Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “deep neural network”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Transfer-Learnt Energy Models for Predicting Electricity Consumption in Buildings with Limited and Sparse Field Data

Modeling energy consumption is critical for energy-efficient utilization of the electric appliances in a building, smart grid programs (like demand-response), and many other smart home applications. State-of-the-art energy modeling techniques either rely on theoretical models, or extensive instrumentation of the building envelope to gather ``big" data to train a deep neural network. While theoretical models are often limited by their estimation accuracy, it is not always feasible to gather a significant amount of field data. In this paper, we explore transfer learning-based strategies to train much more accurate model for energy estimation when using a sparse field data. We transferred knowledge, in the form of data and parameters, from the simulation framework to the field data. We evaluated the efficacy of our approach on field data collected from six commercial buildings and our results indicate that transfer learning-based models trained over one month data can perform comparative (and in some cases better) than the state-of-the-art machine learning and deep learning solutions.

Jain, Milan↗
DOE OSTI2020

Deep Learning Emulation of Subgrid-Scale Processes in Turbulent Shear Flows

Deep neural networks (DNNs) are developed from a data set obtained from the dynamic Smagorinsky model to emulate the subgrid-scale (SGS) viscosity (ν sgs ) and diffusivity (κ sgs ) for turbulent stratified shear flows encountered in the oceans and the atmosphere. These DNNs predict ν sgs and κ sgs from velocities, strain rates, and density gradients such that the evolution of the kinetic energy budget and density variance budget terms is similar to the corresponding values obtained from the original dynamic Smagorinsky model. These DNNs also compute ν sgs and κ sgs ~2–4 times quicker than the dynamic Smagorinsky model resulting in a ~2–2.5 times acceleration of the entire simulation. This study demonstrates the feasibility of deep learning in emulating the subgrid-scale (SGS) phenomenon in geophysical flows accurately in a cost-effective manner. In a broader perspective, deep learning-based surrogate models can present a promising alternative to the traditional parameterizations of the subgrid-scale processes in climate models.

58 GEOSCIENCES↗
DOE OSTI2021

Short- and medium-range orders in Al90Tb10 glass and their relation to the structures of competing crystalline phases

Molecular dynamics simulations using an interatomic potential developed by artificial neural network deep machine learning are performed to study the local structural order in Al90Tb10 metallic glass. We show that more than 80% of the Tb-centered clusters in Al90Tb10 glass have short-range order (SRO) with their 17 first coordination shell atoms stacked in a ‘3661’ or ‘15551’ sequence. Medium-range order (MRO) in Bergman-type packing extended out to the second and third coordination shells is also clearly observed. Analysis of the network formed by the ‘3661’ and ‘15551’ clusters show that ~82% of such SRO units share their faces or vertexes, while only ~6% of neighboring SRO pairs are interpenetrating. Such a network topology is consistent with the Bergman-type MRO around the Tb-centers. Moreover, crystal structure searches using genetic algorithm and the neural network interatomic potential reveal several low-energy metastable crystalline structures in the composition range close to Al 90 Tb 10 . Some of these crystalline structures have the ‘3661’ SRO while others have the ‘15551’ SRO. While the crystalline structures with the ‘3661’ SRO also exhibit the MRO very similar to that observed in the glass, the ones with the ‘15551’ SRO have very different atomic packing in the second and third shells around the Tb centers from that of the Bergman-type MRO observed in the glassy phase.

36 MATERIALS SCIENCE↗
DOE OSTI2024

CrossMP: Enabling Cross-Modality Translation between Single-Cell RNA-Seq and Single-Cell ATAC-Seq through Web-Based Portal

In recent years, there has been a growing interest in profiling multiomic modalities within individual cells simultaneously. One such example is integrating combined single-cell RNA sequencing (scRNA-seq) data and single-cell transposase-accessible chromatin sequencing (scATAC-seq) data. Integrated analysis of diverse modalities has helped researchers make more accurate predictions and gain a more comprehensive understanding than with single-modality analysis. However, generating such multimodal data is technically challenging and expensive, leading to limited availability of single-cell co-assay data. Here, we propose a model for cross-modal prediction between the transcriptome and chromatin profiles in single cells. Our model is based on a deep neural network architecture that learns the latent representations from the source modality and then predicts the target modality. It demonstrates reliable performance in accurately translating between these modalities across multiple paired human scATAC-seq and scRNA-seq datasets. Additionally, we developed CrossMP, a web-based portal allowing researchers to upload their single-cell modality data through an interactive web interface and predict the other type of modality data, using high-performance computing resources plugged at the backend.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

Exploration of Domain Aware Machine Learning for Grid Analytics: Transfer-Learnt Energy Models to Assist Buildings Control with Sparse Field Data

Buildings are a primary consumer of energy in the United States and are also increasingly being perceived as providers of grid services such as load shifting, shedding and modulation. High fidelity models of building energy consumption are needed to set appropriate baselines for measurement and verification (M&V) of controllers designed for energy efficient operation of buildings and to enable buildings to provide grid services via. participation in demand response programs. State-of-the-art building energy modeling techniques either rely on Physics based models, or extensive instrumentation of the building envelope to gather “big” data to train machine learning based models such as deep neural networks. While Physics based models are often limited by their accuracy, it is not always feasible to gather a significant amount of field data required to train machine learning based models with sufficient accuracy. In this paper, we explore the use of transfer learning-based strategies to address unsatisfactory accuracy of models for estimating building energy consumption when available field data for training is sparse or of unacceptable quality. In particular, we transfer knowledge in the form of data and parameters, from Physics based simulation frameworks to the field to improve the model accuracy, thus resulting in a Physics-informed Machine Learning framework. We evaluated the efficacy of our approach on field data collected from six commercial buildings and our results indicate that the proposed transfer learning based models provide comparative (and in some cases better) accuracy than state-of-the-art machine learning and deep learning solutions, with just one month of field data.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2023

Deep learning uncertainty quantification for clinical text classification

Machine learning algorithms are expected to work side-by-side with humans in decision-making pipelines. Thus, the ability of classifiers to make reliable decisions is of paramount importance. Deep neural networks (DNNs) represent the state-of-the-art models to address real-world classification. Although the strength of activation in DNNs is often correlated with the network’s confidence, in-depth analyses are needed to establish whether they are well calibrated. In this paper, we demonstrate the use of DNN-based classification tools to benefit cancer registries by automating information extraction of disease at diagnosis and at surgery from electronic text pathology reports from the US National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) population-based cancer registries. In particular, we introduce multiple methods for selective classification to achieve a target level of accuracy on multiple classification tasks while minimizing the rejection amount—that is, the number of electronic pathology reports for which the model’s predictions are unreliable. We evaluate the proposed methods by comparing our approach with the current in-house deep learning-based abstaining classifier. Overall, all the proposed selective classification methods effectively allow for achieving the targeted level of accuracy or higher in a trade-off analysis aimed to minimize the rejection rate. On in-distribution validation and holdout test data, with all the proposed methods, we achieve on all tasks the required target level of accuracy with a lower rejection rate than the deep abstaining classifier (DAC). Interpreting the results for the out-of-distribution test data is more complex; nevertheless, in this case as well, the rejection rate from the best among the proposed methods achieving 97% accuracy or higher is lower than the rejection rate based on the DAC. We show that although both approaches can flag those samples that should be manually reviewed and labeled by human annotators, the newly proposed methods retain a larger fraction and do so without retraining—thus offering a reduced computational cost compared with the in-house deep learning-based abstaining classifier.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

LENS: Learning Enabled Network Synthesis

RTRC and UMD have developed novel machine learning based methods under the ARPA-E DIFFERENTIATE program for rapid acceleration of hypothesis generation in complex architecture design spaces involving both discrete choices of component inclusion and interconnection and continuous parametric decisions. The project named Learning Enabled Network Synthesis (LENS) further demonstrated the developed methods on challenging electrical power converter design problems by identifying the most suitable circuit topologies and simultaneously selecting the most appropriate components to achieve optimized design of power converter with improved performances. We demonstrated that LENS could enable exploration of very large design space of circuit topologies and components by addressing the limitations of conventional design process in non-linear, high switching speed, multi-dimensional power converter design and optimization. The key innovation developed in LENS is the seamless integration of statistical learning and logical reasoning techniques and building on the individual strengths of these techniques for rapid hypothesis discovery. The main component of LENS comprises of: 1) Graph Reasoning Engine (GRE) to enforce composition rules that rapidly reject all discrete architectures that are composed incorrectly and generates an adaptive database of feasible designs which can be used by ML modules, 2) Graph Generative Learning module which is a deep neural network based generative model for graph architectures which can enable design space exploration beyond the dataset generated by the GRE, 3) Graph Reduced Order Model (ROM) for graph domains for accelerating computation of output metrics, and 4) Active learning and Rule Discovery module for sample efficient learning and extracting logical rules from the learned ML models which will be integrated in the GRE to enhance the filtering effectiveness. LENS approach can be applied to any design domains where designs can be represented as multi-attribute graphs. The LENS team integrated the various technical innovations listed above into an optimization pipeline and exercised the optimization pipeline on the converter design problem. The LENS project demonstrated that the developed AI/ML technologies can be used to generate novel converter circuits >45x faster than experts on chosen use-cases. This can enable faster design space exploration and identification of new designs which are not considered by experts due to the increasing design space complexity. This has significant potential impact on the public and energy needs of the country. It is currently estimated that 30% of all electrical powers generated passes through power converters. The future estimate is that 80% of all power generated would be passing through converters. LENS fills a critical gap in this space since by accelerating the design process the designers would be able to generate more efficient converters which can lead to significant energy savings for the country.

42 ENGINEERING↗
DOE OSTI2025

Toward Drilling the Perfect Geothermal Well: An International Research Coordination Network for Geothermal Drilling Optimization Supported by Deep Machine Learning and Cloud Based Data Aggregation

The EDGE project, supported by the U.S. Department of Energy Geothermal Technologies Office under award DE-EE0008793, established a data-driven framework for improving the efficiency, cost-effectiveness, and reliability of geothermal well drilling. The project focused on developing scalable data infrastructure, advanced machine learning and probabilistic models, and integrated analytics tools to support continuous drilling optimization. A central objective was to reduce geothermal drilling costs by up to seventy percent while minimizing the risk of well failure through predictive diagnostics and adaptive planning. Over the project period, a comprehensive data repository was designed and deployed, incorporating records from over one hundred geothermal wells across varied geological settings. This repository supported both structured and unstructured data and adhered to FAIR data principles, enabling provenance tracking, quality control, and standardized metadata. The project introduced automated ingestion pipelines and a cloud-hosted platform that facilitated access to raw, processed, and derived datasets. This infrastructure served as the foundation for model development and analysis. Machine learning workflows were developed to predict key drilling metrics including rate of penetration, non-productive time, and total drilling costs. Self-organizing maps and dimensionality reduction methods were used to uncover operational patterns and outliers, while supervised learning algorithms such as random forests and deep neural networks were applied to forecast performance outcomes. The models were validated on heterogeneous datasets from both U.S. and Icelandic fields, demonstrating variable but significant predictive accuracy. The results indicated that finer temporal resolution, inclusion of lithological data, and consistency in operational annotations could substantially improve model performance. The project also implemented process mining techniques to reconstruct state-transition models from drilling event logs. These models enabled the identification of deviations from optimal workflows and provided insights into recurring failure modes. Analysis of non-productive time highlighted the impact of equipment failures, geological challenges, and human factors, offering opportunities for targeted mitigation strategies. The EDGE Dashboard was developed as a web-based expert system integrating data visualization, model outputs, and user-driven queries. It provided an accessible interface for operators to explore historical data, evaluate predicted outcomes, and compare drilling scenarios. Initial feedback from project partners suggested that the dashboard could serve as a foundation for more advanced advisory and optimization tools. Overall, the EDGE project demonstrated the feasibility and value of applying modern data science techniques to geothermal drilling. It delivered a set of interoperable tools and models that can support more efficient, lower-risk well development. The findings point toward a viable path for transitioning from advisory analytics to semi-autonomous drilling systems, contingent on continued collaboration, expanded datasets, and field validation. The project results have immediate relevance for drilling operations, data management practices, and future geothermal R&D efforts aimed at achieving reliable, cost-competitive geothermal energy at scale.

15 GEOTHERMAL ENERGY↗
DOE OSTI2021

Prediction of Solar Irradiance and Photovoltaic Solar Energy Product Based on Cloud Coverage Estimation Using Machine Learning Methods

Cloud cover estimation from images taken by sky-facing cameras can be an important input for analyzing current weather conditions and estimating photovoltaic power generation. The constant change in position, shape, and density of clouds, however, makes the development of a robust computational method for cloud cover estimation challenging. Accurately determining the edge of clouds and hence the separation between clouds and clear sky is difficult and often impossible. Toward determining cloud cover for estimating photovoltaic output, we propose using machine learning methods for cloud segmentation. We compare several methods including a classical regression model, deep learning methods, and boosting methods that combine results from the other machine learning models. To train each of the machine learning models with various sky conditions, we supplemented the existing Singapore whole sky imaging segmentation database with hazy and overcast images collected by a camera-equipped Waggle sensor node. We found that the U-Net architecture, one of the deep neural networks we utilized, segmented cloud pixels most accurately. However, the accuracy of segmenting cloud pixels did not guarantee high accuracy of estimating solar irradiance. We confirmed that the cloud cover ratio is directly related to solar irradiance. Additionally, we confirmed that solar irradiance and solar power output are closely related; hence, by predicting solar irradiance, we can estimate solar power output. This study demonstrates that sky-facing cameras with machine learning methods can be used to estimate solar power output. This ground-based approach provides an inexpensive way to understand solar irradiance and estimate production from photovoltaic solar facilities.

14 SOLAR ENERGY↗
DOE OSTI2022

Long-time integration of parametric evolution equations with physics-informed DeepONets

Ordinary and partial differential equations (ODEs/PDEs) play a paramount role in analyzing and simulating complex dynamic processes across all corners of science and engineering. In recent years machine learning tools are aspiring to introduce new effective ways of simulating such equations, however existing approaches are not able to reliably return stable and accurate predictions across long temporal horizons. We aim to address this challenge by introducing an effective framework for learning evolution operators that map random initial conditions to associated ODE/PDE solutions within a short time interval. Such operators can be parametrized by deep neural networks that are trained in an entirely self-supervised manner without requiring one to generate any paired input-output observations. Global long-time predictions across a range of initial conditions can be then obtained by iteratively evaluating the trained model using each prediction as the initial condition for the next evaluation step. Here, this introduces a new approach to temporal domain decomposition that is shown to be effective in performing accurate long-time simulations for a wide range of parametric ODE and PDE systems, from wave propagation, to reaction-diffusion dynamics and stiff chemical kinetics, introducing a new way of rapidly emulating non-equilibrium processes in science and engineering.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Computer Model Calibration with Time Series Data Using Deep Learning and Quantile Regression

Computer models play a key role in many scientific and engineering problems. One major source of uncertainty in computer model experiments is input parameter uncertainty. Computer model calibration is a formal statistical procedure to infer input parameters by combining information from model runs and observational data. The existing standard calibration framework suffers from inferential issues when the model output and observational data are high-dimensional dependent data, such as large time series, due to the difficulty in building an emulator and the nonidentifiability between effects from input parameters and data-model discrepancy. To overcome these challenges, we, in this work, propose a new calibration framework based on a deep neural network (DNN) with long short-term memory layers that directly emulates the inverse relationship between the model output and input parameters. Adopting the “learning with noise” idea, we train our DNN model to filter out the effects from data-model discrepancy on input parameter inference. We also formulate a new way to construct interval predictions for DNN using quantile regression to quantify the uncertainty in input parameter estimates. Through a simulation study and real data application with the Weather Research and Forecasting Model Hydrological modeling system (WRF-Hydro), we show our approach can yield accurate point estimates and well-calibrated interval estimates for input parameters.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

GRIDS-Net: Inverse shape design and identification of scatterers via geometric regularization and physics-embedded deep learning

This study presents a deep learning based methodology for both remote sensing and design of acoustic scatterers. The ability to determine the shape of a scatterer, either in the context of material design or sensing, plays a critical role in many practical engineering problems. This class of inverse problems is extremely challenging due to their high-dimensional, nonlinear, and ill-posed nature. To overcome these technical hurdles, we introduce a geometric regularization approach for deep neural networks (DNN) based on non-uniform rational B-splines (NURBS) and capable of predicting complex 2D scatterer geometries in a parsimonious dimensional representation. Then, this geometric regularization is combined with physics-embedded learning and integrated within a robust convolutional autoencoder (CAE) architecture to accurately predict the shape of 2D scatterers in the context of identification and inverse design problems. Further, an extensive numerical study is presented in order to showcase the remarkable ability of this approach to handle complex scatterer geometries while generating physically-consistent acoustic fields. The study also assesses and contrasts the role played by the (weakly) embedded physics in the convergence of the DNN predictions to a physically consistent inverse design.

42 ENGINEERING↗
DOE OSTI2022

Data-driven modeling of power generation for a coal power plant under cycling

Increased penetration of renewables for power generation has negatively impacted the dynamics of conventional fossil fuel-based power plants. The power plants operating on the base load are forced to cycle, to adjust to the fluctuating power demands. This results in an inefficient operation of the coal power plants, which leads up to higher operating losses. To overcome such operational challenge associated with cycling and to develop an optimal process control, this work analyzes a set of models for predicting power generation. Moreover, the power generation is intrinsically affected by the state of the power plant components, and therefore our model development also incorporates additional power plant process variables while forecasting the power generation. We present and compare multiple state-of-the-art forecasting data-driven methods for power generation to determine the most adequate and accurate model. We also develop an interpretable attention-based transformer model to explain the importance of process variables during training and forecasting. The trained deep neural network (DNN) LSTM model has good accuracy in predicting gross power generation under various prediction horizons with/without cycling events and outperforms the other models for long-term forecasting. The DNN memory-based models show significant superiority over other state-of-the-art machine learning models for short, medium and long range predictions. The transformer-based model with attention enhances the selection of historical data for multi-horizon forecasting, and also allows to interpret the significance of internal power plant components on the power generation. This newly gained insights can be used by operation engineers to anticipate and monitor the health of power plant equipment during high cycling periods.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2021

Encoding Frequency Constraints in Preventive Unit Commitment Using Deep Learning With Region-of-Interest Active Sampling

With the increasing penetration of renewable energy, frequency response and its security are of significant concerns for reliable power system operations. Frequency-constrained unit commitment (FCUC) is proposed to address this challenge. Despite existing efforts in modeling frequency characteristics in unit commitment (UC), current strategies can only handle oversimplified low-order frequency response models and do not consider wide-range operating conditions. This paper presents a generic data-driven framework for FCUC under high renewable penetration. Here, deep neural networks (DNNs) are trained to predict the frequency response using real data or high-fidelity simulation data. Next, the DNN is reformulated as a set of mixed-integer linear constraints to be incorporated into the ordinary UC formulation. In the data generation phase, all possible power injections are considered, and a region-of-interest active sampling is proposed to include power injection samples with frequency nadirs closer to the UFLC threshold, which enhances the accuracy of frequency constraints in FCUC. The proposed FCUC is investigated on the IEEE 39-bus system. Then, a full-order dynamic model simulation using PSS/E verifies the effectiveness of FCUC in frequency-secure generator commitments.

42 ENGINEERING↗
DOE OSTI2019

Phase Segmentation in Atom-Probe Tomography Using Deep Learning-Based Edge Detection

Abstract Atom-probe tomography (APT) facilitates nano- and atomic-scale characterization and analysis of microstructural features. Specifically, APT is well suited to study the interfacial properties of granular or heterophase systems. Traditionally, the identification of the interface between, for precipitate and matrix phases, in APT data has been obtained either by extracting iso-concentration surfaces based on a user-supplied concentration value or by manually perturbing the concentration value until the iso-concentration surface qualitatively matches the interface. These approaches are subjective, not scalable, and may lead to inconsistencies due to local composition inhomogeneities. We introduce a digital image segmentation approach based on deep neural networks that transfer learned knowledge from natural images to automatically segment the data obtained from APT into different phases. This approach not only provides an efficient way to segment the data and extract interfacial properties but does so without the need for expensive interface labeling for training the segmentation model. We consider here a system with a precipitate phase in a matrix and with three different interface modalities—layered, isolated, and interconnected—that are obtained for different relative geometries of the precipitate phase. We demonstrate the accuracy of our segmentation approach through qualitative visualization of the interfaces, as well as through quantitative comparisons with proximity histograms obtained by using more traditional approaches.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

A cross-study analysis of drug response prediction in cancer cell lines

Abstract To enable personalized cancer treatment, machine learning models have been developed to predict drug response as a function of tumor and drug features. However, most algorithm development efforts have relied on cross-validation within a single study to assess model accuracy. While an essential first step, cross-validation within a biological data set typically provides an overly optimistic estimate of the prediction performance on independent test sets. To provide a more rigorous assessment of model generalizability between different studies, we use machine learning to analyze five publicly available cell line-based data sets: National Cancer Institute 60, ancer Therapeutics Response Portal (CTRP), Genomics of Drug Sensitivity in Cancer, Cancer Cell Line Encyclopedia and Genentech Cell Line Screening Initiative (gCSI). Based on observed experimental variability across studies, we explore estimates of prediction upper bounds. We report performance results of a variety of machine learning models, with a multitasking deep neural network achieving the best cross-study generalizability. By multiple measures, models trained on CTRP yield the most accurate predictions on the remaining testing data, and gCSI is the most predictable among the cell line data sets included in this study. With these experiments and further simulations on partial data, two lessons emerge: (1) differences in viability assays can limit model generalizability across studies and (2) drug diversity, more than tumor diversity, is crucial for raising model generalizability in preclinical screening.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Deep-Learning-derived Boundary Layer Height from Meteorological Data over the SGP, GOAMAZON, CACTI

The planetary boundary-layer (PBL) height (PBLH) is an important parameter for various meteorological and climate studies. This study presents a multi-structure deep neural network (DNN) model, designed to estimate PBLH by integrating morning temperature profiles with surface meteorological observations. The DNN model is developed by leveraging a rich data set of PBLH derived from long-standing radiosonde records and augmented with high-resolution micropulse lidar and Doppler lidar observations. We access the performance of the DNN with an ensemble of 10 members, each featuring distinct hidden layer structures, which collectively yield a robust 27-year PBLH data set over the Southern Great Plains from 1994 to 2020. The influence of various meteorological factors on PBLH is rigorously analyzed through the importance test. Moreover, the DNN model's accuracy is evaluated against radiosonde observations and juxtaposed with conventional remote-sensing methodologies, including Doppler lidar, ceilometer, Raman lidar, and micropulse lidar. The DNN model exhibits reliable performance across diverse conditions and demonstrates lower biases relative to remote-sensing methods. In addition, the DNN model, originally trained over a plain region, demonstrates remarkable adaptability when applied to the heterogeneous terrains and climates encountered during the GoAmazon (tropical rainforest) and CACTI (middle-latitude mountain) campaigns. These findings demonstrate the effectiveness of deep learning models in estimating PBLH, enhancing our understanding of boundary-layer dynamics with implications for enhancing the representation of PBL in weather forecasting and climate modeling.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Efficient learning of power grid voltage control strategies via model-based deep reinforcement learning

Here this article proposes a model-based deep reinforcement learning (DRL) method to design emergency control strategies for short-term voltage stability problems in power systems. Recent advances show promising results for model-free DRL-based methods in power systems control problems. But in power systems applications, these model-free methods have certain issues related to training time (clock time) and sample efficiency; both are critical for making state-of-the-art DRL algorithms practically applicable. DRL-agent learns an optimal policy via a trial-and-error method while interacting with the real-world environment. It is also desirable to minimize the direct interaction of the DRL agent with the real-world power grid due to its safety-critical nature. Additionally, the state-of-the-art DRL-based policies are mostly trained using a physics-based grid simulator where dynamic simulation is computationally intensive, lowering the training efficiency. We propose a novel model-based DRL framework where a deep neural network (DNN)-based dynamic surrogate model (SM), instead of a real-world power grid or physics-based simulation, is utilized within the policy learning framework, making the process faster and more sample efficient. However, having stable training in model-based DRL is challenging because of the complex system dynamics of large-scale power systems. We addressed these issues by incorporating imitation learning to have a warm start in policy learning, reward-shaping, and multi-step loss in surrogate model training. Finally, we achieved 97.5% reduction in samples and 87.7% reduction in training time for an application to the IEEE 300-bus test system.

24 POWER TRANSMISSION AND DISTRIBUTION↗