Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Algorithmic Optimization”

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 199 records · Page 11
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Optimal band selection for target detection with a LWIR multispectral imager

Multispectral imaging can offer many benefits in cost, complexity, resolution, size, weight, and power, relative to hyperspectral imaging. When designing a multispectral system, spectral bandpasses can be selected using optimization algorithms configured to maximally separate target detection scores between target and background regions. A hyperspectral image (HSI) can serve as the source of data from which band groupings can be tested for optimality. The output of an adaptive cosine estimator target detection algorithm is used in an objective function. Three optimization algorithms are compared: particle swarm, dual annealing, and differential evolution. A global optimum is also found using a brute force approach on the Livermore Computing Syrah supercomputer. Three materials are investigated: calcite, gypsum, and limestone. This is done for 3-, 4-, and 5-band systems. The data originate from a longwave infrared HSI of a material display board. The optimization algorithms were run 30 times for every scenario. Performance statistics (maximum, minimum, mean, standard deviation, and median) based on the separation values are given. Additional characterization was performed using receiver operator characteristic (ROC) curves and the area under the ROC curve. While good performance was obtained for the three optimization algorithms, the dual annealing algorithm produced the highest and most consistent detection separation scores on average.

47 OTHER INSTRUMENTATION↗
DOE OSTI2016

NEUTRON STAR MASS–RADIUS CONSTRAINTS USING EVOLUTIONARY OPTIMIZATION

The equation of state of cold supra-nuclear-density matter, such as in neutron stars, is an open question in astrophysics. A promising method for constraining the neutron star equation of state is modeling pulse profiles of thermonuclear X-ray burst oscillations from hot spots on accreting neutron stars. The pulse profiles, constructed using spherical and oblate neutron star models, are comparable to what would be observed by a next-generation X-ray timing instrument like ASTROSAT , NICER , or a mission similar to LOFT . In this paper, we showcase the use of an evolutionary optimization algorithm to fit pulse profiles to determine the best-fit masses and radii. By fitting synthetic data, we assess how well the optimization algorithm can recover the input parameters. Multiple Poisson realizations of the synthetic pulse profiles, constructed with 1.6 million counts and no background, were fitted with the Ferret algorithm to analyze both statistical and degeneracy-related uncertainty and to explore how the goodness of fit depends on the input parameters. For the regions of parameter space sampled by our tests, the best-determined parameter is the projected velocity of the spot along the observer’s line of sight, with an accuracy of ≤3% compared to the true value and with ≤5% statistical uncertainty. The next best determined are the mass and radius; for a neutron star with a spin frequency of 600 Hz, the best-fit mass and radius are accurate to ≤5%, with respective uncertainties of ≤7% and ≤10%. The accuracy and precision depend on the observer inclination and spot colatitude, with values of ∼1% achievable in mass and radius if both the inclination and colatitude are ≳60°.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2015

SU-E-T-332: Dosimetric Impact of Photon Energy and Treatment Technique When Knowledge Based Auto-Planning Is Implemented in Radiotherapy of Localized Prostate Cancer

Purpose: The aim of this study was to investigate the dosimetric impact of the combination of photon energy and treatment technique on radiotherapy of localized prostate cancer when knowledge based planning was used. Methods: A total of 16 patients with localized prostate cancer were retrospectively retrieved from database and used for this study. For each patient, four types of treatment plans with different combinations of photon energy (6X and 10X) and treatment techniques (7-field IMRT and 2-arc VMAT) were created using a prostate DVH estimation model in RapidPlan™ and Eclipse treatment planning system (Varian Medical System). For any beam arrangement, DVH objectives and weighting priorities were generated based on the geometric relationship between the OAR and PTV. Photon optimization algorithm was used for plan optimization and AAA algorithm was used for final dose calculation. Plans were evaluated in terms of the pre-defined dosimetric endpoints for PTV, rectum, bladder, penile bulb, and femur heads. A Student’s paired t-test was used for statistical analysis and p > 0.05 was considered statistically significant. Results: For PTV, V95 was statistically similar among all four types of plans, though the mean dose of 10X plans was higher than that of 6X plans. VMAT plans showed higher heterogeneity index than IMRT plans. No statistically significant difference in dosimetry metrics was observed for rectum, bladder, and penile bulb among plan types. For left and right femur, VMAT plans had a higher mean dose than IMRT plans regardless of photon energy, whereas the maximum dose was similar. Conclusion: Overall, the dosimetric endpoints were similar regardless of photon energy and treatment techniques when knowledge based auto planning was used. Given the similarity in dosimetry metrics of rectum, bladder, and penile bulb, the genitourinary and gastrointestinal toxicities should be comparable among the selections of photon energy and treatment techniques.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

A modular GUI-based program for genetic algorithm-based feedback-assisted wavefront shaping

Abstract We have developed a modular graphical user interface (GUI)-based program for use in genetic algorithm-based feedback-assisted wavefront shaping. The program uses a class-based structure to separate out the universal modules (e.g. GUI, multithreading, optimization algorithms) and hardware-specific modules (e.g. code for different SLMs and cameras). This modular design makes the program easily adaptable to a wide range of lab equipment, while providing easy access to a GUI, multithreading, and three optimization algorithms (phase-stepping, simple genetic, and microgenetic).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Parallel Memory-Independent Communication Bounds for SYRK

In this paper, we focus on the parallel communication cost of multiplying a matrix with its transpose, known as a symmetric rank-k update (SYRK). SYRK requires half the computation of general matrix multiplication because of the symmetry of the output matrix. Recent work (Beaumont et al., SPAA '22) has demonstrated that the sequential I/O complexity of SYRK is also a constant factor smaller than that of general matrix multiplication. Inspired by this progress, we establish memory-independent parallel communication lower bounds for SYRK with smaller constants than general matrix multiplication, and we show that these constants are tight by presenting communication-optimal algorithms. The crux of the lower bound proof relies on extending a key geometric inequality to symmetric computations and analytically solving a constrained nonlinear optimization problem. Here, the optimal algorithms use a triangular blocking scheme for parallel distribution of the symmetric output matrix and corresponding computation.

Communication costs↗
DOE OSTI2025

Extreme Longitudinal Compression of Optimized Beams for MEV Ultrafast Electron Diffraction (Final Technical Report)

We worked out the design of a high repetition rate MeV energy ultrafast electron diffraction instrument based on the existing Cornell photoinjector, which can readily be applied to the presented findings. This example is a blueprint of other similarly arranged UED setups. Using particle tracking simulations in conjunction with multiobjective genetic algorithm optimization, we explored the smallest bunch lengths, emittance, and probe spot sizes achievable. As two limits, we defined stroboscopic conditions (with single electrons per pulse) and operation with 10 5 electrons per bunch which may be suitable for single-shot diffraction images. In the stroboscopic case, the flexibility provided by the many cavity bunching and acceleration allows for longitudinal phase space linearization without a higher harmonic field, providing sub-fs bunch lengths at the sample. Given low emittance photoemission conditions, these small bunch lengths can be maintained with probe transverse sizes at the single micron (1 μm) scale and below. In the case of 10 5 electrons per pulse, we simulated state-of-the-art 5D brightness conditions: rms bunch lengths of 10 fs with 3-nm normalized emittances, while permitting repetition rates as high as 1.3 GHz. We showed that in conjunction with collimating apertures, a novel focusing scheme achieves very high-quality emittance compensation for the central core of the beam composing 40% of particles, for a resulting beam size of 5 μm (rms). Finally, to aid in the design of new SRF-based ultrafast electron diffraction machines, we simulated the trade-off between the number of cavities used and achievable bunch length and emittance. In the longitudinal dimension, we made use of the fact that MeV UED requires much lower energy than the 15-MeV maxi mum energy of Cornell’s CBETA injector, and we may therefore use several of the SRF cavities for bunch length compression. In practice, we used a genetic optimization algorithm to choose the phases and amplitudes of the cavities appropriately for optimal bunching. In the zero space charge case, we found that bunching and acceleration are distributed across the six cavities in a way that produces a linearizing effect. And we showed that the ultimate bunch length can be limited by time-of-flight differences arising from transverse size and transverse momentum spread. The space charge code developed and used for this development is now permanent part of the Bmad accelerator simulation code and has already contributed to other developments, e.g., for the EIC electron cooler design.

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Leveraging prior mean models for faster Bayesian optimization of particle accelerators

Tuning particle accelerators is a challenging and time-consuming task that can be automated and carried out efficiently using suitable optimization algorithms, such as model-based Bayesian optimization techniques. One of the major advantages of Bayesian algorithms is the ability to incorporate prior information about beam physics and historical behavior into the model used to make control decisions. In this work, we examine incorporating prior accelerator physics information into Bayesian optimization algorithms by utilizing fast executing, neural network models trained on simulated or historical datasets as prior mean functions in Gaussian process models. We show that in ideal cases, this technique substantially increases convergence speed to optimal solutions in high-dimensional tuning parameter spaces. Additionally, we demonstrate that even in non-ideal cases, where prior models of beam dynamics do not exactly match experimental conditions, the use of this technique can still enhance convergence speed. Finally, we demonstrate how these methods can be used to improve optimization in practical applications, such as transferring information gained from beam dynamics simulations to online control of the LCLS injector, and transferring knowledge gained from experimental measurements across different operating modes, such as accelerating different ion species at the ATLAS heavy ion accelerator.

43 PARTICLE ACCELERATORS↗
DOE OSTI2016

SU-F-T-342: Dosimetric Constraint Prediction Guided Automatic Mulit-Objective Optimization for Intensity Modulated Radiotherapy

Purpose: For intensity modulated radiotherapy, the plan optimization is time consuming with difficulties of selecting objectives and constraints, and their relative weights. A fast and automatic multi-objective optimization algorithm with abilities to predict optimal constraints and manager their trade-offs can help to solve this problem. Our purpose is to develop such a framework and algorithm for a general inverse planning. Methods: There are three main components contained in this proposed multi-objective optimization framework: prediction of initial dosimetric constraints, further adjustment of constraints and plan optimization. We firstly use our previously developed in-house geometry-dosimetry correlation model to predict the optimal patient-specific dosimetric endpoints, and treat them as initial dosimetric constraints. Secondly, we build an endpoint(organ) priority list and a constraint adjustment rule to repeatedly tune these constraints from their initial values, until every single endpoint has no room for further improvement. Lastly, we implement a voxel-independent based FMO algorithm for optimization. During the optimization, a model for tuning these voxel weighting factors respecting to constraints is created. For framework and algorithm evaluation, we randomly selected 20 IMRT prostate cases from the clinic and compared them with our automatic generated plans, in both the efficiency and plan quality. Results: For each evaluated plan, the proposed multi-objective framework could run fluently and automatically. The voxel weighting factor iteration time varied from 10 to 30 under an updated constraint, and the constraint tuning time varied from 20 to 30 for every case until no more stricter constraint is allowed. The average total costing time for the whole optimization procedure is ∼30mins. By comparing the DVHs, better OAR dose sparing could be observed in automatic generated plan, for 13 out of the 20 cases, while others are with competitive results. Conclusion: We have successfully developed a fast and automatic multi-objective optimization for intensity modulated radiotherapy. This work is supported by the National Natural Science Foundation of China (No: 81571771)

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

An ILP based Algorithm for Optimal Customer Selection for Demand Response in SmartGrids

Demand Response (DR) events are initiated by utilities during peak demand periods to curtail consumption. They ensure system reliability and minimize the utility’s expenditure. Selection of the right customers and strategies is critical for a DR event. An effective DR scheduling algorithm minimizes the curtailment error which is the absolute difference between the achieved curtailment value and the target. State-of-the-art heuristics exist for customer selection, however their curtailment errors are unbounded and can be as high as 70%. In this work, we develop an Integer Linear Programming (ILP) formulation for optimally selecting customers and curtailment strategies that minimize the curtailment error during DR events in SmartGrids. We perform experiments on real world data obtained from the University of Southern California’s SmartGrid and show that our algorithm achieves near exact curtailment values with errors in the range of 10 -7 to 10 -5 , which are within the range of numerical errors. We compare our results against the state-of-the-art heuristic being deployed in practice in the USC SmartGrid. We show that for the same set of available customer strategy pairs our algorithm performs 103 to 107 times better in terms of the curtailment errors incurred.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

SU-E-T-225: Correction Matrix for PinPoint Ionization Chamber for Dosimetric Measurements in the Newly Released Incise™ Multileaf Collimator Shaped Small Field for CyberKnife M6™ Machine

Purpose: For small field dosimetry, such as measurements of output factors for cones or MLC-shaped irregular small fields, ion chambers often Result in an underestimation of the dose, due to both the volume averaging effect and the lack of lateral charged particle equilibrium. This work presents a mathematical model for correction matrix for a PTW PinPoint ionization chamber for dosimetric measurements made in the newly released Incise™ Multileaf collimator fields of the CyberKnife M6™ machine. Methods: A correction matrix for a PTW 0.015cc PinPoint ionization chamber was developed by modeling its 3D dose response in twelve cone-shaped circular fields created using the 5mm, 7.5mm, 10mm, 12.5mm, 15mm, 20mm, 25mm, 30mm, 35mm, 40mm, 50mm, 60mm cones in a CyberKnife M6™ machine. For each field size, hundreds of readings were recorded for every 2mm chamber shift in the horizontal plane. The contribution of each dose pixel to a measurement point depended on the radial distance and the angle to the chamber axis. These readings were then compared with the theoretical dose as obtained with Monte Carlo calculation. A penalized least-square optimization algorithm was developed to generate the correction matrix. After the parameter fitting, the mathematical model was validated for MLC-shaped irregular fields. Results: The optimization algorithm used for parameter fitting was stable and the resulted response factors were smooth in spatial domain. After correction with the mathematical model, the chamber reading matched with the calculation for all the tested fields to within 2%. Conclusion: A novel mathematical model has been developed for PinPoint chamber for dosimetric measurements in small MLC-shaped irregular fields. The correction matrix is dependent on detector, treatment unit and the geometry of setup. The model can be applied to non-standard composite fields and provides an access to IMRT point dose validation.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2020

Assessing Historical Variability of South Asian Monsoon Lows and Depressions With an Optimized Tracking Algorithm

Abstract Cyclonic low‐pressure systems (LPS) produce abundant rainfall in South Asia, where they are traditionally categorized as monsoon lows, monsoon depressions, and more intense cyclonic storms. The India Meteorological Department (IMD) has tracked monsoon depressions for over a century, finding a large decline in their number in recent decades, but their methods have changed over time and do not include monsoon lows. This study presents a fast, objective algorithm for identifying monsoon LPS and uses it to assess interannual variability and trends in reanalyses. Variables and thresholds used in the algorithm are selected to best match a subjectively analyzed LPS data set while minimizing disagreement between four reanalyses in a training period. The stream function of 850 hPa horizontal wind is found to be optimal in this sense; it is less noisy than vorticity and represents the complete nondivergent wind, even when flow is not geostrophic. Using this algorithm, LPS statistics are computed for five reanalyses, and none show a detectable trend in monsoon depression counts since 1979. Both the Japanese 55‐year Reanalysis (JRA‐55) and the IMD data set show a step‐like reduction in depression counts when they began using geostationary satellite data, in 1979 and 1982, respectively; the 1958–2018 linear trend in JRA‐55, however, is smaller than in the IMD data set, and its error bar includes 0. There are more LPS in seasons with above‐average monsoon rainfall and in La Niña years, but few other large‐scale modes of interannual variability are found to modulate LPS counts, lifetimes, or track length consistently across reanalyses.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Towards Robustness Guarantees for Feedback-Based Optimization

Feedback-based online optimization algorithms have gained traction in recent years because of their simple implementation, their ability to reject disturbances in real time, and their increased robustness to model mismatch. While the robustness properties have been observed both in simulation and experimental results, the theoretical analysis in the literature is mostly limited to nominal conditions. In this work, we propose a framework to systematically assess the robust stability of feedback-based online optimization algorithms. We leverage tools from monotone operator theory, variational inequalities and classical robust control to obtain tractable numerical tests that guarantee robust convergence properties of online algorithms in feedback with a physical system, even in the presence of disturbances and model uncertainty. The results are illustrated via an academic example and a case study of a power distribution system.

approximation algorithms↗
DOE OSTI2025

Xopt and Badger: a machine learning ecosystem for real-time accelerator control and optimization

Machine learning (ML)-based black-box optimization algorithms have demonstrated significant improvements in accelerator optimization speed, often by orders of magnitude. However, deploying these algorithms in real-time facility control remains challenging due to the specialized expertise and infrastructure required. To bridge this gap, we introduce the Xopt ecosystem, a versatile suite of tools designed to make advanced ML-based optimization accessible to the broader accelerator community. This ecosystem includes Xopt, a modular Python framework that facilitates the integration of ML-based optimization algorithms with arbitrary control problems, and Badger, a graphical user interface built on top of Xopt, which enables seamless deployment of ML algorithms in real-time control systems. The Xopt ecosystem has been successfully applied towards solving challenging real-time control problems at leading international accelerator facilities, including SLAC, LBNL, Argonne, Fermilab, BNL, DESY, and ESRF, demonstrating its effectiveness in real-world optimization tasks. In this presentation, we provide an overview of Xopt’s capabilities and illustrate its impact through case studies from SLAC accelerator facilities including LCLS, LCLS-II, and FACET-II.

Roussel, Ryan [SLAC]↗
DOE OSTI2021

Inference of neutrino flavor evolution through data assimilation and neural differential equations

The evolution of neutrino flavor in dense environments such as core-collapse supernovae and binary compact object mergers constitutes an important and unsolved problem. Its solution has potential implications for the dynamics and heavy-element nucleosynthesis in these environments. In this paper, we build upon recent work to explore inference-based techniques for the estimation of model parameters and neutrino flavor evolution histories. Here, we combine data assimilation, ordinary differential equation solvers, and neural networks to craft an inference approach tailored for nonlinear dynamical systems. Using this architecture, and a simple two-neutrino-beam, two-flavor model, we compare the performances of nine different optimization algorithms and expand upon previous assessments of the efficacy of inference for tackling problems in flavor evolution. We find that employing this new architecture, together with evolutionary optimization algorithms, accurately captures flavor histories in the small-scale model and allows us to quickly explore both model parameters and initial flavor content. In future work we plan to extend these inference techniques to large numbers of neutrinos.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Herbaceous Feedstock 2018 State of Technology Report

The U.S. Department of Energy (DOE) promotes the production of advanced liquid transportation fuels from lignocellulosic biomass by funding fundamental and applied research that advances the state of technology (SOT). As part of its involvement with this mission, Idaho National Laboratory (INL) completes an annual SOT report for biomass feedstock logistics. This report summarizes supply system impacts of Bioenergy Technologies Office (BETO)-funded research and development efforts at INL and elsewhere (such as the High-Tonnage Feedstock Logistics projects (Webb et al. 2013a, Webb et al. 2013b, Webb et al. 2013c, Webb and Sokhansanj 2014, Sokhansanj et al. 2014) that lead to improvements in feedstock supply systems. These include improvements to and observed performance of innovative harvest and collection methods, storage technologies, transportation and handling approaches, and advanced preprocessing technologies. Biomass quality and variability, and the interface between feedstock quality and conversion performance are key drivers in addition to delivered feedstock cost. In this report, we estimate the benefits of R&D improvements to individual supply system unit operations, and present the status of feedstock logistics technology development for converting biomass into biofuels. These analyses are supported by experimental data where possible, and help to align the SOT relative to the cost goals defined in the Multi-Year Program Plan. The 2018 Herbaceous SOT aligned feedstock logistic design with current biorefinery’s design capacity utilized by biochemical conversion platform. Currently biochemical conversion platform utilizes a 725,000 dry ton/year biorefiney design for the techno economic analysis. Hence, feedstock delivered cost in the 2018 Herbaceous SOT is calculated based on biorefinery’s 725,000 dry ton design capacity instead of 800, 000 dry ton capacity utilized in the 2017 Herbaceous SOT. Biomass availabilities in this SOT were updated to year 2018 data from the 2016 Billion-Ton Report (BT16) (DOE 2016a), with the exception of switchgrass, for which the 2018 Herbaceous SOT utilized the 2019 switchgrass availability data from BT16. The BT16 report (DOE 2016a) does not project switchgrass availability in 2018; the soonest switchgrass is available in the BT16 report is 2019. Therefore, availability of switchgrass for this analysis was that projected for 2019. The 2018 Herbaceous SOT incorporates same technologies utilized in the 2017 Herbaceous SOT. However, a sensitivity analysis is performed to understand the impact of variation of process parameters on those technologies on feedstock logistic cost. New R&D data that shows the variations of process parameters affecting process performance is incorporated in the 2018 SOT to measure the variations in delivered feedstock cost. The 2018 Herbaceous SOT has also provided projected delivered feedstock of 2022 design case based on near term technical target under BETO funded R&D project. Finally, updated biorefinery size of 725,000 dry ton/year was incorporated within least-cost formulation model to select optimal siting and depot scales during optimization of the least cost blend. This modification to the optimization algorithm allows the trade-off between the cost of increased supply radius and the savings from selecting biomass from higher producing counties to be assessed. Such optimization has also showed the economic benefit of decentralized depots in comparison to centralized preprocessing co-located with the biorefinery by decoupling the biorefinery and feedstock locations. The 2018 Herbaceous SOT report documents the current modeled cost of a herbaceous feedstock supply system (from harvest to the pretreatment reactor throat, including grower payment) for hydrocarbon fuel production via biochemical conversion, based on equipment and processes now available or potentially available in the near term. The modeled cost also considers both the required quality and the availability of the biomass resources. The 2018 Herbaceous SOT predicts a modeled delivered feedstock cost of $83.67/dry ton (2016$); this is a $0.23/dry ton (2016$) decrease from the 2017 Herbaceous SOT. The modification of biorefinery’s designed capacity and increased projected biomass availability in the same supply shed contributed to this modeled cost reduction. The least-cost formulation model to optimally site and scale local distributed preprocessing depots also contributed to the cost reduction by considering county-level grower payment and distance from the biorefinery as variables in the optimization algorithm. Sensitivity analysis on various process parameters that affect delivered feedstock cost in the 2018 Herbaceous SOT shows that the delivered cost could varies from $80.45-$88.83/dry ton. The top factors that causes such variations are: effective baling rate, bale density, hammer mill throughput, interest rate and storage dry matter loss.

09 BIOMASS FUELS↗
DOE OSTI2024

Efficient First-Order Algorithms for Large-Scale, Non-Smooth Maximum Entropy Models with Application to Wildfire Science

Maximum entropy (MaxEnt) models are a class of statistical models that use the maximum entropy principle to estimate probability distributions from data. Due to the size of modern data sets, MaxEnt models need efficient optimization algorithms to scale well for big data applications. State-of-the-art algorithms for MaxEnt models, however, were not originally designed to handle big data sets; these algorithms either rely on technical devices that may yield unreliable numerical results, scale poorly, or require smoothness assumptions that many practical MaxEnt models lack. In this paper, we present novel optimization algorithms that overcome the shortcomings of state-of-the-art algorithms for training large-scale, non-smooth MaxEnt models. Our proposed first-order algorithms leverage the Kullback–Leibler divergence to train large-scale and non-smooth MaxEnt models efficiently. For MaxEnt models with discrete probability distribution of n elements built from samples, each containing m features, the stepsize parameter estimation and iterations in our algorithms scale on the order of O(mn) operations and can be trivially parallelized. Moreover, the strong ℓ1 convexity of the Kullback–Leibler divergence allows for larger stepsize parameters, thereby speeding up the convergence rate of our algorithms. To illustrate the efficiency of our novel algorithms, we consider the problem of estimating probabilities of fire occurrences as a function of ecological features in the Western US MTBS-Interagency wildfire data set. Our numerical results show that our algorithms outperform the state of the art by one order of magnitude and yield results that agree with physical models of wildfire occurrence and previous statistical analyses of wildfire drivers.

Physics↗
DOE OSTI2023

Direct stellarator coil optimization for nested magnetic surfaces with precise quasi-symmetry

We present a robust optimization algorithm for the design of electromagnetic coils that generate vacuum magnetic fields with nested flux surfaces and precise quasi-symmetry. The method is based on a bilevel optimization problem, where the outer coil optimization is constrained by a set of inner least squares optimization problems whose solutions describe magnetic surfaces. The outer optimization objective targets coils that generate a field with nested magnetic surfaces and good quasi-symmetry. The inner optimization problems identify magnetic surfaces when they exist, and approximate surfaces in the presence of magnetic islands or chaos. We show that this formulation can be used to heal islands and chaos, thus producing coils that result in magnetic fields with precise quasi-symmetry. We show that the method can be initialized with coils from the traditional two-stage coil design process, as well as coils from a near-axis expansion optimization. We present a numerical example where island chains are healed and quasi-symmetry is optimized up to surfaces with aspect ratio 6. Another numerical example illustrates that the aspect ratio of nested flux surfaces with optimized quasi-symmetry can be decreased from 6 to approximately 4. In conclusion, the last example shows that our approach is robust and a cold-start using coils from a near-axis expansion optimization.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2022

A Reinforcement Learning Approach to Parameter Selection for Distributed Optimal Power Flow

With the increasing penetration of distributed energy resources, distributed optimization algorithms have attracted significant attention for power systems applications due to their potential for superior scalability, privacy, and robustness to a single point-of-failure. The Alternating Direction Method of Multipliers (ADMM) is a popular distributed optimization algorithm; however, its convergence performance is highly dependent on the selection of penalty parameters, which are usually chosen heuristically. In this work, we use reinforcement learning (RL) to develop an adaptive penalty parameter selection policy for alternating current optimal power flow (ACOPF) problem solved via ADMM with the goal of minimizing the number of iterations until convergence. We train our RL policy using deep Q-learning and show that this policy can result in significantly accelerated convergence (up to a 59% reduction in the number of iterations compared to existing, curvatureinformed penalty parameter selection methods). Furthermore, we show that our RL policy demonstrates promise for generalizability, performing well under unseen loading schemes as well as under unseen losses of lines and generators (up to a 50% reduction in iterations). This work thus provides a proof-of-concept for using RL for parameter selection in ADMM for power systems applications.

alternating current optimal power flow↗