters for both single and multiple time steps ahead; and (3) validation of the feature selection methods and short-term forecasting models given new data from different systems.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Can a deep-learning model make fast predictions of vacancy formation in diverse materials?

The presence of point defects, such as vacancies, plays an important role in materials design. Here, we explore the extrapolative power of a graph neural network (GNN) to predict vacancy formation energies. We show that a model trained only on perfect materials can also be used to predict vacancy formation energies (E vac ) of defect structures without the need for additional training data. Such GNN-based predictions are considerably faster than density functional theory (DFT) calculations and show potential as a quick pre-screening tool for defect systems. To test this strategy, we developed a DFT dataset of 530 E vac consisting of 3D elemental solids, alloys, oxides, semiconductors, and 2D monolayer materials. We analyzed and discussed the applicability of such direct and fast predictions. We applied the model to predict 192 494 E vac for 55 723 materials in the JARVIS-DFT database. Our work demonstrates how a GNN-model performs on unseen data.

2D materials↗

Comprehensive assessment of deep reinforcement learning approaches for economic dispatch in nuclear-driven microgrids

As the electrical grid integrates more variable renewable energy sources such as wind and solar, the demand for distributed and flexible systems to address this increased variability becomes critical. Nuclear-driven microgrids provide a promising solution by offering stable generation to complement intermittent renewables, ensuring grid reliability and operating efficiency. This paper proposes a recurrent deep reinforcement learning framework for optimal economic dispatch in a nuclear-powered microgrid integrating renewable energy sources, small modular reactors, battery storage systems, and balance-of-plant dynamics. A three-agent control architecture is developed, where demand and renewable energy agents act as forecasters, and a reinforcement learning-based dispatch agent performs real-time energy allocation. A nonlinear programming formulation is first used to generate an optimal baseline for benchmarking. The proposed dispatch controller, based on Proximal Policy Optimization enhanced with Long Short-Term Memory networks, exploits temporal correlations in system dynamics by taking advantage of the time series used as inputs to improve policy robustness under uncertainty. Comparative analysis against established deep reinforcement learning methods, including Proximal Policy Optimization with a feedforward architecture, Soft Actor-Critic, and Twin Delayed Deep Deterministic Policy Gradient, demonstrates superior performance. Numerical results indicate that the proposed controller achieves a 0.39% cost reduction relative to the nonlinear programming benchmark and outperforms other learning-based methods by generating additional revenue of up to 0.35%. All reinforcement learning controllers compute dispatch actions in less than 0.3 s, resulting in a computational speedup of more than three orders of magnitude over the nonlinear programming baseline. The findings of this paper highlight their applicability for real-time operation and control in nuclear-integrated microgrids under volatile operating conditions.

24 POWER TRANSMISSION AND DISTRIBUTION↗